v3.16.5

v3.16.4
v3.16.3
2025-09-17 23:44:27 +00:00 · 2021-06-26 12:27:44 -04:00 · 2021-06-23 21:52:42 -04:00 · 2021-05-06 14:55:03 -04:00 · 2021-04-08 18:09:31 -04:00 · 2021-03-24 18:24:20 -04:00
477 changed files with 57894 additions and 49219 deletions
--- a/3
+++ b/3
@@ -33,3 +33,6 @@ Jay D Dee
 xcouiz@gmail.com

 Cryply
+
+Colin Percival
+Alexander Peslyak
--- a/81
+++ b/81
@@ -1,12 +1,14 @@


-Requirements:
+1. Requirements:
+---------------

 Intel Core2 or newer, or AMD Steamroller or newer CPU. ARM CPUs are not
 supported.
 64 bit Linux operating system. Apple is not supported.

-Building on linux prerequisites:
+2. Building on linux prerequisites:
+-----------------------------------

 It is assumed users know how to install packages on their system and
 be able to compile standard source packages. This is basic Linux and
@@ -20,49 +22,74 @@ http://askubuntu.com/questions/457526/how-to-install-cpuminer-in-ubuntu

 Install any additional dependencies needed by cpuminer-opt. The list below
 are some of the ones that may not be in the default install and need to
-be installed manually. There may be others, read the error messages they
-will give a clue as to the missing package.
+be installed manually. There may be others, read the compiler error messages,
+they will give a clue as to the missing package.

 The following command should install everything you need on Debian based
-distributions such as Ubuntu:
+distributions such as Ubuntu. Fedora and other distributions may have similar
+but different package names. 

-sudo apt-get install build-essential libssl-dev libcurl4-openssl-dev libjansson-dev libgmp-dev automake zlib1g-dev
-
-build-essential  (Development Tools package group on Fedora)
-automake
-libjansson-dev
-libgmp-dev
-libcurl4-openssl-dev
-libssl-dev
-lib-thread
-zlib1g-dev
+$ sudo apt-get install build-essential automake libssl-dev libcurl4-openssl-dev libjansson-dev libgmp-dev zlib1g-dev git

 SHA support on AMD Ryzen CPUs requires gcc version 5 or higher and
-openssl 1.1.0e or higher. Add one of the following, depending on the
-compiler version, to CFLAGS:
-"-march=native" or "-march=znver1" or "-msha".
+openssl 1.1.0e or higher. Add one of the following to CFLAGS for SHA
+support depending on your CPU and compiler version:
+
+"-march=native" is always the best choice
+
+"-march=znver1" for Ryzen 1000 & 2000 series, znver2 for 3000.
+
+"-msha"  Add SHA to other tuning options

 Additional instructions for static compilalation can be found here:
 https://lxadm.com/Static_compilation_of_cpuminer
 Static builds should only considered in a homogeneous HW and SW environment.
 Local builds will always have the best performance and compatibility.

-Extract cpuminer source.
+3. Download cpuminer-opt
+------------------------

-tar xvzf cpuminer-opt-x.y.z.tar.gz
-cd cpuminer-opt-x.y.z
+Download the source code for the latest realease from the official repository.

-Run ./build.sh to build on Linux or execute the following commands.
+https://github.com/JayDDee/cpuminer-opt/releases

-./autogen.sh
-CFLAGS="-O3 -march=native -Wall" ./configure --with-curl
-make
+Extract the source code.

-Start mining.
+$ tar xvzf cpuminer-opt-x.y.z.tar.gz
+
+
+Alternatively it can be cloned from git.
+
+$ git clone https://github.com/JayDDee/cpuminer-opt.git
+ 
+4. Build cpuminer-opt
+---------------------
+
+It is recomended to Build with default options, this will usuallly
+produce the best results.
+
+$ ./build.sh to build on Linux or execute the following commands.
+
+or 
+
+$ ./autogen.sh
+$ CFLAGS="-O3 -march=native -Wall" ./configure --with-curl
+$ make -j n
+
+n is the number of threads.
+
+5. Start mining.
+----------------
+
+$ ./cpuminer -a algo -o url -u username -p password

-./cpuminer -a algo -o url -u username -p password

 Windows
+-------
+
+See also INSTAL_WINDOWS
+
+The following procedure is obsolete and uses an old compiler.

 Precompiled Windows binaries are built on a Linux host using Mingw
 with a more recent compiler than the following Windows hosted procedure.
--- a/148
+++ b/148
@@ -1,5 +1,9 @@
 Instructions for compiling cpuminer-opt for Windows.

+Thwaw intructions nay be out of date. Please consult the wiki for
+the latest:
+
+https://github.com/JayDDee/cpuminer-opt/wiki/Compiling-from-source

 Windows compilation using Visual Studio is not supported. Mingw64 is
 used on a Linux system (bare metal or virtual machine) to cross-compile
@@ -22,14 +26,15 @@ Step by step...

 Refer to Linux compile instructions and install required packages.

-Additionally, install mingw-64.
+Additionally, install mingw-w64.

-sudo apt-get install mingw-w64
+sudo apt-get install mingw-w64 libz-mingw-w64-dev


 2. Create a local library directory for packages to be compiled in the next
-   step. Recommended location is $HOME/usr/lib/
+   step. Suggested location is $HOME/usr/lib/

+$ mkdir $HOME/usr/lib

 3. Download and build other packages for mingw that don't have a mingw64
   version available in the repositories.
@@ -37,67 +42,62 @@ sudo apt-get install mingw-w64
 Download the following source code packages from their respective and
 respected download locations, copy them to ~/usr/lib/ and uncompress them. 

-openssl
-curl
-gmp
+openssl: https://github.com/openssl/openssl/releases

-In most cases the latest vesrion is ok but it's safest to download
-the same major and minor version as included in your distribution.
+curl: https://github.com/curl/curl/releases

-Run the following commands or follow the supplied instructions.
-Do not run "make install" unless you are using ~/usr/lib, which isn't
-recommended.
+gmp: https://gmplib.org/download/gmp/

-Some instructions insist on running "make check". If make check fails
-it may still work, YMMV.
+In most cases the latest version is ok but it's safest to download the same major and minor version as included in your distribution. The following uses versions from Ubuntu 20.04. Change version numbers as required.

-You can speed up "make" by using all CPU cores available with "-j n" where
-n is the number of CPU threads you want to use.
+Run the following commands or follow the supplied instructions. Do not run "make install" unless you are using /usr/lib, which isn't recommended.
+
+Some instructions insist on running "make check". If make check fails it may still work, YMMV.
+
+You can speed up "make" by using all CPU cores available with "-j n" where n is the number of CPU threads you want to use.

 openssl:

-./Configure mingw64 shared --cross-compile-prefix=x86_64-w64-mingw32
-make
+$ ./Configure mingw64 shared --cross-compile-prefix=x86_64-w64-mingw32-
+$ make
+
+Make may fail with an ld error, just ensure libcrypto-1_1-x64.dll is created.

 curl:

-./configure --with-winssl --with-winidn --host=x86_64-w64-mingw32
-make
+$ ./configure --with-winssl --with-winidn --host=x86_64-w64-mingw32
+$ make

 gmp:

-./configure --host=x86_64-w64-mingw32 
-make
-
-
+$ ./configure --host=x86_64-w64-mingw32
+$ make

 4. Tweak the environment.

-This step is required everytime you login or the commands can be added to
-.bashrc.
+This step is required everytime you login or the commands can be added to .bashrc.

-Define some local variables to point to local library. 
+Define some local variables to point to local library.

-export LOCAL_LIB="$HOME/usr/lib"
+$ export LOCAL_LIB="$HOME/usr/lib"

-export LDFLAGS="-L$LOCAL_LIB/curl/lib/.libs -L$LOCAL_LIB/gmp/.libs -L$LOCAL_LIB/openssl"
+$ export LDFLAGS="-L$LOCAL_LIB/curl/lib/.libs -L$LOCAL_LIB/gmp/.libs -L$LOCAL_LIB/openssl"

-export CONFIGURE_ARGS="--with-curl=$LOCAL_LIB/curl --with-crypto=$LOCAL_LIB/openssl --host=x86_64-w64-mingw32"
+$ export CONFIGURE_ARGS="--with-curl=$LOCAL_LIB/curl --with-crypto=$LOCAL_LIB/openssl --host=x86_64-w64-mingw32"

-Create a release directory and copy some dll files previously built.
-This can be done outside of cpuminer-opt and only needs to be done once.
-If the release directory is in cpuminer-opt directory it needs to be
-recreated every a source package is decompressed.
+Adjust for gcc version:

-mkdir release
-cp /usr/x86_64-w64-mingw32/lib/zlib1.dll release/
-cp /usr/x86_64-w64-mingw32/lib/libwinpthread-1.dll release/
-cp /usr/lib/gcc/x86_64-w64-mingw32/7.3-win32/libstdc++-6.dll release/
-cp /usr/lib/gcc/x86_64-w64-mingw32/7.3-win32/libgcc_s_seh-1.dll release/
-cp $LOCAL_LIB/openssl/libcrypto-1_1-x64.dll release/
-cp $LOCAL_LIB/curl/lib/.libs/libcurl-4.dll release/
+$ export GCC_MINGW_LIB="/usr/lib/gcc/x86_64-w64-mingw32/9.3-win32"

+Create a release directory and copy some dll files previously built. This can be done outside of cpuminer-opt and only needs to be done once. If the release directory is in cpuminer-opt directory it needs to be recreated every time a source package is decompressed.

+$ mkdir release
+$ cp /usr/x86_64-w64-mingw32/lib/zlib1.dll release/
+$ cp /usr/x86_64-w64-mingw32/lib/libwinpthread-1.dll release/
+$ cp $GCC_MINGW_LIB/libstdc++-6.dll release/
+$ cp $GCC_MINGW_LIB/libgcc_s_seh-1.dll release/
+$ cp $LOCAL_LIB/openssl/libcrypto-1_1-x64.dll release/
+$ cp $LOCAL_LIB/curl/lib/.libs/libcurl-4.dll release/

 The following steps need to be done every time a new source package is
 opened.
@@ -111,13 +111,73 @@ https://github.com/JayDDee/cpuminer-opt/releases

 Decompress and change to the cpuminer-opt directory.

-
-
-6. Prepare to compile
+6. compile

 Create a link to the locally compiled version of gmp.h

-ln -s $LOCAL_LIB/gmp-version/gmp.h ./gmp.h
+$ ln -s $LOCAL_LIB/gmp-version/gmp.h ./gmp.h
+
+$ ./autogen.sh
+
+Configure the compiler for the CPU architecture of the host machine:
+
+CFLAGS="-O3 -march=native -Wall" ./configure $CONFIGURE_ARGS
+
+or cross compile for a specific CPU architecture:
+
+CFLAGS="-O3 -march=znver1 -Wall" ./configure $CONFIGURE_ARGS
+
+This will compile for AMD Ryzen.
+
+You can compile more generically for a set of specific CPU features if you know what features you want:
+
+CFLAGS="-O3 -maes -msse4.2 -Wall" ./configure $CONFIGURE_ARGS
+
+This will compile for an older CPU that does not have AVX.
+
+You can find several examples in README.txt
+
+If you have a CPU with more than 64 threads and Windows 7 or higher you can enable the CPU Groups feature by adding the following to CFLAGS:
+
+"-D_WIN32_WINNT=0x0601"
+
+Once you have run configure successfully run the compiler with n CPU threads:
+
+$ make -j n
+
+Copy cpuminer.exe to the release directory, compress and copy the release directory to a Windows system and run cpuminer.exe from the command line.
+
+Run cpuminer
+
+In a command windows change directories to the unzipped release folder. to get a list of all options:
+
+cpuminer.exe --help
+
+Command options are specific to where you mine. Refer to the pool's instructions on how to set them.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Create a link to the locally compiled version of gmp.h
+
+$ ln -s $LOCAL_LIB/gmp-version/gmp.h ./gmp.h

 Edit configure.ac to fix lipthread package name.

--- a/Makefile.am
+++ b/Makefile.am
@@ -18,19 +18,9 @@ dist_man_MANS	= cpuminer.1
 cpuminer_SOURCES = \
  cpu-miner.c \
  util.c \
-  uint256.cpp \
  api.c \
  sysinfos.c \
  algo-gate-api.c\
-  crypto/oaes_lib.c \
-  crypto/c_keccak.c \
-  crypto/c_groestl.c \
-  crypto/c_blake256.c \
-  crypto/c_jh.c \
-  crypto/c_skein.c \
-  crypto/hash.c \
-  crypto/aesb.c \
-  crypto/magimath.cpp \
  algo/argon2/argon2a/argon2a.c \
  algo/argon2/argon2a/ar2/argon2.c \
  algo/argon2/argon2a/ar2/opt.c \
@@ -77,31 +67,29 @@ cpuminer_SOURCES = \
  algo/bmw/bmw512-gate.c \
  algo/bmw/bmw512.c \
  algo/bmw/bmw512-4way.c \
-  algo/cryptonight/cryptolight.c \
-  algo/cryptonight/cryptonight-common.c\
-  algo/cryptonight/cryptonight-aesni.c\
-  algo/cryptonight/cryptonight.c\
-  algo/cubehash/sph_cubehash.c \
  algo/cubehash/cubehash_sse2.c\
  algo/cubehash/cube-hash-2way.c \
  algo/echo/sph_echo.c \
+  algo/echo/echo-hash-4way.c \
  algo/echo/aes_ni/hash.c\
  algo/gost/sph_gost.c \
+  algo/groestl/groestl-gate.c \
+  algo/groestl/groestl512-hash-4way.c \
+  algo/groestl/groestl256-hash-4way.c \
  algo/groestl/sph_groestl.c \
  algo/groestl/groestl.c \
+  algo/groestl/groestl-4way.c \
  algo/groestl/myrgr-gate.c \
  algo/groestl/myrgr-4way.c \
  algo/groestl/myr-groestl.c \
  algo/groestl/aes_ni/hash-groestl.c \
  algo/groestl/aes_ni/hash-groestl256.c \
  algo/fugue/sph_fugue.c \
+  algo/fugue/fugue-aesni.c \
  algo/hamsi/sph_hamsi.c \
  algo/hamsi/hamsi-hash-4way.c \
  algo/haval/haval.c \
  algo/haval/haval-hash-4way.c \
-  algo/heavy/sph_hefty1.c \
-  algo/heavy/heavy.c \
-  algo/heavy/bastion.c \
  algo/hodl/aes.c \
  algo/hodl/hodl-gate.c \
  algo/hodl/hodl-wolf.c \
@@ -117,13 +105,15 @@ cpuminer_SOURCES = \
  algo/keccak/keccak-hash-4way.c \
  algo/keccak/keccak-4way.c\
  algo/keccak/keccak-gate.c \
-  algo/keccak/sse2/keccak.c \
-  algo/luffa/sph_luffa.c \
-  algo/luffa/luffa.c \
+  algo/keccak/sha3d-4way.c \
+  algo/keccak/sha3d.c \
+  algo/lanehash/lane.c \
  algo/luffa/luffa_for_sse2.c \
  algo/luffa/luffa-hash-2way.c \
  algo/lyra2/lyra2.c \
  algo/lyra2/sponge.c \
+  algo/lyra2/sponge-2way.c \
+  algo/lyra2/lyra2-hash-2way.c \
  algo/lyra2/lyra2-gate.c \
  algo/lyra2/lyra2rev2.c \
  algo/lyra2/lyra2rev2-4way.c \
@@ -139,13 +129,14 @@ cpuminer_SOURCES = \
  algo/lyra2/allium.c \
  algo/lyra2/phi2-4way.c \
  algo/lyra2/phi2.c \
-  algo/m7m.c \
+  algo/m7m/m7m.c \
+  algo/m7m/magimath.cpp \
  algo/nist5/nist5-gate.c \
  algo/nist5/nist5-4way.c \
  algo/nist5/nist5.c \
  algo/nist5/zr5.c \
+  algo/panama/panama-hash-4way.c \
  algo/panama/sph_panama.c \
-  algo/radiogatun/sph_radiogatun.c \
  algo/quark/quark-gate.c \
  algo/quark/quark.c \
  algo/quark/quark-4way.c \
@@ -168,13 +159,12 @@ cpuminer_SOURCES = \
  algo/ripemd/lbry-4way.c \
  algo/scrypt/scrypt.c \
  algo/scrypt/neoscrypt.c \
-  algo/scrypt/pluck.c \
-  algo/scryptjane/scrypt-jane.c \
  algo/sha/sph_sha2.c \
  algo/sha/sph_sha2big.c \
  algo/sha/sha256-hash-4way.c \
  algo/sha/sha512-hash-4way.c \
-  algo/sha/sha256_hash_11way.c \
+  algo/sha/hmac-sha256-hash.c \
+  algo/sha/hmac-sha256-hash-4way.c \
  algo/sha/sha2.c \
  algo/sha/sha256t-gate.c \
  algo/sha/sha256t-4way.c \
@@ -186,8 +176,8 @@ cpuminer_SOURCES = \
  algo/shavite/sph_shavite.c \
  algo/shavite/sph-shavite-aesni.c \
  algo/shavite/shavite-hash-2way.c \
+  algo/shavite/shavite-hash-4way.c \
  algo/shavite/shavite.c \
-  algo/simd/sph_simd.c \
  algo/simd/nist.c \
  algo/simd/vector.c \
  algo/simd/simd-hash-2way.c \
@@ -198,10 +188,15 @@ cpuminer_SOURCES = \
  algo/skein/skein-gate.c \
  algo/skein/skein2.c \
  algo/skein/skein2-4way.c \
-  algo/skein/skein2-gate.c \
  algo/sm3/sm3.c \
  algo/sm3/sm3-hash-4way.c \
+  algo/swifftx/swifftx.c \
  algo/tiger/sph_tiger.c \
+  algo/verthash/verthash-gate.c \
+  algo/verthash/Verthash.c \
+  algo/verthash/fopen_utf8.c \
+  algo/verthash/tiny_sha3/sha3.c \
+  algo/verthash/tiny_sha3/sha3-4way.c \
  algo/whirlpool/sph_whirlpool.c \
  algo/whirlpool/whirlpool-hash-4way.c \
  algo/whirlpool/whirlpool-gate.c \
@@ -225,7 +220,6 @@ cpuminer_SOURCES = \
  algo/x11/timetravel10-gate.c \
  algo/x11/timetravel10.c \
  algo/x11/timetravel10-4way.c \
-  algo/x11/fresh.c \
  algo/x11/x11evo.c \
  algo/x11/x11evo-4way.c \
  algo/x11/x11evo-gate.c \
@@ -244,7 +238,6 @@ cpuminer_SOURCES = \
  algo/x13/skunk-gate.c \
  algo/x13/skunk-4way.c \
  algo/x13/skunk.c \
-  algo/x13/drop.c \
  algo/x13/x13bcd-4way.c \
  algo/x13/x13bcd.c \
  algo/x14/x14-gate.c \
@@ -263,11 +256,14 @@ cpuminer_SOURCES = \
  algo/x16/x16r-gate.c \
  algo/x16/x16r.c \
  algo/x16/x16r-4way.c \
+  algo/x16/x16rv2.c \
+  algo/x16/x16rv2-4way.c \
  algo/x16/x16rt.c \
  algo/x16/x16rt-4way.c \
  algo/x16/hex.c \
  algo/x16/x21s-4way.c \
  algo/x16/x21s.c \
+  algo/x16/minotaur.c \
  algo/x17/x17-gate.c \
  algo/x17/x17.c \
  algo/x17/x17-4way.c \
@@ -277,12 +273,17 @@ cpuminer_SOURCES = \
  algo/x17/sonoa-gate.c \
  algo/x17/sonoa-4way.c \
  algo/x17/sonoa.c \
-  algo/x20/x20r.c \
+  algo/x22/x22i-4way.c \
+  algo/x22/x22i.c \
+  algo/x22/x22i-gate.c \
+  algo/x22/x25x.c \
+  algo/x22/x25x-4way.c \
  algo/yescrypt/yescrypt.c \
-  algo/yescrypt/sha256_Y.c \
  algo/yescrypt/yescrypt-best.c \
-  algo/yespower/yespower.c \
-  algo/yespower/sha256_p.c \
+  algo/yespower/yespower-gate.c \
+  algo/yespower/yespower-blake2b.c \
+  algo/yespower/crypto/blake2b-yp.c \
+  algo/yespower/yescrypt-r8g.c \
  algo/yespower/yespower-opt.c

 disable_flags =
--- a/README.md
+++ b/README.md
@@ -12,10 +12,24 @@ a false positive, they are flagged simply because they are cryptocurrency
 miners. The source code is open for anyone to inspect. If you don't trust 
 the software, don't use it.

+
+New thread:
+
+https://bitcointalk.org/index.php?topic=5226770.msg53865575#msg53865575
+
+Old thread:
+
 https://bitcointalk.org/index.php?topic=1326803.0

 mailto://jayddee246@gmail.com

+This note is to confirm that bitcointalk users JayDDee and joblo are the
+same person.
+
+I created a new BCT user JayDDee to match my github user id.
+The old thread has been locked but still contains useful information for
+reading.
+
 See file RELEASE_NOTES for change log and INSTALL_LINUX or INSTALL_WINDOWS
 for compile instructions.

@@ -23,25 +37,25 @@ Requirements
 ------------

 1. A x86_64 architecture CPU with a minimum of SSE2 support. This includes
-Intel Core2 and newer and AMD equivalents. In order to take advantage of AES_NI
-optimizations a CPU with AES_NI is required. This includes Intel Westbridge
-and newer and AMD equivalents. Further optimizations are available on some
-algoritms for CPUs with AVX and AVX2, Sandybridge and Haswell respectively.
+Intel Core2 and newer and AMD equivalents. Further optimizations are available
+on some algoritms for CPUs with AES, AVX, AVX2, SHA, AVX512 and VAES.

 Older CPUs are supported by cpuminer-multi by TPruvot but at reduced
 performance.

-ARM CPUs are not supported.
+ARM and Aarch64 CPUs are not supported.

-2. 64 bit Linux OS. Ubuntu and Fedora based distributions, including Mint and
-Centos, are known to work and have all dependencies in their repositories.
-Others may work but may require more effort. Older versions such as Centos 6
-don't work due to missing features. 
+2. 64 bit Linux or Windows OS. Ubuntu and Fedora based distributions,
+including Mint and Centos, are known to work and have all dependencies
+in their repositories. Others may work but may require more effort. Older
+versions such as Centos 6 don't work due to missing features. 
 64 bit Windows OS is supported with mingw_w64 and msys or pre-built binaries.

 MacOS, OSx and Android are not supported.

-3. Stratum pool. Some algos may work wallet mining using getwork or GBT. YMMV.
+3. Stratum pool supporting stratum+tcp:// or stratum+ssl:// protocols or
+RPC getwork using http:// or https://.
+GBT is YMMV.

 Supported Algorithms
 --------------------
@@ -53,7 +67,6 @@ Supported Algorithms
                          argon2d500    argon2d-dyn,  Dynamic (DYN)
                          argon2d4096   argon2d-uis, Unitus, (UIS)
                          axiom         Shabal-256 MemoHash
-                          bastion
                          blake         Blake-256 (SFR)
                          blake2b       Blake2b 256
                          blake2s       Blake-2 S
@@ -64,10 +77,7 @@ Supported Algorithms
                          decred
                          deep          Deepcoin (DCN)
                          dmd-gr        Diamond-Groestl
-                          drop          Dropcoin
-                          fresh         Fresh
                          groestl       Groestl coin
-                          heavy         Heavy
                          hex           x16r-hex
                          hmq1725       Espers
                          hodl          Hodlcoin
@@ -79,26 +89,29 @@ Supported Algorithms
                          lyra2h        Hppcoin
                          lyra2re       lyra2
                          lyra2rev2     lyra2v2
-                          lyra2rev3     lyrav2v3, Vertcoin
+                          lyra2rev3     lyrav2v3
                          lyra2z        
                          lyra2z330     Lyra2 330 rows, Zoin (ZOI)
                          m7m           Magi (XMG)
+                          minotaur      Ringcoin (RNG)
                          myr-gr        Myriad-Groestl
                          neoscrypt     NeoScrypt(128, 2, 1)
                          nist5         Nist5
                          pentablake    Pentablake
-                          phi1612       phi, LUX coin (original algo)
-                          phi2          LUX coin (new algo)
+                          phi1612       phi
+                          phi2          Luxcoin (LUX)
+                          phi2-lux      identical to phi2
                          pluck         Pluck:128 (Supcoin)
                          polytimos     Ninja
+                          power2b       MicroBitcoin (MBC)
                          quark         Quark
                          qubit         Qubit
                          scrypt        scrypt(1024, 1, 1) (default)
                          scrypt:N      scrypt(N, 1, 1)
-                          scryptjane:nf
                          sha256d       Double SHA-256
                          sha256q       Quad SHA-256, Pyrite (PYE)
                          sha256t       Triple SHA-256, Onecoin (OC)
+                          sha3d         Double keccak256 (BSHA3)
                          shavite3      Shavite3
                          skein         Skein+Sha (Skeincoin)
                          skein2        Double Skein (Woodcoin)
@@ -109,6 +122,7 @@ Supported Algorithms
                          tribus        Denarius (DNR)
                          vanilla       blake256r8vnl (VCash)
                          veltor        (VLT)
+                          verthash      Vertcoin
                          whirlpool
                          whirlpoolx
                          x11           Dash
@@ -120,24 +134,53 @@ Supported Algorithms
                          x13sm3        hsr (Hshare)
                          x14           X14
                          x15           X15
-                          x16r          Ravencoin (RVN)
+                          x16r          
+                          x16rv2        
                          x16rt         Gincoin (GIN)
-                          x16rt_veil    Veil (VEIL)
+                          x16rt-veil    Veil (VEIL)
                          x16s          Pigeoncoin (PGN)
                          x17
                          x21s
+                          x22i
+                          x25x
                          xevan         Bitsend (BSD)
                          yescrypt      Globalboost-Y (BSTY)
                          yescryptr8    BitZeny (ZNY)
+                          yescryptr8g   Koto (KOTO)
                          yescryptr16   Eli
                          yescryptr32   WAVI
                          yespower      Cryply
                          yespowerr16   Yenten (YTN)
+                          yespower-b2b  generic yespower + blake2b
                          zr5           Ziftr

+Many variations of scrypt based algos can be mine by specifying their
+parameters:
+
+scryptn2: --algo scrypt --param-n 1048576
+
+cpupower: --algo yespower --param-key "CPUpower: The number of CPU working or available for proof-of-work mining"
+
+power2b: --algo yespower-b2b --param-n 2048 --param-r 32 --param-key "Now I am become Death, the destroyer of worlds"
+
+sugarchain: --algo yespower --param-n 2048 -param-r 32 --param-key "Satoshi Nakamoto 31/Oct/2008 Proof-of-work is essentially one-CPU-one-vote"
+
+yespoweriots: --algo yespower --param-n 2048 --param-key "Iots is committed to the development of IOT"
+
+yespowerlitb: --algo yespower --param-n 2048 --param-r 32 --param-key "LITBpower: The number of LITB working or available for proof-of-work mini"
+
+yespoweric: --algo yespower --param-n 2048 --param-r 32 --param-key "IsotopeC" 
+
+yespowerurx: --algo yespower --param-n 2048 --param-r 32 --param-key "UraniumX"
+
+yespowerltncg: --algo yespower --param-n 2048 --param-r 32 --param-key "LTNCGYES"
+
 Errata
 ------

+Old algorithms that are no longer used frequently will not have the latest
+optimizations.
+
 Cryptonight and variants are no longer supported, use another miner.

 Neoscrypt crashes on Windows, use legacy version.
@@ -155,14 +198,17 @@ Benchmark testing does not work for x11evo.
 Bugs
 ----

-Users are encouraged to post their bug reports on the Bitcoin Talk
-forum at:
+Users are encouraged to post their bug reports using git issues or on the
+Bitcoin Talk forum or opening an issue in git:

 https://bitcointalk.org/index.php?topic=1326803.0

-All problem reports must be accompanied by a proper definition.
+https://github.com/JayDDee/cpuminer-opt/issues
+
+All problem reports must be accompanied by a proper problem definition.
 This should include how the problem occurred, the command line and
-output from the miner showing the startup and any errors.
+output from the miner showing the startup messages and any errors.
+A history is also useful, ie did it work before.

 Donations
 ---------
@@ -170,10 +216,6 @@ Donations
 cpuminer-opt has no fees of any kind but donations are accepted.

 BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
- ETH: 0x72122edabcae9d3f57eab0729305a425f6fef6d0
- LTC: LdUwoHJnux9r9EKqFWNvAi45kQompHk6e8
- BCH: 1QKYkB6atn4P7RFozyziAXLEnurwnUM1cQ
- BTG: GVUyECtRHeC5D58z9F3nGGfVQndwnsPnHQ

 Happy mining!

--- a/README.txt
+++ b/README.txt
@@ -1,8 +1,12 @@
 This file is included in the Windows binary package. Compile instructions
 for Linux and Windows can be found in RELEASE_NOTES.

-cpuminer is a console program that is executed from a DOS command prompt.
-There is no GUI and no mouse support.
+This package is officially avalable only from:
+ https://github.com/JayDDee/cpuminer-opt
+No other sources should be trusted.
+
+cpuminer is a console program that is executed from a DOS or Powershell
+prompt. There is no GUI and no mouse support.

 Miner programs are often flagged as malware by antivirus programs. This is
 a false positive, they are flagged simply because they are cryptocurrency 
@@ -10,12 +14,12 @@ miners. The source code is open for anyone to inspect. If you don't trust
 the software, don't use it.

 Choose the exe that best matches you CPU's features or use trial and
-error to find the fastest one that doesn't crash. Pay attention to
+error to find the fastest one that works. Pay attention to
 the features listed at cpuminer startup to ensure you are mining at
 optimum speed using the best available features.

 Architecture names and compile options used are only provided for Intel
-Core series. Even the newest Pentium and Celeron CPUs are often missing
+Core series. Budget CPUs like Pentium and Celeron are often missing some
 features.

 AMD CPUs older than Piledriver, including Athlon x2 and Phenom II x4, are not
@@ -23,13 +27,42 @@ supported by cpuminer-opt due to an incompatible implementation of SSE2 on
 these CPUs. Some algos may crash the miner with an invalid instruction.
 Users are recommended to use an unoptimized miner such as cpuminer-multi.

-Exe name                Compile flags            Arch name
+More information for Intel and AMD CPU architectures and their features
+can be found on Wikipedia.

-cpuminer-sse2.exe      "-msse2"                  Core2, Nehalem   
-cpuminer-aes-sse42.exe "-march=westmere"         Westmere
-cpuminer-avx.exe       "-march=corei7-avx"       Sandy-Ivybridge
-cpuminer-avx2.exe      "-march=core-avx2"        Haswell, Sky-Kaby-Coffeelake
-cpuminer-zen           "-march=znver1"           AMD Ryzen, Threadripper
+https://en.wikipedia.org/wiki/List_of_Intel_CPU_microarchitectures
+
+https://en.wikipedia.org/wiki/List_of_AMD_CPU_microarchitectures
+
+
+Exe file name                Compile flags              Arch name
+
+cpuminer-sse2.exe            "-msse2"                   Core2, Nehalem   
+cpuminer-aes-sse42.exe       "-march=westmere"          Westmere
+cpuminer-avx.exe             "-march=corei7-avx"        Sandybridge, Ivybridge
+cpuminer-avx2.exe            "-march=core-avx2 -maes"   Haswell(1)
+cpuminer-avx512.exe          "-march=skylake-avx512"    Skylake-X, Cascadelake
+cpuminer-avx512-sha.exe      "-march=cascadelake -msha" Rocketlake(2)
+cpuminer-avx512-sha-vaes.exe "-march=icelake-client"    Icelake, Tigerlake(3)
+cpuminer-zen.exe             "-march=znver1"            AMD Zen1, Zen2
+cpuminer-zen3.exe            "-march=znver2 -mvaes"     Zen3(4)
+
+(1) Haswell includes Broadwell, Skylake, Kabylake, Coffeelake & Cometlake. 
+(2) Rocketlake build uses cascadelake+sha as a workaround until Rocketlake
+    compiler support is avalable.
+(3) Icelake & Tigerlake are only available on some laptops. Mining with a
+    laptop is not recommended.
+(4) Zen3 build uses zen2+vaes as a workaround until Zen3 compiler support is
+    available. Zen2 CPUs should use Zen1 build.
+
+Notes about included DLL files:
+
+Downloading DLL files from alternative sources presents an inherent
+security risk if their source is unknown. All DLL files included have
+been copied from the Ubuntu-20.04 installation or compiled by me from
+source code obtained from the author's official repository. The exact
+procedure is documented in the build instructions for Windows:
+https://github.com/JayDDee/cpuminer-opt/wiki/Compiling-from-source

 If you like this software feel free to donate:

--- a/639
+++ b/639
@@ -1,21 +1,18 @@
 cpuminer-opt is a console program run from the command line using the
 keyboard, not the mouse.

-cpuminer-opt now supports HW SHA acceleration available on AMD Ryzen CPUs.
-This feature requires recent SW including GCC version 5 or higher and
-openssl version 1.1 or higher. It may also require using "-march=znver1"
-compile flag.
-
-cpuminer-opt is a console program, if you're using a mouse you're doing it
-wrong.
+See also README.md for list of supported algorithms,

 Security warning
 ----------------

 Miner programs are often flagged as malware by antivirus programs. This is
-a false positive, they are flagged simply because they are cryptocurrency 
-miners. The source code is open for anyone to inspect. If you don't trust 
-the software, don't use it.
+usually a false positive, they are flagged simply because they are
+cryptocurrency miners. However, some malware masquerading as a miner has
+been spread using the cover that miners are known to be subject to false
+positives ans users will dismiss the AV alert. Always be on alert.
+The source code of cpuminer-opt is open for anyone to inspect.
+If you don't trust the software don't download it.

 The cryptographic hashing code has been taken from trusted sources but has been
 modified for speed at the expense of accepted security practices. This
@@ -25,7 +22,7 @@ required.
 Compile Instructions
 --------------------

-See INSTALL_LINUX or INSTALL_WINDOWS fror compile instruuctions
+See INSTALL_LINUX or INSTALL_WINDOWS for compile instruuctions

 Requirements
 ------------
@@ -33,11 +30,629 @@ Requirements
 Intel Core2 or newer, or AMD Steamroller or newer CPU. ARM CPUs are not
 supported.

-64 bit Linux or Windows operating system. Apple and Android are not supported.
+64 bit Linux or Windows operating system. Apple, Android and Raspberry Pi
+are not supported. FreeBSD YMMV.
+
+Reporting bugs
+--------------
+
+Bugs can be reported by sending am email to JayDDee246@gmail.com or opening
+an issue in git: https://github.com/JayDDee/cpuminer-opt/issues
+
+Please include the following information:
+
+1. CPU model, operating system, cpuminer-opt version (must be latest),
+   binary file for Windows, changes to default build procedure for Linux.
+
+2. Exact command line (except user and pw) and intial output showing
+   the above requested info.
+
+3. Additional program output showing any error messages or other
+   pertinent data.
+
+4. A clear description of the problem including history, scope,
+   persistence or intermittance, and reproduceability. 
+
+In simpler terms:
+
+What is it doing?
+What should it be doing instead?
+Did it work in a previous release?
+Does it happen for all algos? All pools? All options? Solo?
+Does it happen all the time?
+If not what makes it happen or not happen? 

 Change Log
 ----------

+v3.16.5
+
+#329: Fixed GBT incorrect target diff in stats, second attempt.
+Fixed formatting error in share result log when --no-color option is used.
+
+v3.16.4
+
+Faster sha512 and sha256 when not using SHA CPU extension.
+#329: Fixed GBT incorrect target diff in stats.
+
+v3.16.3
+
+#313 Fix compile error with GCC 11.
+Incremental improvements to verthash.
+
+v3.16.2
+
+Verthash: midstate prehash optimization for all architectures.
+Verthash: AVX2 optimization.
+GBT: added support for Bech32 addresses.
+Linux: added CPU frequency to benchmark log.
+Fixed integer overflow in time calculations.
+
+v3.16.1
+
+New options for verthash:
+  --data-file to specify the name, and optionally the path, of the verthash
+              data file, default is "verthash.dat" in the current directory.
+  --verify to perform the data file integrity check at startup, default is
+           not to verify data file integrity.
+Support for creation of default verthash data file if:
+   1) --data-file option is not used,
+   2) no default data file is found in the current directory, and,
+   3) --verify option is used.
+More detailed logs related to verthash data file.
+Small verthash performance improvement.
+Fixed detection of corrupt stats caused by networking issues.
+
+v3.16.0
+
+Added verthash algo.
+
+v3.15.7
+
+Added accepted/stale/rejected percentage to summary log report.
+Added warning if share counters mismatch which could corrupt stats.
+Linux: CPU temperature reporting is more responsive to rising temperature.
+A few AVX2 & AVX512 tweaks.
+Removed some dead code and other cleanup.
+
+v3.15.6
+
+Implement keccak pre-hash optimization for x16* algos.
+Move conditional mining test to before get_new_work in miner thread.
+Add test for share reject reason when solo mining.
+Add support for floating point, as well as integer, "networkhasps" in
+RPC getmininginfo method.
+
+v3.15.5
+
+Fix stratum jobs lost if 2 jobs received in less than one second.
+
+v3.15.4
+
+Fixed yescryptr16 broken in v3.15.3.
+
+v3.15.3
+
+Yescrypt algos now use yespower v0.5, a little faster.
+New implementation of sha256 using SHA CPU extension.
+Replace Openssl with SPH for sha256 & sha512.
+AVX512 optimization for sha256t & sha256q.
+Faster sha256t, sha256q, x21s, x22i & x25x on CPUs with SHA without AVX512.
+AVX512+SHA build for Intel Rocketlake added to Windows binary package. 
+
+v3.15.2
+
+Zen3 AVX2+VAES optimization for x16*, x17, sonoa, xevan, x21s, x22i, x25x,
+allium.
+Zen3 (AVX2+SHA+VAES) build added to Windows binary package.
+
+v3.15.1
+
+Fix compile on AMD Zen3 CPUs with VAES.
+Force new work immediately after solving a block solo.
+
+
+v3.15.0
+
+Fugue optimized with AES, improves many sha3 algos.
+Minotaur algo optimized for all architectures.
+Fixed neoscrypt BUG log.
+ 
+v3.14.3
+
+#265: more mutex changes to reduce blocking with high thread count.
+
+#267: fixed hodl algo potential memory alignment issue,
+      add warning when thread count is not valid for mining hodl algo.
+
+v3.14.2
+
+The second line of the Share Accepted log is no longer displayed,
+new Xnonce log is added and other small log tweaks.
+
+#265: Cleanup use of mutex.
+
+v3.14.1
+
+GBT and getwork log changes:
+ fixed missing TTF in New Block log,
+ ntime no longer byte-swapped for display in New Work log,
+ fixed zero effective hash rate in Periodic Report log,
+ deleted "Current block is..." log.
+
+Renamed stratum "New Job" log to "New Work" to be consistent with the solo
+version of the log. Added more data to both versions.
+
+v3.14.0
+
+Changes to solo mining:
+  - segwit is supported by getblocktemplate,
+  - longpolling is not working and is disabled,
+  - Periodic Report log is output,
+  - New Block log includes TTF estimates,
+  - Stratum thread no longer created when using getwork or GBT.
+
+Fixed BUG log mining sha256d.
+
+v3.13.1.1
+
+Fixed Windows crash mining minotaur algo.
+
+Fixed GCC 10 compile again.
+Added -fno-common to testing to be consistent with GCC 10 default.
+
+v3.13.1
+
+Added minotaur algo for Ringcoin.
+
+v3.13.0.1
+
+Issue #262: Fixed xevan AVX2 invalid shares.
+
+v3.13.0
+
+Updated Windows binaries compiled with GCC 9. Included DLLs also updated.
+Icelake build (cpuminer-avx512-sha-vaes.exe) now included in Windows
+binaries package.
+
+No source code changes.
+
+v3.12.8.2
+
+Fixed x12 AVX2 rejects.
+Fixed phi AVX2 crash.
+
+v3.12.8.1
+
+Issue #261: Fixed yescryptr8g invalid shares.
+
+v3.12.8
+
+Yespower sha256 prehash made thread safe.
+
+Rewrote diff conversion functions from scratch to be simpler and use 
+long double (float80) and int128 arithmetic for improved accuracy and
+precision.
+
+Some code cleanup and assorted small changes.
+
+v3.12.7
+
+Issue #257: fixed a file descriptor leak which caused the CPU temperature
+and frequency query to report zeros after mining for a couple of hours.
+
+Issue #253: stale share reduction for yescrypt, sonoa.
+
+v3.12.6.1
+
+Issue #252: Fixed SSL mining (stratum+tcps://)
+
+Issue #254 Fixed benchmark.
+
+Issue #253: Implemented stale share reduction for yespower, x25x, x22i, x21s,
+x16*, scryptn2, more to come.
+
+v3.12.6
+
+Issue #246: improved stale share detection for getwork.
+
+Improved precision of target_to_diff conversion from 4 digits to 20+.
+
+Display hash and target debug data for all rejected shares.
+
+A graphical representation of CPU affinity is displayed when using --threads.
+
+Added highest and lowest accepted share to summary log.
+
+Other small changes to logs to improve consistency and clarity.
+
+v3.12.5
+
+Issues #246 & #251: fixed incorrect share diff for stratum and getwork,
+fixed incorrect target diff for getwork. Stats should now be correct for
+getwork as well as stratum.
+
+Issue #252: Fixed stratum+tcps not using curl ssl.
+
+Getwork: reduce stale blocks, faster response to new work.
+
+Added ntime to new job/work logs.
+
+README.md now lists the parameters for yespower variations that don't have
+a specific algo name.
+
+v3.12.4.6
+
+Issue #246: fixed getwork repeated new block logs with same height. New work
+for the same block is now reported as "New work" instead of "New block".
+Also added a check that work is new before generating "New work" log.
+
+Added target diff to getwork new block log.
+
+Changed share ratio in share result log to simple fraction, no longer %.
+
+Added debug log to display mininginfo, use -D.
+
+v3.12.4.5
+
+Issue #246: better stale share detection for getwork, and enhanced logging
+of stale shares for stratum & getwork.
+
+Issue #251: fixed incorrect share difficulty and share ratio in share
+result log.
+
+Changed submit log to include share diff and block height.
+
+Small cosmetic changes to logs. 
+
+v3.12.4.4
+
+Issue #246: Fixed net hashrate in getwork block log,
+            removed duplicate getwork block log, 
+            other small tweaks to stats logs for getwork.
+
+Issue #248: Fixed chronic stale shares with scrypt:1048576 (scryptn2). 
+
+v3.12.4.3
+
+Fixed segfault in new block log for getwork.
+
+Disabled silent discarding of stale work after the submit is logged.
+
+v3.12.4.2
+
+Issue #245: fixed getwork stale shares, solo mining with getwork now works.
+
+Issue #246: implemented block and summary logs for getwork.
+
+v3.12.4.1
+
+Issue #245: fix scantime when mining solo with getwork.
+
+Added debug logs for creation of stratum and longpoll threads, use -D to
+enable.
+
+v3.12.4
+
+Issue #244: Change longpoll to ignore job id.
+
+Lyra2rev2 AVX2 +3%, AVX512 +6%.
+
+v3.12.3.1
+
+Issue #241: Fixed regression that broke coinbase address in v3.11.7.
+
+v3.12.3
+
+Issue #238: Fixed skunk AVX2.
+
+Issue #239: Faster AVX2 & AVX512 for skein +44%, skein2 +30%, plus marginal
+increases for skunk, x16r, x16rv2, x16rt, x16rt-veil, x16s, x21s.
+
+Faster anime VAES +57%, AVX512 +21%, AVX2 +3%.
+
+Redesigned code reponsible for #236.
+
+v3.12.2
+
+Fixed xevan, skein, skein2 AVX2, #238.
+
+Reversed polarity of AVX2 vector bit test utilities, and all users, to be
+logically and semantically correct. Follow up to issue #236. 
+
+v3.12.1
+
+Fixed anime AVX2 low difficulty shares, git issue #236.
+
+Periodic summary now reports lost hash rate due to rejected and stale shares,
+displayed only when non-zero.
+
+v3.12.0.1
+
+Fixed hodl rejects, git issue #237.
+
+Fixed debug code added in v3.12.0 to work with AVX2 to be enabled only
+after low difficulty share have been seen to avoid unnecessarily excessive
+log outout.
+
+Added more digits of precision to diff in log output to help diagnose
+low difficulty shares.
+
+v3.12.0
+
+Faster phi2 AVX2 +62%, AVX512 +150% on Intel CPUs. AMD Ryzen AVX2 is
+YMMV due to its inferiour AVX2 implementation.
+
+Fixed Hodl stats, rejects are still an issue since v3.9.5, git issue #237.
+
+API can now be enabled with "-b port" or "--api-bind port".
+It will use the default address 127.0.0.1.
+
+Editorial: Short form options should only be used on the command line to save
+typing. Configuration files and scripts should always use the long form
+"--api-bind addr:port" without relying on any defaults. This is a general
+recommendation that applies to all options for any application.
+
+Removed obsolete cryptonight, all variants, and supporting code for more
+size reduction and faster compiling.
+
+Tweaked the timing of the CPU temperature and frequency log (Linux only).
+
+Added some debug code to collect more info aboout low difficulty rejects,
+git issue #236.
+
+v3.11.9
+
+Fixed x16r invalid shares when Luffa was first in hash order.
+
+API is disabled by default.
+
+New startup message for status of stratum connection, API & extranonce.
+
+New log report for CPU temperature, frequency of fastest and slowest cores.
+
+Compile time is a little shorter and binary file size a little smaller
+using conditional compilation..
+
+Removed code for Bastion, Drop, Heavy, Luffa an Pluck algos and other unused
+code.
+
+v3.11.8
+
+Fixed network hashrate showing incorrect data, should be close now.
+
+Fixed compile errors when using GCC 10 with default flag -fno-common.
+
+Faster x16r, x16rv2, x16rt, x16s, x21s, veil, hex with midstate prehash.
+
+Decoupled sapling usage from block version 5 in yescryptr8g.
+
+More detailed data reporting for low difficulty rejected shares.
+
+v3.11.7
+
+Added yescryptr8g algo for KOTO, including support for block version 5.
+
+Added sha3d algo for BSHA3.
+
+Removed memcmp and clean_job checks from get_new_work, now only check job_id.
+
+Small improvement to sha512 and sha256 parallel implementations that don't
+use SHA.
+
+v3.11.6
+
+Fixed CPU temperature regression from v3.11.5.
+
+More improvements to share log. More compact, highlight incremented counter,
+block height when solved, job id when stale.
+
+v3.11.5
+
+Fixed AVX512 detection that could cause compilation errors on CPUs
+without AVX512.
+
+Fixed "BLOCK SOLVED" log incorrectly displaying "Accepted" when a block
+is solved.
+Added share counter to share submitited & accepted logs
+Added job id to share submitted log.
+Share submitted log is no longer highlighted blue, there was too much blue.
+
+Another CPU temperature fix for Linux.
+
+Added bug reporting tips to RELEASE NOTES.
+
+v3.11.4
+
+Fixed scrypt segfault since v3.9.9.1.
+
+Stale shares counted and reported seperately from other rejected shares.
+
+Display of counters for solved blocks, rejects, stale shares suppressed in
+periodic summary when zero.
+
+v3.11.3
+
+Fixed x12 AVX2 again.
+
+More speed for allium: AVX2 +4%, AVX512 +6%, VAES +14%.
+
+Restored lost speed for x22i & x25x.
+
+v3.11.2
+
+Fixed x11gost (sib) AVX2 invalid shares.
+
+Fixed x16r, x16rv2, x16s, x16rt, x16rt-veil (veil), x21s.
+No shares were submitted when cube, shavite or echo were the first function
+in the hash order.
+
+Fixed all algos reporting stats problems when mining with SSE2.
+
+Faster Lyra2 AVX512: lyra2z +47%, lyra2rev3 +11%, allium +13%, x21s +6% 
+
+Other minor performance improvements.
+
+Known issue:
+
+Lyra2 AVX512 improvements paradoxically reduced performance on x22i and x25x.
+https://github.com/JayDDee/cpuminer-opt/issues/225
+
+v3.11.1
+
+Faster panama for x25x AVX2 & AVX512.
+
+Fixed echo VAES for Xevan.
+
+Removed support for scryptjane algo.
+
+Reverted macro implemtations of hash functions to SPH reference code
+for SSE2 versions of algos.
+
+v3.11.0
+
+Fixed x25x AVX512 lane 4 invalid shares.
+
+AVX512 for hex, phi2.
+
+VAES optimzation for Intel Icelake CPUs for most algos recently optimized
+with AVX512, source code only.
+
+v3.10.7
+
+AVX512 for x25x, lbry, x13bcd (bcd).
+
+v3.10.6
+
+Added support for SSL stratum: stratum+tcps://
+
+Added job id reporting again, but leaner, suppressed with --quiet.
+
+AVX512 for x21s, x22i, lyra2z, allium.
+
+Fixed share overflow warnings mining lbry with Ryzen (SHA).
+
+v3.10.5
+
+AVX512 for x17, sonoa, xevan, hmq1725, lyra2rev3, lyra2rev2. 
+Faster hmq1725 AVX2.
+
+v3.10.4
+
+AVX512 for x16r, x16rv2, x16rt, x16s, x16rt-veil (veil).
+
+v3.10.3
+
+AVX512 for x12, x13, x14, x15.
+Fixed x12 AVX2 invalid shares.
+
+v.10.2
+
+AVX512 added for bmw512, c11, phi1612 (phi), qubit, skunk, x11, x11gost (sib).
+Fixed c11 AVX2 invalid shares.
+
+v3.10.1
+
+AVX512 for blake2b, nist5, quark, tribus.
+
+More broken lane fixes, fixed buffer overflow in skein AVX512, fixed
+quark invalid shares AVX2.
+
+Only the highest ranking feature in a class is listed at startup, lower ranking
+features are available but no longer listed.
+
+v3.10.0
+
+AVX512 is now supported on selected algos, Windows binary is now available.
+AVX512 optimizations are available for argon2d, blake2s, keccak, keccakc,
+skein & skein2.
+
+Fixed CPU temperature for some CPU models (Linux only).
+
+Fixed a bug that caused some lanes not to submit shares.
+
+Fixed some previously undetected buffer overflows.
+
+Lyra2rev2 3% faster SSE2 and AVX2.
+
+Added "-fno-asynchronous-unwind-tables" to AVX512 build script for Windows
+to fix known mingw issue.
+
+Changed AVX2 build script to explicitly add AES to address change in
+behaviour in GCC 9. 
+
+v3.9.11
+
+Added x22i & x25x algos.
+Blake2s 2% faster AVX2 with Intel CPU, slower with Ryzen v1, v2 ?
+
+v3.9.10
+
+Faster X* algos with AVX2.
+Small improvements to summary stats report.
+
+v3.9.9.1
+
+Fixed a day1 bug that could cause the miner to idle for up to 2 minutes
+under certain circumstances.
+
+Redesigned summary stats report now includes session statistics.
+
+More robust handling of statistics to reduce corruption.
+
+Removed --hide-diff option.
+
+Better handling of cpu-affinity with more than 64 CPUs.
+
+v3.9.9
+
+Added power2b algo for MicroBitcoin.
+Added generic yespower-b2b (yespower + blake2b) algo to be used with
+the parameters introduced in v3.9.7 for yespower & yescrypt.
+Display additional info when a share is rejected.
+Some low level enhancements and minor tweaking of log output.
+RELEASE_NOTES (this file) and README.md  added to Windows release package.
+
+v3.9.8.1
+
+Summary log report will be generated on stratum diff change or after 5 minutes,
+whichever comes first, to prevent incorrect data in the report.
+
+Removed phi2-lux alias (introduced in v3.9.8) due to Luxcoin's planned fork
+to a new algo. The new Luxcoin algo is not supported by cpuminer-opt.
+Until the fork Luxcoin can be mined using phi2 algo.
+
+--hide-diff option is deprecated and has no effect. It will be removed in a
+future release.
+
+v3.9.8
+
+Changes to log output to provide data more relevant to actual mining
+performance.
+phi2 can now handle pools with a mix of coins that use and don't use roots.
+phi2-lux added as an alias for phi2 as they are identical except for roots.
+Add x16rv2 algo for Ravencoin fork.
+
+v3.9.7
+
+Command line option changes:
+
+"-R" is no longer used as a shortcut for "--retry-pause", users must
+use the long option.
+
+New options:
+
+-N, --param-n: set the N parameter for yescrypt, yespower or scrypt algos
+-R, --param-r: set the R parameter for yescrypt or yespower algos, scrypt is
+     hardcoded with R=1
+-K, --param-key: set the client key/pers parameter for yescrypt/yespower algos.
+
+These options can be used to mine yescrypt or yespower variations using
+the generic yescrypt or yespower algo name and specifying the parameters
+manually. They can even be used to mine variations that aren't formally
+supported by a unique algo name. Existing algos can continue to to be mined
+using their original name without parameters.
+
 v3.9.6.2

 New algo blake2b.
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.15.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.16.1 -*- Autoconf -*-

-# Copyright (C) 1996-2017 Free Software Foundation, Inc.
+# Copyright (C) 1996-2018 Free Software Foundation, Inc.

 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])

-# Copyright (C) 2002-2017 Free Software Foundation, Inc.
+# Copyright (C) 2002-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.15'
+[am__api_version='1.16'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.15.1], [],
+m4_if([$1], [1.16.1], [],
      [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])

@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.15.1])dnl
+[AM_AUTOMAKE_VERSION([1.16.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
  [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])

 # Figure out how to run the assembler.                      -*- Autoconf -*-

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -78,7 +78,7 @@ _AM_IF_OPTION([no-dependencies],, [_AM_DEPENDENCIES([CCAS])])dnl

 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -130,7 +130,7 @@ am_aux_dir=`cd "$ac_aux_dir" && pwd`

 # AM_CONDITIONAL                                            -*- Autoconf -*-

-# Copyright (C) 1997-2017 Free Software Foundation, Inc.
+# Copyright (C) 1997-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -161,7 +161,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])

-# Copyright (C) 1999-2017 Free Software Foundation, Inc.
+# Copyright (C) 1999-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -352,13 +352,12 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl

 # Generate code to set up dependency tracking.              -*- Autoconf -*-

-# Copyright (C) 1999-2017 Free Software Foundation, Inc.
+# Copyright (C) 1999-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.

-
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
@@ -366,49 +365,41 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
  # Older Autoconf quotes --file arguments for eval, but not when files
  # are listed without --file.  Let's play safe and only enable the eval
  # if we detect the quoting.
-  case $CONFIG_FILES in
-  *\'*) eval set x "$CONFIG_FILES" ;;
-  *)   set x $CONFIG_FILES ;;
-  esac
+  # TODO: see whether this extra hack can be removed once we start
+  # requiring Autoconf 2.70 or later.
+  AS_CASE([$CONFIG_FILES],
+          [*\'*], [eval set x "$CONFIG_FILES"],
+          [*], [set x $CONFIG_FILES])
  shift
-  for mf
+  # Used to flag and report bootstrapping failures.
+  am_rc=0
+  for am_mf
  do
    # Strip MF so we end up with the name of the file.
-    mf=`echo "$mf" | sed -e 's/:.*$//'`
-    # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
-    # some people rename them; so instead we look at the file content.
-    # Grep'ing the first line is not enough: some people post-process
-    # each Makefile.in and add a new line on top of each file to say so.
-    # Grep'ing the whole file is not good either: AIX grep has a line
+    am_mf=`AS_ECHO(["$am_mf"]) | sed -e 's/:.*$//'`
+    # Check whether this is an Automake generated Makefile which includes
+    # dependency-tracking related rules and includes.
+    # Grep'ing the whole file directly is not great: AIX grep has a line
    # limit of 2048, but all sed's we know have understand at least 4000.
-    if sed -n 's,^#.*generated by automake.*,X,p' "$mf" | grep X >/dev/null 2>&1; then
-      dirpart=`AS_DIRNAME("$mf")`
-    else
-      continue
-    fi
-    # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
-    DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
-    test -z "$DEPDIR" && continue
-    am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
-    am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # Find all dependency output files, they are included files with
-    # $(DEPDIR) in their names.  We invoke sed twice because it is the
-    # simplest approach to changing $(DEPDIR) to its actual value in the
-    # expansion.
-    for file in `sed -n "
-      s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
-      # Make sure the directory exists.
-      test -f "$dirpart/$file" && continue
-      fdir=`AS_DIRNAME(["$file"])`
-      AS_MKDIR_P([$dirpart/$fdir])
-      # echo "creating $dirpart/$file"
-      echo '# dummy' > "$dirpart/$file"
-    done
+    sed -n 's,^am--depfiles:.*,X,p' "$am_mf" | grep X >/dev/null 2>&1 \
+      || continue
+    am_dirpart=`AS_DIRNAME(["$am_mf"])`
+    am_filepart=`AS_BASENAME(["$am_mf"])`
+    AM_RUN_LOG([cd "$am_dirpart" \
+      && sed -e '/# am--include-marker/d' "$am_filepart" \
+        | $MAKE -f - am--depfiles]) || am_rc=$?
  done
+  if test $am_rc -ne 0; then
+    AC_MSG_FAILURE([Something went wrong bootstrapping makefile fragments
+    for automatic dependency tracking.  Try re-running configure with the
+    '--disable-dependency-tracking' option to at least be able to build
+    the package (albeit without support for automatic dependency tracking).])
+  fi
+  AS_UNSET([am_dirpart])
+  AS_UNSET([am_filepart])
+  AS_UNSET([am_mf])
+  AS_UNSET([am_rc])
+  rm -f conftest-deps.mk
 }
 ])# _AM_OUTPUT_DEPENDENCY_COMMANDS

@@ -417,18 +408,17 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # -----------------------------
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
-# This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each '.P' file that we will
-# need in order to bootstrap the dependency handling code.
+# This code is only required when automatic dependency tracking is enabled.
+# This creates each '.Po' and '.Plo' makefile fragment that we'll need in
+# order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
     [test x"$AMDEP_TRUE" != x"" || _AM_OUTPUT_DEPENDENCY_COMMANDS],
-     [AMDEP_TRUE="$AMDEP_TRUE" ac_aux_dir="$ac_aux_dir"])
-])
+     [AMDEP_TRUE="$AMDEP_TRUE" MAKE="${MAKE-make}"])])

 # Do all the work for Automake.                             -*- Autoconf -*-

-# Copyright (C) 1996-2017 Free Software Foundation, Inc.
+# Copyright (C) 1996-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -515,8 +505,8 @@ AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
 AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # For better backward compatibility.  To be removed once Automake 1.9.x
 # dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+# <https://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <https://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
 # We need awk for the "check" target (and possibly the TAP driver).  The
 # system "awk" is bad on some platforms.
@@ -583,7 +573,7 @@ END
 Aborting the configuration process, to ensure you take notice of the issue.

 You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
+that behaves properly: <https://www.gnu.org/software/coreutils/>.

 If you want to complete the configuration process using your problematic
 'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
@@ -625,7 +615,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -646,7 +636,7 @@ if test x"${install_sh+set}" != xset; then
 fi
 AC_SUBST([install_sh])])

-# Copyright (C) 2003-2017 Free Software Foundation, Inc.
+# Copyright (C) 2003-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -668,7 +658,7 @@ AC_SUBST([am__leading_dot])])
 # Add --enable-maintainer-mode option to configure.         -*- Autoconf -*-
 # From Jim Meyering

-# Copyright (C) 1996-2017 Free Software Foundation, Inc.
+# Copyright (C) 1996-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -703,7 +693,7 @@ AC_MSG_CHECKING([whether to enable maintainer-specific portions of Makefiles])

 # Check to see how 'make' treats includes.	            -*- Autoconf -*-

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -711,49 +701,42 @@ AC_MSG_CHECKING([whether to enable maintainer-specific portions of Makefiles])

 # AM_MAKE_INCLUDE()
 # -----------------
-# Check to see how make treats includes.
+# Check whether make has an 'include' directive that can support all
+# the idioms we need for our automatic dependency tracking code.
 AC_DEFUN([AM_MAKE_INCLUDE],
-[am_make=${MAKE-make}
-cat > confinc << 'END'
+[AC_MSG_CHECKING([whether ${MAKE-make} supports the include directive])
+cat > confinc.mk << 'END'
 am__doit:
-	@echo this is the am__doit target
+	@echo this is the am__doit target >confinc.out
 .PHONY: am__doit
 END
-# If we don't find an include directive, just comment out the code.
-AC_MSG_CHECKING([for style of include used by $am_make])
 am__include="#"
 am__quote=
-_am_result=none
-# First try GNU make style include.
-echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
-case `$am_make -s -f confmf 2> /dev/null` in #(
-*the\ am__doit\ target*)
-  am__include=include
-  am__quote=
-  _am_result=GNU
-  ;;
-esac
-# Now try BSD make style include.
-if test "$am__include" = "#"; then
-   echo '.include "confinc"' > confmf
-   case `$am_make -s -f confmf 2> /dev/null` in #(
-   *the\ am__doit\ target*)
-     am__include=.include
-     am__quote="\""
-     _am_result=BSD
-     ;;
-   esac
-fi
-AC_SUBST([am__include])
-AC_SUBST([am__quote])
-AC_MSG_RESULT([$_am_result])
-rm -f confinc confmf
-])
+# BSD make does it like this.
+echo '.include "confinc.mk" # ignored' > confmf.BSD
+# Other make implementations (GNU, Solaris 10, AIX) do it like this.
+echo 'include confinc.mk # ignored' > confmf.GNU
+_am_result=no
+for s in GNU BSD; do
+  AM_RUN_LOG([${MAKE-make} -f confmf.$s && cat confinc.out])
+  AS_CASE([$?:`cat confinc.out 2>/dev/null`],
+      ['0:this is the am__doit target'],
+      [AS_CASE([$s],
+          [BSD], [am__include='.include' am__quote='"'],
+          [am__include='include' am__quote=''])])
+  if test "$am__include" != "#"; then
+    _am_result="yes ($s style)"
+    break
+  fi
+done
+rm -f confinc.* confmf.*
+AC_MSG_RESULT([${_am_result}])
+AC_SUBST([am__include])])
+AC_SUBST([am__quote])])

 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-

-# Copyright (C) 1997-2017 Free Software Foundation, Inc.
+# Copyright (C) 1997-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -792,7 +775,7 @@ fi

 # Helper functions for option handling.                     -*- Autoconf -*-

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -821,7 +804,7 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])

-# Copyright (C) 1999-2017 Free Software Foundation, Inc.
+# Copyright (C) 1999-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -868,7 +851,7 @@ AC_LANG_POP([C])])
 # For backward compatibility.
 AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -887,7 +870,7 @@ AC_DEFUN([AM_RUN_LOG],

 # Check to make sure that the build environment is sane.    -*- Autoconf -*-

-# Copyright (C) 1996-2017 Free Software Foundation, Inc.
+# Copyright (C) 1996-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -968,7 +951,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])

-# Copyright (C) 2009-2017 Free Software Foundation, Inc.
+# Copyright (C) 2009-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1028,7 +1011,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])

-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1056,7 +1039,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])

-# Copyright (C) 2006-2017 Free Software Foundation, Inc.
+# Copyright (C) 2006-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1075,7 +1058,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])

 # Check how to create a tarball.                            -*- Autoconf -*-

-# Copyright (C) 2004-2017 Free Software Foundation, Inc.
+# Copyright (C) 2004-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
--- a/algo-gate-api.c
+++ b/algo-gate-api.c
@@ -15,8 +15,6 @@
 #include <stdbool.h>
 #include <memory.h>
 #include <unistd.h>
-#include <openssl/sha.h>
-//#include "miner.h"
 #include "algo-gate-api.h"

 // Define null and standard functions.
@@ -90,39 +88,174 @@ void algo_not_implemented()
 }

 // default null functions
-
+// deprecated, use generic as default
 int null_scanhash()
 {
   applog(LOG_WARNING,"SWERR: undefined scanhash function in algo_gate");
   return 0;
 }

-void null_hash()
+// Default generic scanhash can be used in many cases.
+int scanhash_generic( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t edata[20] __attribute__((aligned(64)));
+   uint32_t hash[8] __attribute__((aligned(64)));
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 1;
+   uint32_t n = first_nonce;
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+
+   mm128_bswap32_80( edata, pdata );
+   do
+   {
+      edata[19] = n;
+      if ( likely( algo_gate.hash( hash, edata, thr_id ) ) )
+      if ( unlikely( valid_hash( hash, ptarget ) && !bench ) )
+      {
+         pdata[19] = bswap_32( n );
+         submit_solution( work, hash, mythr );
+      }
+      n++;
+   } while ( n < last_nonce && !work_restart[thr_id].restart );
+   *hashes_done = n - first_nonce;
+   pdata[19] = n;
+   return 0;
+}
+
+#if defined(__AVX2__)
+
+//int scanhash_4way_64_64( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+//int scanhash_4way_64_640( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+int scanhash_4way_64in_32out( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t hash32[8*4] __attribute__ ((aligned (64)));
+   uint32_t vdata[20*4] __attribute__ ((aligned (64)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   uint32_t *hash32_d7 = &(hash32[ 7*4 ]);
+   uint32_t *pdata = work->data;
+   const uint32_t *ptarget = work->target;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 4;
+   __m256i  *noncev = (__m256i*)vdata + 9;
+   uint32_t n = first_nonce;
+   const int thr_id = mythr->id;
+   const uint32_t targ32_d7 = ptarget[7];
+   const bool bench = opt_benchmark;
+
+   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+   *noncev = mm256_intrlv_blend_32(
+                   _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
+   do
+   {
+      if ( likely( algo_gate.hash( hash32, vdata, thr_id ) ) )
+      for ( int lane = 0; lane < 4; lane++ )
+      if ( unlikely( hash32_d7[ lane ] <= targ32_d7 && !bench ) )
+      {
+         extr_lane_4x32( lane_hash, hash32, lane, 256 );
+         if ( valid_hash( lane_hash, ptarget ) )
+         {
+            pdata[19] = bswap_32( n + lane );
+            submit_solution( work, lane_hash, mythr );
+         }
+      }
+      *noncev = _mm256_add_epi32( *noncev,
+                                  m256_const1_64( 0x0000000400000000 ) );
+      n += 4;
+   } while ( likely( ( n <= last_nonce ) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
+//int scanhash_8way_32_32( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+#endif
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+//int scanhash_8way_64_64( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+//int scanhash_8way_64_640( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+int scanhash_8way_64in_32out( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t hash32[8*8] __attribute__ ((aligned (128)));
+   uint32_t vdata[20*8] __attribute__ ((aligned (64)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   uint32_t *hash32_d7 = &(hash32[7*8]);
+   uint32_t *pdata = work->data;
+   const uint32_t *ptarget = work->target;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 8;
+   __m512i  *noncev = (__m512i*)vdata + 9;
+   uint32_t n = first_nonce;
+   const int thr_id = mythr->id;
+   const uint32_t targ32_d7 = ptarget[7];
+   const bool bench = opt_benchmark;
+
+   mm512_bswap32_intrlv80_8x64( vdata, pdata );
+   *noncev = mm512_intrlv_blend_32(
+              _mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
+                                n+3, 0, n+2, 0, n+1, 0, n,   0 ), *noncev );
+   do
+   {
+      if ( likely( algo_gate.hash( hash32, vdata, thr_id ) ) )
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( ( hash32_d7[ lane ] <= targ32_d7 ) && !bench ) )
+      {
+         extr_lane_8x32( lane_hash, hash32, lane, 256 );
+         if ( likely( valid_hash( lane_hash, ptarget ) ) )
+         {
+            pdata[19] = bswap_32( n + lane );
+            submit_solution( work, lane_hash, mythr );
+         }
+      }
+      *noncev = _mm512_add_epi32( *noncev,
+                                  m512_const1_64( 0x0000000800000000 ) );
+      n += 8;
+   } while ( likely( ( n < last_nonce ) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
+//int scanhash_16way_32_32( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr )
+
+#endif
+
+
+
+int null_hash()
 {
   applog(LOG_WARNING,"SWERR: null_hash unsafe null function");
-};
-void null_hash_suw()
-{
-  applog(LOG_WARNING,"SWERR: null_hash_suw unsafe null function");
+   return 0;
 };

 void init_algo_gate( algo_gate_t* gate )
 {
   gate->miner_thread_init       = (void*)&return_true;
-   gate->scanhash                = (void*)&null_scanhash;
+   gate->scanhash                = (void*)&scanhash_generic;
   gate->hash                    = (void*)&null_hash;
-   gate->hash_suw                = (void*)&null_hash_suw;
   gate->get_new_work            = (void*)&std_get_new_work;
-   gate->get_nonceptr            = (void*)&std_get_nonceptr;
   gate->work_decode             = (void*)&std_le_work_decode;
   gate->decode_extra_data       = (void*)&do_nothing;
-   gate->wait_for_diff           = (void*)&std_wait_for_diff;
-   gate->get_max64               = (void*)&get_max64_0x1fffffLL;
   gate->gen_merkle_root         = (void*)&sha256d_gen_merkle_root;
-   gate->stratum_gen_work        = (void*)&std_stratum_gen_work;
   gate->build_stratum_request   = (void*)&std_le_build_stratum_request;
   gate->malloc_txs_request      = (void*)&std_malloc_txs_request;
-   gate->set_target              = (void*)&std_set_target;
   gate->submit_getwork_result   = (void*)&std_le_submit_getwork_result;
   gate->build_block_header      = (void*)&std_build_block_header;
   gate->build_extraheader       = (void*)&std_build_extraheader;
@@ -132,7 +265,6 @@ void init_algo_gate( algo_gate_t* gate )
   gate->resync_threads          = (void*)&do_nothing;
   gate->do_this_thread          = (void*)&return_true;
   gate->longpoll_rpc_call       = (void*)&std_longpoll_rpc_call;
-   gate->stratum_handle_response = (void*)&std_stratum_handle_response;
   gate->get_work_data_size      = (void*)&std_get_work_data_size;
   gate->optimizations           = EMPTY_SET;
   gate->ntime_index             = STD_NTIME_INDEX;
@@ -145,9 +277,11 @@ void init_algo_gate( algo_gate_t* gate )
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wimplicit-function-declaration"

-// called by each thread that uses the gate
+// Called once by main
 bool register_algo_gate( int algo, algo_gate_t *gate )
 {
+  bool rc = false;
+
  if ( NULL == gate )
  {
    applog(LOG_ERR,"FAIL: algo_gate registration failed, NULL gate\n");
@@ -156,111 +290,108 @@ bool register_algo_gate( int algo, algo_gate_t *gate )

  init_algo_gate( gate );

-  switch (algo)
+  switch ( algo )
  {
-    case ALGO_ALLIUM:        register_allium_algo        ( gate ); break;
-    case ALGO_ANIME:         register_anime_algo         ( gate ); break;
-    case ALGO_ARGON2:        register_argon2_algo        ( gate ); break;
-    case ALGO_ARGON2D250:    register_argon2d_crds_algo  ( gate ); break;
-    case ALGO_ARGON2D500:    register_argon2d_dyn_algo   ( gate ); break;
-    case ALGO_ARGON2D4096:   register_argon2d4096_algo   ( gate ); break;
-    case ALGO_AXIOM:         register_axiom_algo         ( gate ); break;
-    case ALGO_BASTION:       register_bastion_algo       ( gate ); break;
-    case ALGO_BLAKE:         register_blake_algo         ( gate ); break;
-    case ALGO_BLAKE2B:       register_blake2b_algo       ( gate ); break;
-    case ALGO_BLAKE2S:       register_blake2s_algo       ( gate ); break;
-    case ALGO_BLAKECOIN:     register_blakecoin_algo     ( gate ); break;
-    case ALGO_BMW512:        register_bmw512_algo        ( gate ); break;
-    case ALGO_C11:           register_c11_algo           ( gate ); break;
-    case ALGO_CRYPTOLIGHT:   register_cryptolight_algo   ( gate ); break;
-    case ALGO_CRYPTONIGHT:   register_cryptonight_algo   ( gate ); break;
-    case ALGO_CRYPTONIGHTV7: register_cryptonightv7_algo ( gate ); break;
-    case ALGO_DECRED:        register_decred_algo        ( gate ); break;
-    case ALGO_DEEP:          register_deep_algo          ( gate ); break;
-    case ALGO_DMD_GR:        register_dmd_gr_algo        ( gate ); break;
-    case ALGO_DROP:          register_drop_algo          ( gate ); break;
-    case ALGO_FRESH:         register_fresh_algo         ( gate ); break;
-    case ALGO_GROESTL:       register_groestl_algo       ( gate ); break;
-    case ALGO_HEAVY:         register_heavy_algo         ( gate ); break;
-    case ALGO_HEX:           register_hex_algo           ( gate ); break;
-    case ALGO_HMQ1725:       register_hmq1725_algo       ( gate ); break;
-    case ALGO_HODL:          register_hodl_algo          ( gate ); break;
-    case ALGO_JHA:           register_jha_algo           ( gate ); break;
-    case ALGO_KECCAK:        register_keccak_algo        ( gate ); break;
-    case ALGO_KECCAKC:       register_keccakc_algo       ( gate ); break;
-    case ALGO_LBRY:          register_lbry_algo          ( gate ); break;
-    case ALGO_LUFFA:         register_luffa_algo         ( gate ); break;
-    case ALGO_LYRA2H:        register_lyra2h_algo        ( gate ); break;
-    case ALGO_LYRA2RE:       register_lyra2re_algo       ( gate ); break;
-    case ALGO_LYRA2REV2:     register_lyra2rev2_algo     ( gate ); break;
-    case ALGO_LYRA2REV3:     register_lyra2rev3_algo     ( gate ); break;
-    case ALGO_LYRA2Z:        register_lyra2z_algo        ( gate ); break;
-    case ALGO_LYRA2Z330:     register_lyra2z330_algo     ( gate ); break;
-    case ALGO_M7M:           register_m7m_algo           ( gate ); break;
-    case ALGO_MYR_GR:        register_myriad_algo        ( gate ); break;
-    case ALGO_NEOSCRYPT:     register_neoscrypt_algo     ( gate ); break;
-    case ALGO_NIST5:         register_nist5_algo         ( gate ); break;
-    case ALGO_PENTABLAKE:    register_pentablake_algo    ( gate ); break;
-    case ALGO_PHI1612:       register_phi1612_algo       ( gate ); break;
-    case ALGO_PHI2:          register_phi2_algo          ( gate ); break;
-    case ALGO_PLUCK:         register_pluck_algo         ( gate ); break;
-    case ALGO_POLYTIMOS:     register_polytimos_algo     ( gate ); break;
-    case ALGO_QUARK:         register_quark_algo         ( gate ); break;
-    case ALGO_QUBIT:         register_qubit_algo         ( gate ); break;
-    case ALGO_SCRYPT:        register_scrypt_algo        ( gate ); break;
-    case ALGO_SCRYPTJANE:    register_scryptjane_algo    ( gate ); break;
-    case ALGO_SHA256D:       register_sha256d_algo       ( gate ); break;
-    case ALGO_SHA256Q:       register_sha256q_algo       ( gate ); break;
-    case ALGO_SHA256T:       register_sha256t_algo       ( gate ); break;
-    case ALGO_SHAVITE3:      register_shavite_algo       ( gate ); break;
-    case ALGO_SKEIN:         register_skein_algo         ( gate ); break;
-    case ALGO_SKEIN2:        register_skein2_algo        ( gate ); break;
-    case ALGO_SKUNK:         register_skunk_algo         ( gate ); break;
-    case ALGO_SONOA:         register_sonoa_algo         ( gate ); break;
-    case ALGO_TIMETRAVEL:    register_timetravel_algo    ( gate ); break;
-    case ALGO_TIMETRAVEL10:  register_timetravel10_algo  ( gate ); break;
-    case ALGO_TRIBUS:        register_tribus_algo        ( gate ); break;
-    case ALGO_VANILLA:       register_vanilla_algo       ( gate ); break;
-    case ALGO_VELTOR:        register_veltor_algo        ( gate ); break;
-    case ALGO_WHIRLPOOL:     register_whirlpool_algo     ( gate ); break;
-    case ALGO_WHIRLPOOLX:    register_whirlpoolx_algo    ( gate ); break;
-    case ALGO_X11:           register_x11_algo           ( gate ); break;
-    case ALGO_X11EVO:        register_x11evo_algo        ( gate ); break;
-    case ALGO_X11GOST:       register_x11gost_algo       ( gate ); break;
-    case ALGO_X12:           register_x12_algo           ( gate ); break;
-    case ALGO_X13:           register_x13_algo           ( gate ); break;
-    case ALGO_X13BCD:        register_x13bcd_algo        ( gate ); break;
-    case ALGO_X13SM3:        register_x13sm3_algo        ( gate ); break;
-    case ALGO_X14:           register_x14_algo           ( gate ); break;
-    case ALGO_X15:           register_x15_algo           ( gate ); break;
-    case ALGO_X16R:          register_x16r_algo          ( gate ); break;
-    case ALGO_X16RT:         register_x16rt_algo         ( gate ); break;
-    case ALGO_X16RT_VEIL:    register_x16rt_veil_algo    ( gate ); break;
-    case ALGO_X16S:          register_x16s_algo          ( gate ); break;
-    case ALGO_X17:           register_x17_algo           ( gate ); break;
-    case ALGO_X21S:          register_x21s_algo          ( gate ); break;
-    case ALGO_XEVAN:         register_xevan_algo         ( gate ); break;
-/*    case ALGO_YESCRYPT:     register_yescrypt_05_algo     ( gate ); break;
-     case ALGO_YESCRYPTR8:   register_yescryptr8_05_algo   ( gate ); break;
-     case ALGO_YESCRYPTR16:  register_yescryptr16_05_algo  ( gate ); break;
-     case ALGO_YESCRYPTR32:  register_yescryptr32_05_algo  ( gate ); break;
-*/
-    case ALGO_YESCRYPT:      register_yescrypt_algo      ( gate ); break;
-    case ALGO_YESCRYPTR8:    register_yescryptr8_algo    ( gate ); break;
-    case ALGO_YESCRYPTR16:   register_yescryptr16_algo   ( gate ); break;
-    case ALGO_YESCRYPTR32:   register_yescryptr32_algo   ( gate ); break;
-    case ALGO_YESPOWER:      register_yespower_algo      ( gate ); break;
-    case ALGO_YESPOWERR16:   register_yespowerr16_algo   ( gate ); break;
-    case ALGO_ZR5:           register_zr5_algo           ( gate ); break;
+    case ALGO_ALLIUM:       rc = register_allium_algo        ( gate ); break;
+    case ALGO_ANIME:        rc = register_anime_algo         ( gate ); break;
+    case ALGO_ARGON2:       rc = register_argon2_algo        ( gate ); break;
+    case ALGO_ARGON2D250:   rc = register_argon2d_crds_algo  ( gate ); break;
+    case ALGO_ARGON2D500:   rc = register_argon2d_dyn_algo   ( gate ); break;
+    case ALGO_ARGON2D4096:  rc = register_argon2d4096_algo   ( gate ); break;
+    case ALGO_AXIOM:        rc = register_axiom_algo         ( gate ); break;
+    case ALGO_BLAKE:        rc = register_blake_algo         ( gate ); break;
+    case ALGO_BLAKE2B:      rc = register_blake2b_algo       ( gate ); break;
+    case ALGO_BLAKE2S:      rc = register_blake2s_algo       ( gate ); break;
+    case ALGO_BLAKECOIN:    rc = register_blakecoin_algo     ( gate ); break;
+    case ALGO_BMW512:       rc = register_bmw512_algo        ( gate ); break;
+    case ALGO_C11:          rc = register_c11_algo           ( gate ); break;
+    case ALGO_DECRED:       rc = register_decred_algo        ( gate ); break;
+    case ALGO_DEEP:         rc = register_deep_algo          ( gate ); break;
+    case ALGO_DMD_GR:       rc = register_dmd_gr_algo        ( gate ); break;
+    case ALGO_GROESTL:      rc = register_groestl_algo       ( gate ); break;
+    case ALGO_HEX:          rc = register_hex_algo           ( gate ); break;
+    case ALGO_HMQ1725:      rc = register_hmq1725_algo       ( gate ); break;
+    case ALGO_HODL:         rc = register_hodl_algo          ( gate ); break;
+    case ALGO_JHA:          rc = register_jha_algo           ( gate ); break;
+    case ALGO_KECCAK:       rc = register_keccak_algo        ( gate ); break;
+    case ALGO_KECCAKC:      rc = register_keccakc_algo       ( gate ); break;
+    case ALGO_LBRY:         rc = register_lbry_algo          ( gate ); break;
+    case ALGO_LYRA2H:       rc = register_lyra2h_algo        ( gate ); break;
+    case ALGO_LYRA2RE:      rc = register_lyra2re_algo       ( gate ); break;
+    case ALGO_LYRA2REV2:    rc = register_lyra2rev2_algo     ( gate ); break;
+    case ALGO_LYRA2REV3:    rc = register_lyra2rev3_algo     ( gate ); break;
+    case ALGO_LYRA2Z:       rc = register_lyra2z_algo        ( gate ); break;
+    case ALGO_LYRA2Z330:    rc = register_lyra2z330_algo     ( gate ); break;
+    case ALGO_M7M:          rc = register_m7m_algo           ( gate ); break;
+    case ALGO_MINOTAUR:     rc = register_minotaur_algo      ( gate ); break;
+    case ALGO_MYR_GR:       rc = register_myriad_algo        ( gate ); break;
+    case ALGO_NEOSCRYPT:    rc = register_neoscrypt_algo     ( gate ); break;
+    case ALGO_NIST5:        rc = register_nist5_algo         ( gate ); break;
+    case ALGO_PENTABLAKE:   rc = register_pentablake_algo    ( gate ); break;
+    case ALGO_PHI1612:      rc = register_phi1612_algo       ( gate ); break;
+    case ALGO_PHI2:         rc = register_phi2_algo          ( gate ); break;
+    case ALGO_POLYTIMOS:    rc = register_polytimos_algo     ( gate ); break;
+    case ALGO_POWER2B:      rc = register_power2b_algo       ( gate ); break;
+    case ALGO_QUARK:        rc = register_quark_algo         ( gate ); break;
+    case ALGO_QUBIT:        rc = register_qubit_algo         ( gate ); break;
+    case ALGO_SCRYPT:       rc = register_scrypt_algo        ( gate ); break;
+    case ALGO_SHA256D:      rc = register_sha256d_algo       ( gate ); break;
+    case ALGO_SHA256Q:      rc = register_sha256q_algo       ( gate ); break;
+    case ALGO_SHA256T:      rc = register_sha256t_algo       ( gate ); break;
+    case ALGO_SHA3D:        rc = register_sha3d_algo         ( gate ); break;
+    case ALGO_SHAVITE3:     rc = register_shavite_algo       ( gate ); break;
+    case ALGO_SKEIN:        rc = register_skein_algo         ( gate ); break;
+    case ALGO_SKEIN2:       rc = register_skein2_algo        ( gate ); break;
+    case ALGO_SKUNK:        rc = register_skunk_algo         ( gate ); break;
+    case ALGO_SONOA:        rc = register_sonoa_algo         ( gate ); break;
+    case ALGO_TIMETRAVEL:   rc = register_timetravel_algo    ( gate ); break;
+    case ALGO_TIMETRAVEL10: rc = register_timetravel10_algo  ( gate ); break;
+    case ALGO_TRIBUS:       rc = register_tribus_algo        ( gate ); break;
+    case ALGO_VANILLA:      rc = register_vanilla_algo       ( gate ); break;
+    case ALGO_VELTOR:       rc = register_veltor_algo        ( gate ); break;
+    case ALGO_VERTHASH:     rc = register_verthash_algo      ( gate ); break;
+    case ALGO_WHIRLPOOL:    rc = register_whirlpool_algo     ( gate ); break;
+    case ALGO_WHIRLPOOLX:   rc = register_whirlpoolx_algo    ( gate ); break;
+    case ALGO_X11:          rc = register_x11_algo           ( gate ); break;
+    case ALGO_X11EVO:       rc = register_x11evo_algo        ( gate ); break;
+    case ALGO_X11GOST:      rc = register_x11gost_algo       ( gate ); break;
+    case ALGO_X12:          rc = register_x12_algo           ( gate ); break;
+    case ALGO_X13:          rc = register_x13_algo           ( gate ); break;
+    case ALGO_X13BCD:       rc = register_x13bcd_algo        ( gate ); break;
+    case ALGO_X13SM3:       rc = register_x13sm3_algo        ( gate ); break;
+    case ALGO_X14:          rc = register_x14_algo           ( gate ); break;
+    case ALGO_X15:          rc = register_x15_algo           ( gate ); break;
+    case ALGO_X16R:         rc = register_x16r_algo          ( gate ); break;
+    case ALGO_X16RV2:       rc = register_x16rv2_algo        ( gate ); break;
+    case ALGO_X16RT:        rc = register_x16rt_algo         ( gate ); break;
+    case ALGO_X16RT_VEIL:   rc = register_x16rt_veil_algo    ( gate ); break;
+    case ALGO_X16S:         rc = register_x16s_algo          ( gate ); break;
+    case ALGO_X17:          rc = register_x17_algo           ( gate ); break;
+    case ALGO_X21S:         rc = register_x21s_algo          ( gate ); break;
+    case ALGO_X22I:         rc = register_x22i_algo          ( gate ); break;
+    case ALGO_X25X:         rc = register_x25x_algo          ( gate ); break;
+    case ALGO_XEVAN:        rc = register_xevan_algo         ( gate ); break;
+    case ALGO_YESCRYPT:     rc = register_yescrypt_05_algo   ( gate ); break;
+//    case ALGO_YESCRYPT:      register_yescrypt_algo      ( gate ); break;
+    case ALGO_YESCRYPTR8:   rc = register_yescryptr8_05_algo ( gate ); break;
+//    case ALGO_YESCRYPTR8:    register_yescryptr8_algo    ( gate ); break;
+    case ALGO_YESCRYPTR8G:  rc = register_yescryptr8g_algo   ( gate ); break;
+    case ALGO_YESCRYPTR16:  rc = register_yescryptr16_05_algo( gate ); break;
+//    case ALGO_YESCRYPTR16:   register_yescryptr16_algo   ( gate ); break;
+    case ALGO_YESCRYPTR32:  rc = register_yescryptr32_05_algo( gate ); break;
+//    case ALGO_YESCRYPTR32:   register_yescryptr32_algo   ( gate ); break;
+    case ALGO_YESPOWER:     rc = register_yespower_algo      ( gate ); break;
+    case ALGO_YESPOWERR16:  rc = register_yespowerr16_algo   ( gate ); break;
+    case ALGO_YESPOWER_B2B: rc = register_yespower_b2b_algo  ( gate ); break;
+    case ALGO_ZR5:          rc = register_zr5_algo           ( gate ); break;
   default:
-      applog(LOG_ERR,"FAIL: algo_gate registration failed, unknown algo %s.\n", algo_names[opt_algo] );
+      applog(LOG_ERR,"BUG: unregistered algorithm %s.\n", algo_names[opt_algo] );
      return false;
  } // switch

- // ensure required functions were defined.
-  if (  gate->scanhash == (void*)&null_scanhash )
+  if ( !rc )
  {
-    applog(LOG_ERR, "FAIL: Required algo_gate functions undefined\n");
+    applog(LOG_ERR, "FAIL: %s algorithm failed to initialize\n", algo_names[opt_algo] );
    return false;
  }
  return true;
@@ -269,30 +400,6 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
 // restore warnings
 #pragma GCC diagnostic pop

-// override std defaults with jr2 defaults
-bool register_json_rpc2( algo_gate_t *gate )
-{
-  applog(LOG_WARNING,"\nCryptonight algorithm and variants are no longer");
-  applog(LOG_WARNING,"supported by cpuminer-opt. Shares submitted will");
-  applog(LOG_WARNING,"likely be rejected. Proceed at your own risk.\n");
-
-  gate->wait_for_diff           = (void*)&do_nothing;
-  gate->get_new_work            = (void*)&jr2_get_new_work;
-  gate->get_nonceptr            = (void*)&jr2_get_nonceptr;
-  gate->stratum_gen_work        = (void*)&jr2_stratum_gen_work;
-  gate->build_stratum_request   = (void*)&jr2_build_stratum_request;
-  gate->submit_getwork_result   = (void*)&jr2_submit_getwork_result;
-  gate->longpoll_rpc_call       = (void*)&jr2_longpoll_rpc_call;
-  gate->work_decode             = (void*)&jr2_work_decode;
-  gate->stratum_handle_response = (void*)&jr2_stratum_handle_response;
-  gate->nonce_index             = JR2_NONCE_INDEX;
-  jsonrpc_2 = true;   // still needed
-  opt_extranonce = false;
-//  have_gbt = false;
-  return true;
- }
-
-// run the alternate hash function for a specific algo
 void exec_hash_function( int algo, void *output, const void *pdata )
 {
  algo_gate_t gate;   
@@ -312,39 +419,35 @@ void exec_hash_function( int algo, void *output, const void *pdata )
 const char* const algo_alias_map[][2] =
 {
 //   alias                proper
-  { "argon2d-crds",      "argon2d250"   },
-  { "argon2d-dyn",       "argon2d500"   },
-  { "argon2d-uis",       "argon2d4096"  },
-  { "bitcore",           "timetravel10" },
-  { "bitzeny",           "yescryptr8"   },
-  { "blake256r8",        "blakecoin"    },
-  { "blake256r8vnl",     "vanilla"      },
-  { "blake256r14",       "blake"        },
-  { "blake256r14dcr",    "decred"       },
-  { "cryptonote",        "cryptonight"  },
-  { "cryptonight-light", "cryptolight"  },
-  { "diamond",           "dmd-gr"       },
-  { "droplp",            "drop"         },
-  { "espers",            "hmq1725"      },
-  { "flax",              "c11"          },
-  { "hsr",               "x13sm3"       },
-  { "jackpot",           "jha"          },
-  { "jane",              "scryptjane"   }, 
-  { "lyra2",             "lyra2re"      },
-  { "lyra2v2",           "lyra2rev2"    },
-  { "lyra2v3",           "lyra2rev3"    },
-  { "myrgr",             "myr-gr"       },
-  { "myriad",            "myr-gr"       },
-  { "neo",               "neoscrypt"    },
-  { "phi",               "phi1612"      },
-//  { "sia",               "blake2b"      },
-  { "sib",               "x11gost"      },
-  { "timetravel8",       "timetravel"   },
-  { "veil",              "x16rt-veil"   },
-  { "x16r-hex",          "hex"          },
-  { "yenten",            "yescryptr16"  },
-  { "ziftr",             "zr5"          },
-  { NULL,                NULL           }   
+  { "argon2d-dyn",       "argon2d500"     },
+  { "argon2d-uis",       "argon2d4096"    },
+  { "bcd",               "x13bcd"         },
+  { "bitcore",           "timetravel10"   },
+  { "bitzeny",           "yescryptr8"     },
+  { "blake256r8",        "blakecoin"      },
+  { "blake256r8vnl",     "vanilla"        },
+  { "blake256r14",       "blake"          },
+  { "blake256r14dcr",    "decred"         },
+  { "diamond",           "dmd-gr"         },
+  { "espers",            "hmq1725"        },
+  { "flax",              "c11"            },
+  { "hsr",               "x13sm3"         },
+  { "jackpot",           "jha"            },
+  { "lyra2",             "lyra2re"        },
+  { "lyra2v2",           "lyra2rev2"      },
+  { "lyra2v3",           "lyra2rev3"      },
+  { "myrgr",             "myr-gr"         },
+  { "myriad",            "myr-gr"         },
+  { "neo",               "neoscrypt"      },
+  { "phi",               "phi1612"        },
+  { "scryptn2",          "scrypt:1048576" },
+  { "sib",               "x11gost"        },
+  { "timetravel8",       "timetravel"     },
+  { "veil",              "x16rt-veil"     },
+  { "x16r-hex",          "hex"            },
+  { "yenten",            "yescryptr16"    },
+  { "ziftr",             "zr5"            },
+  { NULL,                NULL             }   
 };

 // if arg is a valid alias for a known algo it is updated with the proper
@@ -357,7 +460,7 @@ void get_algo_alias( char** algo_or_alias )
    if ( !strcasecmp( *algo_or_alias, algo_alias_map[i][ ALIAS ] ) )
    {
      // found valid alias, return proper name
-      *algo_or_alias = (char* const)( algo_alias_map[i][ PROPER ] );
+      *algo_or_alias = (char*)( algo_alias_map[i][ PROPER ] );
      return;
    }
 }
@@ -365,40 +468,3 @@ void get_algo_alias( char** algo_or_alias )
 #undef ALIAS
 #undef PROPER

-bool submit_solution( struct work *work, void *hash,
-                      struct thr_info *thr )
-{
-     work_set_target_ratio( work, hash );
-     if ( submit_work( thr, work ) )
-     {
-         if ( !opt_quiet )
-            applog( LOG_BLUE, "Share %d submitted by thread %d, job %s.",
-                    accepted_share_count + rejected_share_count + 1,
-                    thr->id, work->job_id );
-         return true;
-     }
-     else
-          applog( LOG_WARNING, "Failed to submit share." );
-     return false;
-}
-
-bool submit_lane_solution( struct work *work, void *hash,
-                           struct thr_info *thr, int lane )
-{
-     work_set_target_ratio( work, hash );
-     if ( submit_work( thr, work ) )
-     {
-         if ( !opt_quiet )
-//            applog( LOG_BLUE, "Share %d submitted by thread %d, lane %d.",
-//                    accepted_share_count + rejected_share_count + 1,
-//                    thr->id, lane );
-            applog( LOG_BLUE, "Share %d submitted by thread %d, lane %d, job %s.",
-                    accepted_share_count + rejected_share_count + 1, thr->id,
-                    lane, work->job_id );
-         return true;
-     }
-     else
-          applog( LOG_WARNING, "Failed to submit share." );
-     return false;
-}
-
--- a/algo-gate-api.h
+++ b/algo-gate-api.h
@@ -35,7 +35,7 @@
 //    6. Determine if other non existant functions are required.
 //    That is determined by the need to add code in cpu-miner.c
 //    that applies only to the new algo. That is forbidden. All
-//    algo specific code must be in theh algo's file.
+//    algo specific code must be in the algo's file.
 //
 //    7. If new functions need to be added to the gate add the type
 //    to the structure, declare a null instance in this file and define
@@ -48,10 +48,10 @@
 //    instances as they are defined by default, or unsafe functions that
 //    are not needed by the algo.
 //
-//    9. Add an case entry to the switch/case in function register_gate
+//    9. Add a case entry to the switch/case in function register_gate
 //    in file algo-gate-api.c for the new algo.
 //
-//    10 If a new function type was defined add an entry to ini talgo_gate
+//    10 If a new function type was defined add an entry to init algo_gate
 //    to initialize the new function to its null instance described in step 7.
 //
 //    11. If the new algo has aliases add them to the alias array in
@@ -75,7 +75,7 @@

 // my hack at creating a set data type using bit masks. Set inclusion,
 // exclusion union and intersection operations are provided for convenience. In // some cases it may be desireable to use boolean algebra directly on the
-// data to perfomr set operations. Sets can be represented as single
+// data to perform set operations. Sets can be represented as single
 // elements, a bitwise OR of multiple elements, a bitwise OR of multiple
 // set variables or constants, or combinations of the above.
 // Examples:
@@ -85,14 +85,17 @@

 typedef  uint32_t set_t;

-#define EMPTY_SET       0
-#define SSE2_OPT        1
-#define AES_OPT         2  
-#define SSE42_OPT       4
-#define AVX_OPT         8
-#define AVX2_OPT     0x10
-#define SHA_OPT      0x20
-#define AVX512_OPT   0x40
+#define EMPTY_SET        0
+#define SSE2_OPT         1
+#define AES_OPT          2  
+#define SSE42_OPT        4
+#define AVX_OPT          8   // Sandybridge
+#define AVX2_OPT      0x10   // Haswell, Zen1
+#define SHA_OPT       0x20   // Zen1, Icelake (sha256)
+#define AVX512_OPT    0x40   // Skylake-X (AVX512[F,VL,DQ,BW])
+#define VAES_OPT      0x80   // Icelake (VAES & AVX512)
+#define VAES256_OPT   0x100  // Zen3 (VAES without AVX512)
+

 // return set containing all elements from sets a & b
 inline set_t set_union ( set_t a, set_t b ) { return a | b; }
@@ -108,45 +111,62 @@ inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }

 typedef struct
 {
-// special case, only one target, provides a callback for scanhash to
-// submit work with less overhead.
-// bool (*submit_work )             ( struct thr_info*, const struct work* );
-
-// mandatory functions, must be overwritten
-// Added a 5th arg for the thread_info structure to replace the int thr id
-// in the first arg. Both will co-exist during the trasition.
-//int ( *scanhash ) ( int, struct work*, uint32_t, uint64_t* );
+// Mandatory functions, one of these is mandatory. If a generic scanhash
+// is used a custom target hash function must be registered, with a custom
+// scanhash the target hash function can be called directly and doesn't need
+// to be registered with the gate. 
 int ( *scanhash ) ( struct work*, uint32_t, uint64_t*, struct thr_info* );

-// optional unsafe, must be overwritten if algo uses function
-void ( *hash )     ( void*, const void*, uint32_t ) ;
-void ( *hash_suw ) ( void*, const void* );
+int ( *hash )     ( void*, const void*, int );

 //optional, safe to use default in most cases
-bool ( *miner_thread_init )      ( int );
-void ( *stratum_gen_work )       ( struct stratum_ctx*, struct work* );
-void ( *get_new_work )           ( struct work*, struct work*, int, uint32_t*,
-                                   bool );
-uint32_t *( *get_nonceptr )      ( uint32_t* );
-void ( *decode_extra_data )      ( struct work*, uint64_t* );
-void ( *wait_for_diff )          ( struct stratum_ctx* );
-int64_t ( *get_max64 )           ();
-bool ( *work_decode )            ( const json_t*, struct work* );
-void ( *set_target)              ( struct work*, double );
-bool ( *submit_getwork_result )  ( CURL*, struct work* );
-void ( *gen_merkle_root )        ( char*, struct stratum_ctx* );
-void ( *build_extraheader )      ( struct work*, struct stratum_ctx* );
-void ( *build_block_header )     ( struct work*, uint32_t, uint32_t*,
-	                           uint32_t*, uint32_t, uint32_t );
-void ( *build_stratum_request )  ( char*, struct work*, struct stratum_ctx* );
-char* ( *malloc_txs_request )    ( struct work* );
-void ( *set_work_data_endian )   ( struct work* );
-double ( *calc_network_diff )    ( struct work* );
-bool ( *ready_to_mine )          ( struct work*, struct stratum_ctx*, int );
-void ( *resync_threads )         ( struct work* );
-bool ( *do_this_thread )         ( int );
-json_t* (*longpoll_rpc_call)     ( CURL*, int*, char* );
-bool ( *stratum_handle_response )( json_t* );
+
+// Called once by each miner thread to allocate thread local buffers and
+// other initialization specific to miner threads.
+bool ( *miner_thread_init )     ( int );
+
+// Get thread local copy of blockheader with unique nonce.
+void ( *get_new_work )          ( struct work*, struct work*, int, uint32_t* );
+
+// Decode getwork blockheader
+bool ( *work_decode )           ( struct work* );
+
+// Extra getwork data
+void ( *decode_extra_data )     ( struct work*, uint64_t* );
+
+bool ( *submit_getwork_result ) ( CURL*, struct work* );
+
+void ( *gen_merkle_root )       ( char*, struct stratum_ctx* );
+
+// Increment extranonce
+void ( *build_extraheader )     ( struct work*, struct stratum_ctx* );
+
+void ( *build_block_header )    ( struct work*, uint32_t, uint32_t*,
+	                                uint32_t*, uint32_t, uint32_t,
+                                   unsigned char* );
+
+// Build mining.submit message
+void ( *build_stratum_request ) ( char*, struct work*, struct stratum_ctx* );
+
+char* ( *malloc_txs_request )   ( struct work* );
+
+// Big endian or little endian
+void ( *set_work_data_endian )  ( struct work* );
+
+double ( *calc_network_diff )   ( struct work* );
+
+// Wait for first work
+bool ( *ready_to_mine )         ( struct work*, struct stratum_ctx*, int );
+
+// Diverge mining threads
+bool ( *do_this_thread )        ( int );
+
+// After do_this_thread
+void ( *resync_threads )        ( int, struct work* );
+
+// No longer needed
+json_t* (*longpoll_rpc_call)      ( CURL*, int*, char* );
+
 set_t optimizations;
 int  ( *get_work_data_size )     ();
 int  ntime_index;
@@ -184,88 +204,96 @@ void four_way_not_tested();
 #define STD_WORK_DATA_SIZE 128
 #define STD_WORK_CMP_SIZE 76

-#define JR2_NONCE_INDEX 39  // 8 bit offset
+//#define JR2_NONCE_INDEX 39  // 8 bit offset

 // These indexes are only used with JSON RPC2 and are not gated.
-#define JR2_WORK_CMP_INDEX_2 43
-#define JR2_WORK_CMP_SIZE_2 33
+//#define JR2_WORK_CMP_INDEX_2 43
+//#define JR2_WORK_CMP_SIZE_2 33

-// allways returns failure
+// deprecated, use generic instead
 int null_scanhash();

-// Allow algos to submit from scanhash loop.
-bool submit_solution( struct work *work, void *hash,
-                      struct thr_info *thr );
-bool submit_lane_solution( struct work *work, void *hash,
-                          struct thr_info *thr, int lane );
+// Default generic, may be used in many cases.
+// N-way is more complicated, requires many different implementations
+// depending on architecture, input format, and output format.
+// Naming convention is scanhash_[N]way_[input format]in_[output format]out
+// N = number of lanes
+// input/output format:
+//    32: 32 bit interleaved parallel lanes
+//    64: 64 bit interleaved parallel lanes
+//    640: input only, not interleaved, contiguous serial 640 bit lanes.
+//    256: output only, not interleaved, contiguous serial 256 bit lanes.

- 
-bool submit_work( struct thr_info *thr, const struct work *work_in );
+int scanhash_generic( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
+
+#if defined(__AVX2__)
+
+//int scanhash_4way_64in_64out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+//int scanhash_4way_64in_256out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+int scanhash_4way_64in_32out( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
+
+//int scanhash_8way_32in_32out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+#endif
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+//int scanhash_8way_64in_64out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+//int scanhash_8way_64in_256out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+int scanhash_8way_64in_32out( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
+
+//int scanhash_16way_32in_32out( struct work *work, uint32_t max_nonce,
+//                      uint64_t *hashes_done, struct thr_info *mythr );
+
+#endif

 // displays warning
-void null_hash    ();
-void null_hash_suw();
+int null_hash();

 // optional safe targets, default listed first unless noted.

-void std_wait_for_diff();
-
-uint32_t *std_get_nonceptr( uint32_t *work_data );
-uint32_t *jr2_get_nonceptr( uint32_t *work_data );
-
 void std_get_new_work( struct work *work, struct work *g_work, int thr_id,
-                       uint32_t* end_nonce_ptr, bool clean_job );
-void jr2_get_new_work( struct work *work, struct work *g_work, int thr_id,
                       uint32_t* end_nonce_ptr );

-void std_stratum_gen_work( struct stratum_ctx *sctx, struct work *work );
-void jr2_stratum_gen_work( struct stratum_ctx *sctx, struct work *work );
-
 void sha256d_gen_merkle_root( char *merkle_root, struct stratum_ctx *sctx );
 void SHA256_gen_merkle_root ( char *merkle_root, struct stratum_ctx *sctx );

-// pick your favorite or define your own
-int64_t get_max64_0x1fffffLL(); // default
-int64_t get_max64_0x40LL();
-int64_t get_max64_0x3ffff();
-int64_t get_max64_0x3fffffLL();
-int64_t get_max64_0x1ffff();
-int64_t get_max64_0xffffLL();
-
-void std_set_target(    struct work *work, double job_diff );
-void alt_set_target(    struct work* work, double job_diff );
-void scrypt_set_target( struct work *work, double job_diff );
-
-bool std_le_work_decode( const json_t *val, struct work *work );
-bool std_be_work_decode( const json_t *val, struct work *work );
-bool jr2_work_decode( const json_t *val, struct work *work );
+bool std_le_work_decode( struct work *work );
+bool std_be_work_decode( struct work *work );

 bool std_le_submit_getwork_result( CURL *curl, struct work *work );
 bool std_be_submit_getwork_result( CURL *curl, struct work *work );
-bool jr2_submit_getwork_result( CURL *curl, struct work *work );

 void std_le_build_stratum_request( char *req, struct work *work );
 void std_be_build_stratum_request( char *req, struct work *work );
-void jr2_build_stratum_request   ( char *req, struct work *work );

 char* std_malloc_txs_request( struct work *work );

-// Default is do_nothing (assumed LE)
+// Default is do_nothing, little endian is assumed
 void set_work_data_big_endian( struct work *work );

 double std_calc_network_diff( struct work *work );

 void std_build_block_header( struct work* g_work, uint32_t version,
-	                     uint32_t *prevhash,  uint32_t *merkle_root,
-   	                     uint32_t ntime, uint32_t nbits );
+	                          uint32_t *prevhash,  uint32_t *merkle_root,
+   	                       uint32_t ntime,      uint32_t nbits,
+                             unsigned char *final_sapling_hash );

 void std_build_extraheader( struct work *work, struct stratum_ctx *sctx );

 json_t* std_longpoll_rpc_call( CURL *curl, int *err, char *lp_url );
-json_t* jr2_longpoll_rpc_call( CURL *curl, int *err );
-
-bool std_stratum_handle_response( json_t *val );
-bool jr2_stratum_handle_response( json_t *val );

 bool std_ready_to_mine( struct work* work, struct stratum_ctx* stratum,
                        int thr_id );
@@ -278,19 +306,16 @@ int std_get_work_data_size();
 // by calling the algo's register function.
 bool register_algo_gate( int algo, algo_gate_t *gate );

-// Override any default gate functions that are applicable and do any other
-// algo-specific initialization.
+// Called by algos to verride any default gate functions that are applicable
+// and do any other algo-specific initialization.
 // The register functions for all the algos can be declared here to reduce
 // compiler warnings but that's just more work for devs adding new algos.
 bool register_algo( algo_gate_t *gate );

-// Overrides a common set of functions used by RPC2 and other RPC2-specific
-// init. Called by algo's register function before initializing algo-specific
-// functions and data.
-bool register_json_rpc2( algo_gate_t *gate );
-
 // use this to call the hash function of an algo directly, ie util.c test.
 void exec_hash_function( int algo, void *output, const void *pdata );

-void get_algo_alias( char** algo_or_alias );
+// Validate a string as a known algo and alias, updates arg to proper
+// algo name if valid alias, NULL if invalid alias or algo.
+void get_algo_alias( char **algo_or_alias );

--- a/algo/argon2/argon2a/argon2a.c
+++ b/algo/argon2/argon2a/argon2a.c
@@ -62,9 +62,7 @@ int scanhash_argon2( struct work* work, uint32_t max_nonce,
 		argon2hash(hash, endiandata);
 		if (hash[7] <= Htarg && fulltest(hash, ptarget)) {
 			pdata[19] = nonce;
-			*hashes_done = pdata[19] - first_nonce;
-			work_set_target_ratio(work, hash);
-			return 1;
+         submit_solution( work, hash, mythr );
 		}
 		nonce++;
 	} while (nonce < max_nonce && !work_restart[thr_id].restart);
@@ -74,19 +72,14 @@ int scanhash_argon2( struct work* work, uint32_t max_nonce,
 	return 0;
 }

-int64_t argon2_get_max64 ()
-{
-  return 0x1ffLL;
-}
-
 bool register_argon2_algo( algo_gate_t* gate )
 {
  gate->optimizations = SSE2_OPT | AVX_OPT | AVX2_OPT;
  gate->scanhash        = (void*)&scanhash_argon2;
  gate->hash            = (void*)&argon2hash;
  gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  gate->set_target      = (void*)&scrypt_set_target;
-  gate->get_max64       = (void*)&argon2_get_max64;
+  opt_target_factor = 65536.0;
+
  return true;
 };

--- a/algo/argon2/argon2d/argon2d-gate.c
+++ b/algo/argon2/argon2d/argon2d-gate.c
@@ -1,4 +1,5 @@
 #include "argon2d-gate.h"
+#include "simd-utils.h"
 #include "argon2d/argon2.h"

 static const size_t INPUT_BYTES = 80;  // Lenth of a block header in bytes. Input Length = Salt Length (salt = input)
@@ -36,7 +37,7 @@ void argon2d_crds_hash( void *output, const void *input )
 int scanhash_argon2d_crds( struct work *work, uint32_t max_nonce,
                      uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t _ALIGN(64) endiandata[20];
+   uint32_t _ALIGN(64) edata[20];
   uint32_t _ALIGN(64) hash[8];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
@@ -45,11 +46,11 @@ int scanhash_argon2d_crds( struct work *work, uint32_t max_nonce,
   const uint32_t Htarg = ptarget[7];
   uint32_t nonce = first_nonce;

-   swab32_array( endiandata, pdata, 20 );
+   swab32_array( edata, pdata, 20 );

   do {
-      be32enc(&endiandata[19], nonce);
-      argon2d_crds_hash( hash, endiandata );
+      be32enc(&edata[19], nonce);
+      argon2d_crds_hash( hash, edata );
      if ( hash[7] <= Htarg && fulltest( hash, ptarget ) && !opt_benchmark )
      {
          pdata[19] = nonce;
@@ -67,8 +68,8 @@ bool register_argon2d_crds_algo( algo_gate_t* gate )
 {
        gate->scanhash = (void*)&scanhash_argon2d_crds;
        gate->hash = (void*)&argon2d_crds_hash;
-        gate->set_target = (void*)&scrypt_set_target;
        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
        return true;
 }

@@ -103,31 +104,32 @@ void argon2d_dyn_hash( void *output, const void *input )
 int scanhash_argon2d_dyn( struct work *work, uint32_t max_nonce,
                      uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t _ALIGN(64) endiandata[20];
+   uint32_t _ALIGN(64) edata[20];
   uint32_t _ALIGN(64) hash[8];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
-   const uint32_t first_nonce = pdata[19];
-   const uint32_t Htarg = ptarget[7];
+   const int thr_id = mythr->id; 
+   const uint32_t first_nonce = (const uint32_t)pdata[19];
+   const uint32_t last_nonce = (const uint32_t)max_nonce;
   uint32_t nonce = first_nonce;
+   const bool bench = opt_benchmark;

-   swab32_array( endiandata, pdata, 20 );
-
+   mm128_bswap32_80( edata, pdata );
   do
   {
-      be32enc(&endiandata[19], nonce);
-      argon2d_dyn_hash( hash, endiandata );
-      if ( hash[7] <= Htarg && fulltest( hash, ptarget ) && !opt_benchmark )
+      edata[19] = nonce;
+      argon2d_dyn_hash( hash, edata );
+      if ( unlikely( valid_hash( (uint64_t*)hash, (uint64_t*)ptarget )
+           && !bench ) )
      {
-          pdata[19] = nonce;
+          pdata[19] = bswap_32( nonce );;
          submit_solution( work, hash, mythr );
      }
      nonce++;
-  } while (nonce < max_nonce && !work_restart[thr_id].restart);
+  } while ( likely( nonce < last_nonce && !work_restart[thr_id].restart ) );

   pdata[19] = nonce;
-   *hashes_done = pdata[19] - first_nonce + 1;
+   *hashes_done = pdata[19] - first_nonce;
   return 0;
 }

@@ -135,8 +137,8 @@ bool register_argon2d_dyn_algo( algo_gate_t* gate )
 {
        gate->scanhash = (void*)&scanhash_argon2d_dyn;
        gate->hash = (void*)&argon2d_dyn_hash;
-        gate->set_target = (void*)&scrypt_set_target;
        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
        return true;
 }

@@ -146,47 +148,42 @@ int scanhash_argon2d4096( struct work *work, uint32_t max_nonce,
                           uint64_t *hashes_done, struct thr_info *mythr )
 {
   uint32_t _ALIGN(64) vhash[8];
-   uint32_t _ALIGN(64) endiandata[20];
+   uint32_t _ALIGN(64) edata[20];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = (const uint32_t)max_nonce;
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;  // thr_id arg is deprecated
   uint32_t t_cost = 1; // 1 iteration
   uint32_t m_cost = 4096; // use 4MB
   uint32_t parallelism = 1; // 1 thread, 2 lanes
+   const bool bench = opt_benchmark;

-   for ( int i = 0; i < 19; i++ )
-      be32enc( &endiandata[i], pdata[i] );
+   mm128_bswap32_80( edata, pdata );

   do {
-      be32enc( &endiandata[19], n );
-      argon2d_hash_raw( t_cost, m_cost, parallelism, (char*) endiandata, 80,
-                 (char*) endiandata, 80, (char*) vhash, 32, ARGON2_VERSION_13 );
-      if ( vhash[7] < Htarg && fulltest( vhash, ptarget ) && !opt_benchmark )
+      edata[19] = n;
+      argon2d_hash_raw( t_cost, m_cost, parallelism, (char*) edata, 80,
+                 (char*) edata, 80, (char*) vhash, 32, ARGON2_VERSION_13 );
+      if ( unlikely( valid_hash( vhash, ptarget ) && !bench ) )
      {
-         pdata[19] = n;
+         be32enc( &pdata[19], n );
         submit_solution( work, vhash, mythr );
      }
      n++;
+   } while ( likely( n < last_nonce && !work_restart[thr_id].restart ) );

-   } while (n < max_nonce && !work_restart[thr_id].restart);
-
-   *hashes_done = n - first_nonce + 1;
+   *hashes_done = n - first_nonce;
   pdata[19] = n;
-
   return 0;
 }

-int64_t get_max64_0x1ff() { return 0x1ff; }
-
 bool register_argon2d4096_algo( algo_gate_t* gate )
 {
        gate->scanhash = (void*)&scanhash_argon2d4096;
-        gate->set_target = (void*)&scrypt_set_target;
-        gate->get_max64  = (void*)&get_max64_0x1ff;
        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
        return true;
 }

--- a/algo/argon2/argon2d/argon2d/opt.c
+++ b/algo/argon2/argon2d/argon2d/opt.c
@@ -21,7 +21,7 @@

 #include "argon2.h"
 #include "core.h"
-
+#include "simd-utils.h"
 #include "../blake2/blake2.h"
 #include "../blake2/blamka-round-opt.h"

@@ -37,24 +37,28 @@

 #if defined(__AVX512F__)

-static void fill_block(__m512i *state, const block *ref_block,
-                       block *next_block, int with_xor) {
+static void fill_block( __m512i *state, const block *ref_block,
+                       block *next_block, int with_xor )
+{
    __m512i block_XY[ARGON2_512BIT_WORDS_IN_BLOCK];
    unsigned int i;

-    if (with_xor) {
-        for (i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++) {
-            state[i] = _mm512_xor_si512(
-                state[i], _mm512_loadu_si512((const __m512i *)ref_block->v + i));
-            block_XY[i] = _mm512_xor_si512(
-                state[i], _mm512_loadu_si512((const __m512i *)next_block->v + i));
-        }
-    } else {
-        for (i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++) {
-            block_XY[i] = state[i] = _mm512_xor_si512(
-                state[i], _mm512_loadu_si512((const __m512i *)ref_block->v + i));
+    if ( with_xor )
+    {
+        for ( i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++ )
+        {
+            state[i] = _mm512_xor_si512( state[i],
+                      _mm512_load_si512( (const __m512i*)ref_block->v + i ) );
+            block_XY[i] = _mm512_xor_si512( state[i],
+                      _mm512_load_si512( (const __m512i*)next_block->v + i ) );
        }
    }
+    else
+    {
+        for ( i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++ )
+            block_XY[i] = state[i] = _mm512_xor_si512( state[i],
+                      _mm512_load_si512( (const __m512i*)ref_block->v + i ) );
+    }

    BLAKE2_ROUND_1( state[ 0], state[ 1], state[ 2], state[ 3],
                    state[ 4], state[ 5], state[ 6], state[ 7] );
@@ -66,23 +70,10 @@ static void fill_block(__m512i *state, const block *ref_block,
    BLAKE2_ROUND_2( state[ 1], state[ 3], state[ 5], state[ 7],
                    state[ 9], state[11], state[13], state[15] );

-/*
-    for (i = 0; i < 2; ++i) {
-        BLAKE2_ROUND_1(
-            state[8 * i + 0], state[8 * i + 1], state[8 * i + 2], state[8 * i + 3],
-            state[8 * i + 4], state[8 * i + 5], state[8 * i + 6], state[8 * i + 7]);
-    }
-
-    for (i = 0; i < 2; ++i) {
-        BLAKE2_ROUND_2(
-            state[2 * 0 + i], state[2 * 1 + i], state[2 * 2 + i], state[2 * 3 + i],
-            state[2 * 4 + i], state[2 * 5 + i], state[2 * 6 + i], state[2 * 7 + i]);
-    }
-*/
-
-    for (i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++) {
-        state[i] = _mm512_xor_si512(state[i], block_XY[i]);
-        _mm512_storeu_si512((__m512i *)next_block->v + i, state[i]);
+    for ( i = 0; i < ARGON2_512BIT_WORDS_IN_BLOCK; i++ )
+    {
+        state[i] = _mm512_xor_si512( state[i], block_XY[i] );
+        _mm512_store_si512( (__m512i*)next_block->v + i, state[i] );
    }
 }

@@ -125,18 +116,6 @@ static void fill_block(__m256i *state, const block *ref_block,
    BLAKE2_ROUND_2( state[ 3], state[ 7], state[11], state[15],
                    state[19], state[23], state[27], state[31] );

-/*
-    for (i = 0; i < 4; ++i) {
-        BLAKE2_ROUND_1(state[8 * i + 0], state[8 * i + 4], state[8 * i + 1], state[8 * i + 5],
-                       state[8 * i + 2], state[8 * i + 6], state[8 * i + 3], state[8 * i + 7]);
-    }
-
-    for (i = 0; i < 4; ++i) {
-        BLAKE2_ROUND_2(state[ 0 + i], state[ 4 + i], state[ 8 + i], state[12 + i],
-                       state[16 + i], state[20 + i], state[24 + i], state[28 + i]);
-    }
-*/
-
    for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
        state[i] = _mm256_xor_si256(state[i], block_XY[i]);
        _mm256_store_si256((__m256i *)next_block->v + i, state[i]);
@@ -153,14 +132,14 @@ static void fill_block(__m128i *state, const block *ref_block,
    if (with_xor) {
        for (i = 0; i < ARGON2_OWORDS_IN_BLOCK; i++) {
            state[i] = _mm_xor_si128(
-                state[i], _mm_loadu_si128((const __m128i *)ref_block->v + i));
+                state[i], _mm_load_si128((const __m128i *)ref_block->v + i));
            block_XY[i] = _mm_xor_si128(
-                state[i], _mm_loadu_si128((const __m128i *)next_block->v + i));
+                state[i], _mm_load_si128((const __m128i *)next_block->v + i));
        }
    } else {
        for (i = 0; i < ARGON2_OWORDS_IN_BLOCK; i++) {
            block_XY[i] = state[i] = _mm_xor_si128(
-                state[i], _mm_loadu_si128((const __m128i *)ref_block->v + i));
+                state[i], _mm_load_si128((const __m128i *)ref_block->v + i));
        }
    }

@@ -198,22 +177,9 @@ static void fill_block(__m128i *state, const block *ref_block,
    BLAKE2_ROUND( state[ 7], state[15], state[23], state[31],  
                  state[39], state[47], state[55], state[63] );

-/*
-    for (i = 0; i < 8; ++i) {
-        BLAKE2_ROUND(state[8 * i + 0], state[8 * i + 1], state[8 * i + 2],
-            state[8 * i + 3], state[8 * i + 4], state[8 * i + 5],
-            state[8 * i + 6], state[8 * i + 7]);
-    }
-
-    for (i = 0; i < 8; ++i) {
-        BLAKE2_ROUND(state[8 * 0 + i], state[8 * 1 + i], state[8 * 2 + i],
-            state[8 * 3 + i], state[8 * 4 + i], state[8 * 5 + i],
-            state[8 * 6 + i], state[8 * 7 + i]);
-    }
-*/
    for (i = 0; i < ARGON2_OWORDS_IN_BLOCK; i++) {
        state[i] = _mm_xor_si128(state[i], block_XY[i]);
-        _mm_storeu_si128((__m128i *)next_block->v + i, state[i]);
+        _mm_store_si128((__m128i *)next_block->v + i, state[i]);
    }
 }

--- a/algo/argon2/argon2d/blake2/blamka-round-opt.h
+++ b/algo/argon2/argon2d/blake2/blamka-round-opt.h
@@ -184,10 +184,10 @@ static BLAKE2_INLINE __m128i fBlaMka(__m128i x, __m128i y) {

 #include <immintrin.h>

-#define  rotr32  mm256_swap32_64
-#define  rotr24  mm256_ror3x8_64
-#define  rotr16  mm256_ror1x16_64
-#define  rotr63( x ) mm256_rol_64( x, 1 )
+#define  rotr32( x )  mm256_ror_64( x, 32 )
+#define  rotr24( x )  mm256_ror_64( x, 24 )
+#define  rotr16( x )  mm256_ror_64( x, 16 )
+#define  rotr63( x )  mm256_rol_64( x,  1 )

 //#define rotr32(x)   _mm256_shuffle_epi32(x, _MM_SHUFFLE(2, 3, 0, 1))
 //#define rotr24(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10, 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10))
@@ -427,14 +427,14 @@ static __m512i muladd(__m512i x, __m512i y)
 #define SWAP_QUARTERS(A0, A1) \
    do { \
        SWAP_HALVES(A0, A1); \
-        A0 = _mm512_permutexvar_epi64(_mm512_setr_epi64(0, 1, 4, 5, 2, 3, 6, 7), A0); \
-        A1 = _mm512_permutexvar_epi64(_mm512_setr_epi64(0, 1, 4, 5, 2, 3, 6, 7), A1); \
+        A0 = _mm512_shuffle_i64x2( A0, A0, 0xd8 ); \
+        A1 = _mm512_shuffle_i64x2( A1, A1, 0xd8 ); \
    } while((void)0, 0)

 #define UNSWAP_QUARTERS(A0, A1) \
    do { \
-        A0 = _mm512_permutexvar_epi64(_mm512_setr_epi64(0, 1, 4, 5, 2, 3, 6, 7), A0); \
-        A1 = _mm512_permutexvar_epi64(_mm512_setr_epi64(0, 1, 4, 5, 2, 3, 6, 7), A1); \
+        A0 = _mm512_shuffle_i64x2( A0, A0, 0xd8 ); \
+        A1 = _mm512_shuffle_i64x2( A1, A1, 0xd8 ); \
        SWAP_HALVES(A0, A1); \
    } while((void)0, 0)

--- a/algo/blake/blake-4way.c
+++ b/algo/blake/blake-4way.c
@@ -13,7 +13,7 @@ void blakehash_4way(void *state, const void *input)
     uint32_t vhash[8*4] __attribute__ ((aligned (64)));
     blake256r14_4way_context ctx;
     memcpy( &ctx, &blake_4w_ctx, sizeof ctx );
-     blake256r14_4way( &ctx, input + (64<<2), 16 );
+     blake256r14_4way_update( &ctx, input + (64<<2), 16 );
     blake256r14_4way_close( &ctx, vhash );
     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
 }
@@ -36,7 +36,7 @@ int scanhash_blake_4way( struct work *work, uint32_t max_nonce,

   mm128_bswap32_intrlv80_4x32( vdata, pdata );
   blake256r14_4way_init( &blake_4w_ctx );
-   blake256r14_4way( &blake_4w_ctx, vdata, 64 );
+   blake256r14_4way_update( &blake_4w_ctx, vdata, 64 );

   do {
      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
@@ -48,7 +48,7 @@ int scanhash_blake_4way( struct work *work, uint32_t max_nonce,
      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;

@@ -107,7 +107,7 @@ int scanhash_blake_8way( struct work *work, uint32_t max_nonce,
      if ( (hash+i)[7] <= HTarget && fulltest( hash+i, ptarget ) )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 8;

--- a/algo/blake/blake-gate.c
+++ b/algo/blake/blake-gate.c
@@ -1,18 +1,8 @@
 #include "blake-gate.h"

-int64_t blake_get_max64 ()
-{
-  return 0x7ffffLL;
-}
-
 bool register_blake_algo( algo_gate_t* gate )
 {
  gate->optimizations = AVX2_OPT;
-  gate->get_max64 = (void*)&blake_get_max64;
-//#if defined (__AVX2__) && defined (FOUR_WAY)
-//   gate->optimizations = SSE2_OPT | AVX2_OPT;
-//  gate->scanhash  = (void*)&scanhash_blake_8way;
-//  gate->hash      = (void*)&blakehash_8way;
 #if defined(BLAKE_4WAY)
  four_way_not_tested();
  gate->scanhash  = (void*)&scanhash_blake_4way;
--- a/algo/blake/blake-hash-4way.h
+++ b/algo/blake/blake-hash-4way.h
@@ -37,8 +37,6 @@
 #ifndef __BLAKE_HASH_4WAY__
 #define __BLAKE_HASH_4WAY__ 1

-//#ifdef __SSE4_2__
-
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -51,49 +49,45 @@ extern "C"{

 #define SPH_SIZE_blake512   512

-// With SSE4.2 only Blake-256 4 way is available.
-// With AVX2 Blake-256 8way & Blake-512 4 way are also available.
-
-// Blake-256 4 way
+//////////////////////////
+//
+//   Blake-256 4 way SSE2

 typedef struct {
   unsigned char buf[64<<2];
   uint32_t H[8<<2];
-   uint32_t S[4<<2];
-//   __m128i buf[16] __attribute__ ((aligned (64)));
-//   __m128i H[8];
-//   __m128i S[4];    
   size_t ptr;
   uint32_t T0, T1;
   int rounds;   // 14 for blake, 8 for blakecoin & vanilla
 } blake_4way_small_context __attribute__ ((aligned (64)));

-// Default 14 rounds
+// Default, 14 rounds, blake, decred
 typedef blake_4way_small_context blake256_4way_context;
 void blake256_4way_init(void *ctx);
-void blake256_4way(void *ctx, const void *data, size_t len);
+void blake256_4way_update(void *ctx, const void *data, size_t len);
 void blake256_4way_close(void *ctx, void *dst);

 // 14 rounds, blake, decred
 typedef blake_4way_small_context blake256r14_4way_context;
 void blake256r14_4way_init(void *cc);
-void blake256r14_4way(void *cc, const void *data, size_t len);
+void blake256r14_4way_update(void *cc, const void *data, size_t len);
 void blake256r14_4way_close(void *cc, void *dst);

 // 8 rounds, blakecoin, vanilla
 typedef blake_4way_small_context blake256r8_4way_context;
 void blake256r8_4way_init(void *cc);
-void blake256r8_4way(void *cc, const void *data, size_t len);
+void blake256r8_4way_update(void *cc, const void *data, size_t len);
 void blake256r8_4way_close(void *cc, void *dst);

 #ifdef __AVX2__

-// Blake-256 8 way
+//////////////////////////
+//
+//   Blake-256 8 way AVX2

 typedef struct {
   __m256i buf[16] __attribute__ ((aligned (64)));
   __m256i H[8];
-   __m256i S[4];
   size_t ptr;
   sph_u32 T0, T1;
   int rounds;   // 14 for blake, 8 for blakecoin & vanilla
@@ -102,39 +96,92 @@ typedef struct {
 // Default 14 rounds
 typedef blake_8way_small_context blake256_8way_context;
 void blake256_8way_init(void *cc);
-void blake256_8way(void *cc, const void *data, size_t len);
+void blake256_8way_update(void *cc, const void *data, size_t len);
 void blake256_8way_close(void *cc, void *dst);

 // 14 rounds, blake, decred
 typedef blake_8way_small_context blake256r14_8way_context;
 void blake256r14_8way_init(void *cc);
-void blake256r14_8way(void *cc, const void *data, size_t len);
+void blake256r14_8way_update(void *cc, const void *data, size_t len);
 void blake256r14_8way_close(void *cc, void *dst);

 // 8 rounds, blakecoin, vanilla
 typedef blake_8way_small_context blake256r8_8way_context;
 void blake256r8_8way_init(void *cc);
-void blake256r8_8way(void *cc, const void *data, size_t len);
+void blake256r8_8way_update(void *cc, const void *data, size_t len);
 void blake256r8_8way_close(void *cc, void *dst);

-// Blake-512 4 way
+// Blake-512 4 way AVX2

 typedef struct {
-   __m256i buf[16] __attribute__ ((aligned (64)));
+   __m256i buf[16];
   __m256i H[8];
   __m256i S[4];   
   size_t ptr;
   sph_u64 T0, T1;
-} blake_4way_big_context;
+} blake_4way_big_context __attribute__ ((aligned (128)));

 typedef blake_4way_big_context blake512_4way_context;

-void blake512_4way_init(void *cc);
-void blake512_4way(void *cc, const void *data, size_t len);
-void blake512_4way_close(void *cc, void *dst);
-void blake512_4way_addbits_and_close(
-	void *cc, unsigned ub, unsigned n, void *dst);
+void blake512_4way_init( blake_4way_big_context *sc );
+void blake512_4way_update( void *cc, const void *data, size_t len );
+void blake512_4way_close( void *cc, void *dst );
+void blake512_4way_full( blake_4way_big_context *sc, void * dst,
+                         const void *data, size_t len );

+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+////////////////////////////
+//
+//   Blake-256 16 way AVX512
+
+typedef struct {
+   __m512i buf[16];
+   __m512i H[8];
+   size_t ptr;
+   uint32_t T0, T1;
+   int rounds;   // 14 for blake, 8 for blakecoin & vanilla
+} blake_16way_small_context __attribute__ ((aligned (128)));
+
+// Default 14 rounds
+typedef blake_16way_small_context blake256_16way_context;
+void blake256_16way_init(void *cc);
+void blake256_16way_update(void *cc, const void *data, size_t len);
+void blake256_16way_close(void *cc, void *dst);
+
+// 14 rounds, blake, decred
+typedef blake_16way_small_context blake256r14_16way_context;
+void blake256r14_16way_init(void *cc);
+void blake256r14_16way_update(void *cc, const void *data, size_t len);
+void blake256r14_16way_close(void *cc, void *dst);
+
+// 8 rounds, blakecoin, vanilla
+typedef blake_16way_small_context blake256r8_16way_context;
+void blake256r8_16way_init(void *cc);
+void blake256r8_16way_update(void *cc, const void *data, size_t len);
+void blake256r8_16way_close(void *cc, void *dst);
+
+////////////////////////////
+//
+//// Blake-512 8 way AVX512
+
+typedef struct {
+   __m512i buf[16];
+   __m512i H[8];
+   __m512i S[4];
+   size_t ptr;
+   sph_u64 T0, T1;
+} blake_8way_big_context __attribute__ ((aligned (128)));
+
+typedef blake_8way_big_context blake512_8way_context;
+
+void blake512_8way_init( blake_8way_big_context *sc );
+void blake512_8way_update( void *cc, const void *data, size_t len );
+void blake512_8way_close( void *cc, void *dst );
+void blake512_8way_full( blake_8way_big_context *sc, void * dst,
+                        const void *data, size_t len );
+
+#endif  // AVX512
 #endif  // AVX2

 #ifdef __cplusplus
--- a/algo/blake/blake256-hash-4way.c
+++ b/algo/blake/blake256-hash-4way.c
@@ -304,16 +304,17 @@ static const sph_u32 CS[16] = {

 #endif

+// Blake-256 4 way

 #define GS_4WAY( m0, m1, c0, c1, a, b, c, d ) \
 do { \
-   a = _mm_add_epi32( _mm_add_epi32( _mm_xor_si128( \
-                                   _mm_set1_epi32( c1 ), m0 ), b ), a ); \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), \
+                      _mm_xor_si128( _mm_set1_epi32( c1 ), m0 ) ); \
   d = mm128_ror_32( _mm_xor_si128( d, a ), 16 ); \
   c = _mm_add_epi32( c, d ); \
   b = mm128_ror_32( _mm_xor_si128( b, c ), 12 ); \
-   a = _mm_add_epi32( _mm_add_epi32( _mm_xor_si128( \
-                                   _mm_set1_epi32( c0 ), m1 ), b ), a ); \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), \
+                      _mm_xor_si128( _mm_set1_epi32( c0 ), m1 ) ); \
   d = mm128_ror_32( _mm_xor_si128( d, a ), 8 ); \
   c = _mm_add_epi32( c, d ); \
   b = mm128_ror_32( _mm_xor_si128( b, c ), 7 ); \
@@ -321,7 +322,8 @@ do { \

 #if SPH_COMPACT_BLAKE_32

-// Blake-256 4 way
+// Not used
+#if 0

 #define ROUND_S_4WAY(r)   do { \
 	GS_4WAY(M[sigma[r][0x0]], M[sigma[r][0x1]], \
@@ -342,6 +344,8 @@ do { \
 		CS[sigma[r][0xE]], CS[sigma[r][0xF]], V3, V4, V9, VE); \
 } while (0)

+#endif
+
 #else

 #define ROUND_S_4WAY(r)   do { \
@@ -359,7 +363,6 @@ do { \

 #define DECL_STATE32_4WAY \
 	__m128i H0, H1, H2, H3, H4, H5, H6, H7; \
-	__m128i S0, S1, S2, S3; \
        uint32_t T0, T1;

 #define READ_STATE32_4WAY(state)   do { \
@@ -371,10 +374,6 @@ do { \
 		H5 = casti_m128i( state->H, 5 ); \
 		H6 = casti_m128i( state->H, 6 ); \
 		H7 = casti_m128i( state->H, 7 ); \
-		S0 = casti_m128i( state->S, 0 ); \
-		S1 = casti_m128i( state->S, 1 ); \
-		S2 = casti_m128i( state->S, 2 ); \
-		S3 = casti_m128i( state->S, 3 ); \
 		T0 = (state)->T0; \
 		T1 = (state)->T1; \
 	} while (0)
@@ -388,17 +387,13 @@ do { \
 		casti_m128i( state->H, 5 ) = H5; \
 		casti_m128i( state->H, 6 ) = H6; \
 		casti_m128i( state->H, 7 ) = H7; \
-		casti_m128i( state->S, 0 ) = S0; \
-		casti_m128i( state->S, 1 ) = S1; \
-		casti_m128i( state->S, 2 ) = S2; \
-		casti_m128i( state->S, 3 ) = S3; \
 		(state)->T0 = T0; \
 		(state)->T1 = T1; \
 	} while (0)

 #if SPH_COMPACT_BLAKE_32
 // not used
-
+#if 0
 #define COMPRESS32_4WAY( rounds )   do { \
 	__m128i M[16]; \
 	__m128i V0, V1, V2, V3, V4, V5, V6, V7; \
@@ -441,6 +436,7 @@ do { \
        H7 = _mm_xor_si128( _mm_xor_si128( \
                                   _mm_xor_si128( S3, V7 ), VF ), H7 ); \
 	} while (0)
+#endif

 #else

@@ -508,10 +504,10 @@ do { \
   V5 = H5; \
   V6 = H6; \
   V7 = H7; \
-   V8 = _mm_xor_si128( S0, m128_const1_64( 0x243F6A88243F6A88 ) ); \
-   V9 = _mm_xor_si128( S1, m128_const1_64( 0x85A308D385A308D3 ) ); \
-   VA = _mm_xor_si128( S2, m128_const1_64( 0x13198A2E13198A2E ) ); \
-   VB = _mm_xor_si128( S3, m128_const1_64( 0x0370734403707344 ) ); \
+   V8 = m128_const1_64( 0x243F6A88243F6A88 ); \
+   V9 = m128_const1_64( 0x85A308D385A308D3 ); \
+   VA = m128_const1_64( 0x13198A2E13198A2E ); \
+   VB = m128_const1_64( 0x0370734403707344 ); \
   VC = _mm_xor_si128( _mm_set1_epi32( T0 ), \
                           m128_const1_64( 0xA4093822A4093822 ) ); \
   VD = _mm_xor_si128( _mm_set1_epi32( T0 ), \
@@ -538,14 +534,14 @@ do { \
      ROUND_S_4WAY(2); \
      ROUND_S_4WAY(3); \
   } \
-   H0 = mm128_xor4( V8, V0, S0, H0 ); \
-   H1 = mm128_xor4( V9, V1, S1, H1 ); \
-   H2 = mm128_xor4( VA, V2, S2, H2 ); \
-   H3 = mm128_xor4( VB, V3, S3, H3 ); \
-   H4 = mm128_xor4( VC, V4, S0, H4 ); \
-   H5 = mm128_xor4( VD, V5, S1, H5 ); \
-   H6 = mm128_xor4( VE, V6, S2, H6 ); \
-   H7 = mm128_xor4( VF, V7, S3, H7 ); \
+   H0 = _mm_xor_si128( _mm_xor_si128( V8, V0 ), H0 ); \
+   H1 = _mm_xor_si128( _mm_xor_si128( V9, V1 ), H1 ); \
+   H2 = _mm_xor_si128( _mm_xor_si128( VA, V2 ), H2 ); \
+   H3 = _mm_xor_si128( _mm_xor_si128( VB, V3 ), H3 ); \
+   H4 = _mm_xor_si128( _mm_xor_si128( VC, V4 ), H4 ); \
+   H5 = _mm_xor_si128( _mm_xor_si128( VD, V5 ), H5 ); \
+   H6 = _mm_xor_si128( _mm_xor_si128( VE, V6 ), H6 ); \
+   H7 = _mm_xor_si128( _mm_xor_si128( VF, V7 ), H7 ); \
 } while (0)

 #endif
@@ -556,13 +552,13 @@ do { \

 #define GS_8WAY( m0, m1, c0, c1, a, b, c, d ) \
 do { \
-   a = _mm256_add_epi32( _mm256_add_epi32( _mm256_xor_si256( \
-                 _mm256_set1_epi32( c1 ), m0 ), b ), a ); \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), \
+                         _mm256_xor_si256( _mm256_set1_epi32( c1 ), m0 ) ); \
   d = mm256_ror_32( _mm256_xor_si256( d, a ), 16 ); \
   c = _mm256_add_epi32( c, d ); \
   b = mm256_ror_32( _mm256_xor_si256( b, c ), 12 ); \
-   a = _mm256_add_epi32( _mm256_add_epi32( _mm256_xor_si256( \
-                 _mm256_set1_epi32( c0 ), m1 ), b ), a ); \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), \
+                         _mm256_xor_si256( _mm256_set1_epi32( c0 ), m1 ) ); \
   d = mm256_ror_32( _mm256_xor_si256( d, a ), 8 ); \
   c = _mm256_add_epi32( c, d ); \
   b = mm256_ror_32( _mm256_xor_si256( b, c ), 7 ); \
@@ -581,7 +577,6 @@ do { \

 #define DECL_STATE32_8WAY \
   __m256i H0, H1, H2, H3, H4, H5, H6, H7; \
-   __m256i S0, S1, S2, S3; \
   sph_u32 T0, T1;

 #define READ_STATE32_8WAY(state) \
@@ -594,10 +589,6 @@ do { \
   H5 = (state)->H[5]; \
   H6 = (state)->H[6]; \
   H7 = (state)->H[7]; \
-   S0 = (state)->S[0]; \
-   S1 = (state)->S[1]; \
-   S2 = (state)->S[2]; \
-   S3 = (state)->S[3]; \
   T0 = (state)->T0; \
   T1 = (state)->T1; \
 } while (0)
@@ -612,10 +603,6 @@ do { \
   (state)->H[5] = H5; \
   (state)->H[6] = H6; \
   (state)->H[7] = H7; \
-   (state)->S[0] = S0; \
-   (state)->S[1] = S1; \
-   (state)->S[2] = S2; \
-   (state)->S[3] = S3; \
   (state)->T0 = T0; \
   (state)->T1 = T1; \
 } while (0)
@@ -635,10 +622,10 @@ do { \
   V5 = H5; \
   V6 = H6; \
   V7 = H7; \
-   V8 = _mm256_xor_si256( S0, m256_const1_64( 0x243F6A88243F6A88 ) ); \
-   V9 = _mm256_xor_si256( S1, m256_const1_64( 0x85A308D385A308D3 ) ); \
-   VA = _mm256_xor_si256( S2, m256_const1_64( 0x13198A2E13198A2E ) ); \
-   VB = _mm256_xor_si256( S3, m256_const1_64( 0x0370734403707344 ) ); \
+   V8 = m256_const1_64( 0x243F6A88243F6A88 ); \
+   V9 = m256_const1_64( 0x85A308D385A308D3 ); \
+   VA = m256_const1_64( 0x13198A2E13198A2E ); \
+   VB = m256_const1_64( 0x0370734403707344 ); \
   VC = _mm256_xor_si256( _mm256_set1_epi32( T0 ),\
                              m256_const1_64( 0xA4093822A4093822 ) ); \
   VD = _mm256_xor_si256( _mm256_set1_epi32( T0 ),\
@@ -647,7 +634,7 @@ do { \
                              m256_const1_64( 0x082EFA98082EFA98 ) ); \
   VF = _mm256_xor_si256( _mm256_set1_epi32( T1 ), \
                              m256_const1_64( 0xEC4E6C89EC4E6C89 ) ); \
-   shuf_bswap32 = m256_const_64( 0x0c0d0e0f08090a0b, 0x0405060700010203, \
+   shuf_bswap32 = m256_const_64( 0x1c1d1e1f18191a1b, 0x1415161710111213, \
                                 0x0c0d0e0f08090a0b, 0x0405060700010203 ); \
   M0 = _mm256_shuffle_epi8( * buf    , shuf_bswap32 ); \
   M1 = _mm256_shuffle_epi8( *(buf+ 1), shuf_bswap32 ); \
@@ -682,17 +669,155 @@ do { \
      ROUND_S_8WAY(2); \
      ROUND_S_8WAY(3); \
   } \
-   H0 = mm256_xor4( V8, V0, S0, H0 ); \
-   H1 = mm256_xor4( V9, V1, S1, H1 ); \
-   H2 = mm256_xor4( VA, V2, S2, H2 ); \
-   H3 = mm256_xor4( VB, V3, S3, H3 ); \
-   H4 = mm256_xor4( VC, V4, S0, H4 ); \
-   H5 = mm256_xor4( VD, V5, S1, H5 ); \
-   H6 = mm256_xor4( VE, V6, S2, H6 ); \
-   H7 = mm256_xor4( VF, V7, S3, H7 ); \
+   H0 = _mm256_xor_si256( _mm256_xor_si256( V8, V0 ), H0 ); \
+   H1 = _mm256_xor_si256( _mm256_xor_si256( V9, V1 ), H1 ); \
+   H2 = _mm256_xor_si256( _mm256_xor_si256( VA, V2 ), H2 ); \
+   H3 = _mm256_xor_si256( _mm256_xor_si256( VB, V3 ), H3 ); \
+   H4 = _mm256_xor_si256( _mm256_xor_si256( VC, V4 ), H4 ); \
+   H5 = _mm256_xor_si256( _mm256_xor_si256( VD, V5 ), H5 ); \
+   H6 = _mm256_xor_si256( _mm256_xor_si256( VE, V6 ), H6 ); \
+   H7 = _mm256_xor_si256( _mm256_xor_si256( VF, V7 ), H7 ); \
 } while (0)


+#endif
+
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+// Blaske-256 16 way AVX512
+
+#define GS_16WAY( m0, m1, c0, c1, a, b, c, d ) \
+do { \
+   a = _mm512_add_epi32( _mm512_add_epi32( a, b ), \
+                         _mm512_xor_si512( _mm512_set1_epi32( c1 ), m0 ) ); \
+   d = mm512_ror_32( _mm512_xor_si512( d, a ), 16 ); \
+   c = _mm512_add_epi32( c, d ); \
+   b = mm512_ror_32( _mm512_xor_si512( b, c ), 12 ); \
+   a = _mm512_add_epi32( _mm512_add_epi32( a, b ), \
+                         _mm512_xor_si512( _mm512_set1_epi32( c0 ), m1 ) ); \
+   d = mm512_ror_32( _mm512_xor_si512( d, a ), 8 ); \
+   c = _mm512_add_epi32( c, d ); \
+   b = mm512_ror_32( _mm512_xor_si512( b, c ), 7 ); \
+} while (0)
+
+#define ROUND_S_16WAY(r)   do { \
+        GS_16WAY(Mx(r, 0), Mx(r, 1), CSx(r, 0), CSx(r, 1), V0, V4, V8, VC); \
+        GS_16WAY(Mx(r, 2), Mx(r, 3), CSx(r, 2), CSx(r, 3), V1, V5, V9, VD); \
+        GS_16WAY(Mx(r, 4), Mx(r, 5), CSx(r, 4), CSx(r, 5), V2, V6, VA, VE); \
+        GS_16WAY(Mx(r, 6), Mx(r, 7), CSx(r, 6), CSx(r, 7), V3, V7, VB, VF); \
+        GS_16WAY(Mx(r, 8), Mx(r, 9), CSx(r, 8), CSx(r, 9), V0, V5, VA, VF); \
+        GS_16WAY(Mx(r, A), Mx(r, B), CSx(r, A), CSx(r, B), V1, V6, VB, VC); \
+        GS_16WAY(Mx(r, C), Mx(r, D), CSx(r, C), CSx(r, D), V2, V7, V8, VD); \
+        GS_16WAY(Mx(r, E), Mx(r, F), CSx(r, E), CSx(r, F), V3, V4, V9, VE); \
+} while (0)
+
+#define DECL_STATE32_16WAY \
+   __m512i H0, H1, H2, H3, H4, H5, H6, H7; \
+   sph_u32 T0, T1;
+
+#define READ_STATE32_16WAY(state) \
+do { \
+   H0 = (state)->H[0]; \
+   H1 = (state)->H[1]; \
+   H2 = (state)->H[2]; \
+   H3 = (state)->H[3]; \
+   H4 = (state)->H[4]; \
+   H5 = (state)->H[5]; \
+   H6 = (state)->H[6]; \
+   H7 = (state)->H[7]; \
+   T0 = (state)->T0; \
+   T1 = (state)->T1; \
+} while (0)
+
+#define WRITE_STATE32_16WAY(state) \
+do { \
+   (state)->H[0] = H0; \
+   (state)->H[1] = H1; \
+   (state)->H[2] = H2; \
+   (state)->H[3] = H3; \
+   (state)->H[4] = H4; \
+   (state)->H[5] = H5; \
+   (state)->H[6] = H6; \
+   (state)->H[7] = H7; \
+   (state)->T0 = T0; \
+   (state)->T1 = T1; \
+} while (0)
+
+#define COMPRESS32_16WAY( rounds ) \
+do { \
+   __m512i M0, M1, M2, M3, M4, M5, M6, M7; \
+   __m512i M8, M9, MA, MB, MC, MD, ME, MF; \
+   __m512i V0, V1, V2, V3, V4, V5, V6, V7; \
+   __m512i V8, V9, VA, VB, VC, VD, VE, VF; \
+   __m512i shuf_bswap32; \
+   V0 = H0; \
+   V1 = H1; \
+   V2 = H2; \
+   V3 = H3; \
+   V4 = H4; \
+   V5 = H5; \
+   V6 = H6; \
+   V7 = H7; \
+   V8 = m512_const1_64( 0x243F6A88243F6A88 ); \
+   V9 = m512_const1_64( 0x85A308D385A308D3 ); \
+   VA = m512_const1_64( 0x13198A2E13198A2E ); \
+   VB = m512_const1_64( 0x0370734403707344 ); \
+   VC = _mm512_xor_si512( _mm512_set1_epi32( T0 ),\
+                              m512_const1_64( 0xA4093822A4093822 ) ); \
+   VD = _mm512_xor_si512( _mm512_set1_epi32( T0 ),\
+                              m512_const1_64( 0x299F31D0299F31D0 ) ); \
+   VE = _mm512_xor_si512( _mm512_set1_epi32( T1 ), \
+                              m512_const1_64( 0x082EFA98082EFA98 ) ); \
+   VF = _mm512_xor_si512( _mm512_set1_epi32( T1 ), \
+                              m512_const1_64( 0xEC4E6C89EC4E6C89 ) ); \
+   shuf_bswap32 = m512_const_64( 0x3c3d3e3f38393a3b, 0x3435363730313233, \
+                                 0x2c2d2e2f28292a2b, 0x2425262720212223, \
+                                 0x1c1d1e1f18191a1b, 0x1415161710111213, \
+                                 0x0c0d0e0f08090a0b, 0x0405060700010203 ); \
+   M0 = _mm512_shuffle_epi8( * buf    , shuf_bswap32 ); \
+   M1 = _mm512_shuffle_epi8( *(buf+ 1), shuf_bswap32 ); \
+   M2 = _mm512_shuffle_epi8( *(buf+ 2), shuf_bswap32 ); \
+   M3 = _mm512_shuffle_epi8( *(buf+ 3), shuf_bswap32 ); \
+   M4 = _mm512_shuffle_epi8( *(buf+ 4), shuf_bswap32 ); \
+   M5 = _mm512_shuffle_epi8( *(buf+ 5), shuf_bswap32 ); \
+   M6 = _mm512_shuffle_epi8( *(buf+ 6), shuf_bswap32 ); \
+   M7 = _mm512_shuffle_epi8( *(buf+ 7), shuf_bswap32 ); \
+   M8 = _mm512_shuffle_epi8( *(buf+ 8), shuf_bswap32 ); \
+   M9 = _mm512_shuffle_epi8( *(buf+ 9), shuf_bswap32 ); \
+   MA = _mm512_shuffle_epi8( *(buf+10), shuf_bswap32 ); \
+   MB = _mm512_shuffle_epi8( *(buf+11), shuf_bswap32 ); \
+   MC = _mm512_shuffle_epi8( *(buf+12), shuf_bswap32 ); \
+   MD = _mm512_shuffle_epi8( *(buf+13), shuf_bswap32 ); \
+   ME = _mm512_shuffle_epi8( *(buf+14), shuf_bswap32 ); \
+   MF = _mm512_shuffle_epi8( *(buf+15), shuf_bswap32 ); \
+   ROUND_S_16WAY(0); \
+   ROUND_S_16WAY(1); \
+   ROUND_S_16WAY(2); \
+   ROUND_S_16WAY(3); \
+   ROUND_S_16WAY(4); \
+   ROUND_S_16WAY(5); \
+   ROUND_S_16WAY(6); \
+   ROUND_S_16WAY(7); \
+   if (rounds == 14) \
+   { \
+      ROUND_S_16WAY(8); \
+      ROUND_S_16WAY(9); \
+      ROUND_S_16WAY(0); \
+      ROUND_S_16WAY(1); \
+      ROUND_S_16WAY(2); \
+      ROUND_S_16WAY(3); \
+   } \
+   H0 = _mm512_xor_si512( _mm512_xor_si512( V8, V0 ), H0 ); \
+   H1 = _mm512_xor_si512( _mm512_xor_si512( V9, V1 ), H1 ); \
+   H2 = _mm512_xor_si512( _mm512_xor_si512( VA, V2 ), H2 ); \
+   H3 = _mm512_xor_si512( _mm512_xor_si512( VB, V3 ), H3 ); \
+   H4 = _mm512_xor_si512( _mm512_xor_si512( VC, V4 ), H4 ); \
+   H5 = _mm512_xor_si512( _mm512_xor_si512( VD, V5 ), H5 ); \
+   H6 = _mm512_xor_si512( _mm512_xor_si512( VE, V6 ), H6 ); \
+   H7 = _mm512_xor_si512( _mm512_xor_si512( VF, V7 ), H7 ); \
+} while (0)
+
 #endif

 // Blake-256 4 way
@@ -703,7 +828,6 @@ static void
 blake32_4way_init( blake_4way_small_context *ctx, const uint32_t *iv,
                   const uint32_t *salt, int rounds )
 {
-   __m128i zero = m128_zero;
   casti_m128i( ctx->H, 0 ) = m128_const1_64( 0x6A09E6676A09E667 );
   casti_m128i( ctx->H, 1 ) = m128_const1_64( 0xBB67AE85BB67AE85 );
   casti_m128i( ctx->H, 2 ) = m128_const1_64( 0x3C6EF3723C6EF372 );
@@ -712,18 +836,14 @@ blake32_4way_init( blake_4way_small_context *ctx, const uint32_t *iv,
   casti_m128i( ctx->H, 5 ) = m128_const1_64( 0x9B05688C9B05688C );
   casti_m128i( ctx->H, 6 ) = m128_const1_64( 0x1F83D9AB1F83D9AB );
   casti_m128i( ctx->H, 7 ) = m128_const1_64( 0x5BE0CD195BE0CD19 );
-
-   casti_m128i( ctx->S, 0 ) = zero;
-   casti_m128i( ctx->S, 1 ) = zero;
-   casti_m128i( ctx->S, 2 ) = zero;
-   casti_m128i( ctx->S, 3 ) = zero;
   ctx->T0 = ctx->T1 = 0;
   ctx->ptr = 0;
   ctx->rounds = rounds;
 }

 static void
-blake32_4way( blake_4way_small_context *ctx, const void *data, size_t len )
+blake32_4way( blake_4way_small_context *ctx, const void *data,
+              size_t len )
 {
   __m128i *buf = (__m128i*)ctx->buf;
   size_t  bptr = ctx->ptr<<2;
@@ -824,7 +944,6 @@ static void
 blake32_8way_init( blake_8way_small_context *sc, const sph_u32 *iv,
                   const sph_u32 *salt, int rounds )
 {
-   __m256i zero = m256_zero;
   casti_m256i( sc->H, 0 ) = m256_const1_64( 0x6A09E6676A09E667 );
   casti_m256i( sc->H, 1 ) = m256_const1_64( 0xBB67AE85BB67AE85 );
   casti_m256i( sc->H, 2 ) = m256_const1_64( 0x3C6EF3723C6EF372 );
@@ -833,10 +952,6 @@ blake32_8way_init( blake_8way_small_context *sc, const sph_u32 *iv,
   casti_m256i( sc->H, 5 ) = m256_const1_64( 0x9B05688C9B05688C );
   casti_m256i( sc->H, 6 ) = m256_const1_64( 0x1F83D9AB1F83D9AB );
   casti_m256i( sc->H, 7 ) = m256_const1_64( 0x5BE0CD195BE0CD19 );
-   casti_m256i( sc->S, 0 ) = zero;
-   casti_m256i( sc->S, 1 ) = zero;
-   casti_m256i( sc->S, 2 ) = zero;
-   casti_m256i( sc->S, 3 ) = zero;
   sc->T0 = sc->T1 = 0;
   sc->ptr = 0;
   sc->rounds = rounds;
@@ -940,6 +1055,179 @@ blake32_8way_close( blake_8way_small_context *sc, unsigned ub, unsigned n,

 #endif

+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+//Blake-256 16 way AVX512
+
+static void
+blake32_16way_init( blake_16way_small_context *sc, const sph_u32 *iv,
+                   const sph_u32 *salt, int rounds )
+{
+   casti_m512i( sc->H, 0 ) = m512_const1_64( 0x6A09E6676A09E667 );
+   casti_m512i( sc->H, 1 ) = m512_const1_64( 0xBB67AE85BB67AE85 );
+   casti_m512i( sc->H, 2 ) = m512_const1_64( 0x3C6EF3723C6EF372 );
+   casti_m512i( sc->H, 3 ) = m512_const1_64( 0xA54FF53AA54FF53A );
+   casti_m512i( sc->H, 4 ) = m512_const1_64( 0x510E527F510E527F );
+   casti_m512i( sc->H, 5 ) = m512_const1_64( 0x9B05688C9B05688C );
+   casti_m512i( sc->H, 6 ) = m512_const1_64( 0x1F83D9AB1F83D9AB );
+   casti_m512i( sc->H, 7 ) = m512_const1_64( 0x5BE0CD195BE0CD19 );
+   sc->T0 = sc->T1 = 0;
+   sc->ptr = 0;
+   sc->rounds = rounds;
+}
+
+static void
+blake32_16way( blake_16way_small_context *sc, const void *data, size_t len )
+{
+   __m512i *vdata = (__m512i*)data;
+   __m512i *buf;
+   size_t ptr;
+   const int buf_size = 64;   // number of elements, sizeof/4
+   DECL_STATE32_16WAY
+   buf = sc->buf;
+   ptr = sc->ptr;
+   if ( len < buf_size - ptr )
+   {
+        memcpy_512( buf + (ptr>>2), vdata, len>>2 );
+        ptr += len;
+        sc->ptr = ptr;
+        return;
+   }
+   READ_STATE32_16WAY(sc);
+   while ( len > 0 )
+   {
+      size_t clen;
+
+      clen = buf_size - ptr;
+      if (clen > len)
+           clen = len;
+      memcpy_512( buf + (ptr>>2), vdata, clen>>2 );
+      ptr += clen;
+      vdata += (clen>>2);
+      len -= clen;
+      if ( ptr == buf_size )
+      {
+          if ( ( T0 = T0 + 512 ) < 512 )
+                T1 = T1 + 1;
+          COMPRESS32_16WAY( sc->rounds );
+          ptr = 0;
+      }
+   }
+   WRITE_STATE32_16WAY(sc);
+   sc->ptr = ptr;
+}
+
+static void
+blake32_16way_close( blake_16way_small_context *sc, unsigned ub, unsigned n,
+                    void *dst, size_t out_size_w32 )
+{
+   __m512i buf[16];
+   size_t ptr;
+   unsigned bit_len;
+   sph_u32 th, tl;
+
+   ptr = sc->ptr;
+   bit_len = ((unsigned)ptr << 3);
+   buf[ptr>>2] = m512_const1_64( 0x0000008000000080ULL );
+   tl = sc->T0 + bit_len;
+   th = sc->T1;
+
+   if ( ptr == 0 )
+   {
+        sc->T0 = 0xFFFFFE00UL;
+        sc->T1 = 0xFFFFFFFFUL;
+   }
+   else if ( sc->T0 == 0 )
+   {
+        sc->T0 = 0xFFFFFE00UL + bit_len;
+        sc->T1 = sc->T1 - 1;
+   }
+   else
+        sc->T0 -= 512 - bit_len;
+
+   if ( ptr <= 52 )
+   {
+       memset_zero_512( buf + (ptr>>2) + 1, (52 - ptr) >> 2 );
+       if ( out_size_w32 == 8 )
+           buf[52>>2] = _mm512_or_si512( buf[52>>2],
+                                m512_const1_64( 0x0100000001000000ULL ) );
+       buf[+56>>2] = mm512_bswap_32( _mm512_set1_epi32( th ) );
+       buf[+60>>2] = mm512_bswap_32( _mm512_set1_epi32( tl ) );
+       blake32_16way( sc, buf + (ptr>>2), 64 - ptr );
+   }
+   else
+   {
+        memset_zero_512( buf + (ptr>>2) + 1, (60-ptr) >> 2 );
+        blake32_16way( sc, buf + (ptr>>2), 64 - ptr );
+        sc->T0 = 0xFFFFFE00UL;
+        sc->T1 = 0xFFFFFFFFUL;
+        memset_zero_512( buf, 56>>2 );
+       if ( out_size_w32 == 8 )
+           buf[52>>2] = m512_const1_64( 0x0100000001000000ULL );
+        buf[56>>2] = mm512_bswap_32( _mm512_set1_epi32( th ) );
+        buf[60>>2] = mm512_bswap_32( _mm512_set1_epi32( tl ) );
+        blake32_16way( sc, buf, 64 );
+   }
+   mm512_block_bswap_32( (__m512i*)dst, (__m512i*)sc->H );
+}
+
+void
+blake256_16way_init(void *cc)
+{
+   blake32_16way_init( cc, IV256, salt_zero_8way_small, 14 );
+}
+
+void
+blake256_16way_update(void *cc, const void *data, size_t len)
+{
+        blake32_16way(cc, data, len);
+}
+
+void
+blake256_16way_close(void *cc, void *dst)
+{
+        blake32_16way_close(cc, 0, 0, dst, 8);
+}
+
+void blake256r14_16way_init(void *cc)
+{
+   blake32_16way_init( cc, IV256, salt_zero_8way_small, 14 );
+}
+
+void
+blake256r14_16way_update(void *cc, const void *data, size_t len)
+{
+   blake32_16way(cc, data, len);
+}
+
+void
+blake256r14_16way_close(void *cc, void *dst)
+{
+   blake32_16way_close(cc, 0, 0, dst, 8);
+}
+
+void blake256r8_16way_init(void *cc)
+{
+   blake32_16way_init( cc, IV256, salt_zero_8way_small, 8 );
+}
+
+void
+blake256r8_16way_update(void *cc, const void *data, size_t len)
+{
+   blake32_16way(cc, data, len);
+}
+
+void
+blake256r8_16way_close(void *cc, void *dst)
+{
+   blake32_16way_close(cc, 0, 0, dst, 8);
+}
+
+#endif // AVX512
+
+
+
 // Blake-256 4 way

 // default 14 rounds, backward copatibility
@@ -950,7 +1238,7 @@ blake256_4way_init(void *ctx)
 }

 void
-blake256_4way(void *ctx, const void *data, size_t len)
+blake256_4way_update(void *ctx, const void *data, size_t len)
 {
 	blake32_4way(ctx, data, len);
 }
@@ -972,7 +1260,7 @@ blake256_8way_init(void *cc)
 }

 void
-blake256_8way(void *cc, const void *data, size_t len)
+blake256_8way_update(void *cc, const void *data, size_t len)
 {
        blake32_8way(cc, data, len);
 }
@@ -992,7 +1280,7 @@ void blake256r14_4way_init(void *cc)
 }

 void
-blake256r14_4way(void *cc, const void *data, size_t len)
+blake256r14_4way_update(void *cc, const void *data, size_t len)
 {
   blake32_4way(cc, data, len);
 }
@@ -1011,7 +1299,7 @@ void blake256r14_8way_init(void *cc)
 }

 void
-blake256r14_8way(void *cc, const void *data, size_t len)
+blake256r14_8way_update(void *cc, const void *data, size_t len)
 {
   blake32_8way(cc, data, len);
 }
@@ -1031,7 +1319,7 @@ void blake256r8_4way_init(void *cc)
 }

 void
-blake256r8_4way(void *cc, const void *data, size_t len)
+blake256r8_4way_update(void *cc, const void *data, size_t len)
 {
   blake32_4way(cc, data, len);
 }
@@ -1050,7 +1338,7 @@ void blake256r8_8way_init(void *cc)
 }

 void
-blake256r8_8way(void *cc, const void *data, size_t len)
+blake256r8_8way_update(void *cc, const void *data, size_t len)
 {
   blake32_8way(cc, data, len);
 }
--- a/algo/blake/blake2b-4way.c
+++ b/algo/blake/blake2b-4way.c
@@ -4,13 +4,59 @@
 */

 #include "blake2b-gate.h"
-
-#if defined(BLAKE2B_4WAY)
-
 #include <string.h>
 #include <stdint.h>
 #include "blake2b-hash-4way.h"

+#if defined(BLAKE2B_8WAY)
+
+int scanhash_blake2b_8way( struct work *work, uint32_t max_nonce,
+                           uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t hash[8*8] __attribute__ ((aligned (128)));;
+   uint32_t vdata[20*8] __attribute__ ((aligned (64)));;
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   blake2b_8way_ctx ctx __attribute__ ((aligned (64)));
+   uint32_t *hash7 = &(hash[49]);   // 3*16+1
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   int thr_id = mythr->id;
+   __m512i  *noncev = (__m512i*)vdata + 9;   // aligned
+   const uint32_t Htarg = ptarget[7];
+   const uint32_t first_nonce = pdata[19];
+
+   uint32_t n = first_nonce;
+
+   mm512_bswap32_intrlv80_8x64( vdata, pdata );
+
+   do {
+      *noncev = mm512_intrlv_blend_32( mm512_bswap_32(
+                _mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
+                                  n+3, 0, n+2, 0, n+1, 0, n  , 0 ) ), *noncev );
+
+      blake2b_8way_init( &ctx );
+      blake2b_8way_update( &ctx, vdata, 80 );
+      blake2b_8way_final( &ctx, hash );
+
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( hash7[ lane<<1 ] <= Htarg )
+      {
+          extr_lane_8x64( lane_hash, hash, lane, 256 );
+          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+          {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+          }
+      }
+      n += 8;
+   } while ( (n < max_nonce-8) && !work_restart[thr_id].restart);
+
+   *hashes_done = n - first_nonce + 1;
+   return 0;
+}
+
+#elif defined(BLAKE2B_4WAY)
+
 // Function not used, code inlined.
 void blake2b_4way_hash(void *output, const void *input)
 {
@@ -48,13 +94,13 @@ int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
      blake2b_4way_final( &ctx, hash );

      for ( int lane = 0; lane < 4; lane++ )
-      if ( hash7[ lane<<1 ] < Htarg )
+      if ( hash7[ lane<<1 ] <= Htarg )
      {
          extr_lane_4x64( lane_hash, hash, lane, 256 );
          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 4;
--- a/algo/blake/blake2b-gate.c
+++ b/algo/blake/blake2b-gate.c
@@ -1,24 +1,19 @@
 #include "blake2b-gate.h"

-/*
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blake2s_get_max64 ()
-{
-   return 0x7ffffLL;
-}
-*/

 bool register_blake2b_algo( algo_gate_t* gate )
 {
-#if defined(BLAKE2B_4WAY)
+#if defined(BLAKE2B_8WAY)
+  gate->scanhash  = (void*)&scanhash_blake2b_8way;
+//  gate->hash      = (void*)&blake2b_8way_hash;
+#elif defined(BLAKE2B_4WAY)
  gate->scanhash  = (void*)&scanhash_blake2b_4way;
  gate->hash      = (void*)&blake2b_4way_hash;
 #else
  gate->scanhash  = (void*)&scanhash_blake2b;
  gate->hash      = (void*)&blake2b_hash;
 #endif
-//  gate->get_max64 = (void*)&blake2s_get_max64;
-  gate->optimizations =  AVX2_OPT;
+  gate->optimizations =  AVX2_OPT | AVX512_OPT;
  return true;
 };

--- a/algo/blake/blake2b-gate.h
+++ b/algo/blake/blake2b-gate.h
@@ -4,13 +4,21 @@
 #include <stdint.h>
 #include "algo-gate-api.h"

-#if defined(__AVX2__)
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+  #define BLAKE2B_8WAY
+#elif defined(__AVX2__)
  #define BLAKE2B_4WAY
 #endif

 bool register_blake2b_algo( algo_gate_t* gate );

-#if defined(BLAKE2B_4WAY)
+#if defined(BLAKE2B_8WAY)
+
+//void blake2b_8way_hash( void *state, const void *input );
+int scanhash_blake2b_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+
+#elif defined(BLAKE2B_4WAY)

 void blake2b_4way_hash( void *state, const void *input );
 int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
--- a/algo/blake/blake2b-hash-4way.c
+++ b/algo/blake/blake2b-hash-4way.c
@@ -35,6 +35,180 @@

 #if defined(__AVX2__)

+static const uint8_t sigma[12][16] =
+{
+      { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
+      { 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
+      { 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
+      { 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
+      { 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
+      { 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
+      { 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
+      { 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
+      { 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
+      { 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
+      { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
+      { 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
+};
+
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+#define B2B8W_G(a, b, c, d, x, y) \
+{ \
+   v[a] = _mm512_add_epi64( _mm512_add_epi64( v[a], v[b] ), x ); \
+   v[d] = mm512_ror_64( _mm512_xor_si512( v[d], v[a] ), 32 ); \
+   v[c] = _mm512_add_epi64( v[c], v[d] ); \
+   v[b] = mm512_ror_64( _mm512_xor_si512( v[b], v[c] ), 24 ); \
+   v[a] = _mm512_add_epi64( _mm512_add_epi64( v[a], v[b] ), y ); \
+   v[d] = mm512_ror_64( _mm512_xor_si512( v[d], v[a] ), 16 ); \
+   v[c] = _mm512_add_epi64( v[c], v[d] ); \
+   v[b] = mm512_ror_64( _mm512_xor_si512( v[b], v[c] ), 63 ); \
+}
+
+static void blake2b_8way_compress( blake2b_8way_ctx *ctx, int last )
+{  
+   __m512i v[16], m[16];
+
+   v[ 0] = ctx->h[0];
+   v[ 1] = ctx->h[1];
+   v[ 2] = ctx->h[2];
+   v[ 3] = ctx->h[3];
+   v[ 4] = ctx->h[4];
+   v[ 5] = ctx->h[5];
+   v[ 6] = ctx->h[6];
+   v[ 7] = ctx->h[7];
+   v[ 8] = m512_const1_64( 0x6A09E667F3BCC908 );
+   v[ 9] = m512_const1_64( 0xBB67AE8584CAA73B );
+   v[10] = m512_const1_64( 0x3C6EF372FE94F82B );
+   v[11] = m512_const1_64( 0xA54FF53A5F1D36F1 );
+   v[12] = m512_const1_64( 0x510E527FADE682D1 );
+   v[13] = m512_const1_64( 0x9B05688C2B3E6C1F );
+   v[14] = m512_const1_64( 0x1F83D9ABFB41BD6B );
+   v[15] = m512_const1_64( 0x5BE0CD19137E2179 );
+
+   v[12] = _mm512_xor_si512( v[12], _mm512_set1_epi64( ctx->t[0] ) );
+   v[13] = _mm512_xor_si512( v[13], _mm512_set1_epi64( ctx->t[1] ) );
+
+   if ( last )
+      v[14] = mm512_not( v[14] );
+
+   m[ 0] = ctx->b[ 0];
+   m[ 1] = ctx->b[ 1];
+   m[ 2] = ctx->b[ 2];
+   m[ 3] = ctx->b[ 3];
+   m[ 4] = ctx->b[ 4];
+   m[ 5] = ctx->b[ 5];
+   m[ 6] = ctx->b[ 6];
+   m[ 7] = ctx->b[ 7];
+   m[ 8] = ctx->b[ 8];
+   m[ 9] = ctx->b[ 9];
+   m[10] = ctx->b[10];
+   m[11] = ctx->b[11];
+   m[12] = ctx->b[12];
+   m[13] = ctx->b[13];
+   m[14] = ctx->b[14];
+   m[15] = ctx->b[15];
+
+   for ( int i = 0; i < 12; i++ )
+   {
+      B2B8W_G( 0, 4,  8, 12, m[ sigma[i][ 0] ], m[ sigma[i][ 1] ] );
+      B2B8W_G( 1, 5,  9, 13, m[ sigma[i][ 2] ], m[ sigma[i][ 3] ] );
+      B2B8W_G( 2, 6, 10, 14, m[ sigma[i][ 4] ], m[ sigma[i][ 5] ] );
+      B2B8W_G( 3, 7, 11, 15, m[ sigma[i][ 6] ], m[ sigma[i][ 7] ] );
+      B2B8W_G( 0, 5, 10, 15, m[ sigma[i][ 8] ], m[ sigma[i][ 9] ] );
+      B2B8W_G( 1, 6, 11, 12, m[ sigma[i][10] ], m[ sigma[i][11] ] );
+      B2B8W_G( 2, 7,  8, 13, m[ sigma[i][12] ], m[ sigma[i][13] ] );
+      B2B8W_G( 3, 4,  9, 14, m[ sigma[i][14] ], m[ sigma[i][15] ] );
+   }
+
+   ctx->h[0] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[0], v[0] ), v[ 8] );
+   ctx->h[1] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[1], v[1] ), v[ 9] );
+   ctx->h[2] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[2], v[2] ), v[10] );
+   ctx->h[3] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[3], v[3] ), v[11] );
+   ctx->h[4] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[4], v[4] ), v[12] );
+   ctx->h[5] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[5], v[5] ), v[13] );
+   ctx->h[6] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[6], v[6] ), v[14] );
+   ctx->h[7] = _mm512_xor_si512( _mm512_xor_si512( ctx->h[7], v[7] ), v[15] );
+}
+
+int blake2b_8way_init( blake2b_8way_ctx *ctx )
+{
+   size_t i;
+
+   ctx->h[0] = m512_const1_64( 0x6A09E667F3BCC908 );
+   ctx->h[1] = m512_const1_64( 0xBB67AE8584CAA73B );
+   ctx->h[2] = m512_const1_64( 0x3C6EF372FE94F82B );
+   ctx->h[3] = m512_const1_64( 0xA54FF53A5F1D36F1 );
+   ctx->h[4] = m512_const1_64( 0x510E527FADE682D1 );
+   ctx->h[5] = m512_const1_64( 0x9B05688C2B3E6C1F );
+   ctx->h[6] = m512_const1_64( 0x1F83D9ABFB41BD6B );
+   ctx->h[7] = m512_const1_64( 0x5BE0CD19137E2179 );
+
+   ctx->h[0] = _mm512_xor_si512( ctx->h[0], m512_const1_64( 0x01010020 ) );
+
+   ctx->t[0] = 0;
+   ctx->t[1] = 0;
+   ctx->c = 0;
+   ctx->outlen = 32;
+
+   for ( i = 0; i < 16; i++ )
+     ctx->b[i] = m512_zero;
+
+   return 0;
+}
+
+
+void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
+                          size_t inlen )
+{
+   __m512i* in =(__m512i*)input;
+
+   size_t i, c;
+   c = ctx->c >> 3;
+
+   for ( i = 0; i < (inlen >> 3); i++ )
+   {
+      if ( ctx->c == 128 )
+      {
+         ctx->t[0] += ctx->c;
+         if ( ctx->t[0] < ctx->c )
+            ctx->t[1]++;
+         blake2b_8way_compress( ctx, 0 );
+         ctx->c = 0;
+      }
+      ctx->b[ c++ ] = in[i];
+      ctx->c += 8;
+   }
+}
+
+void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out )
+{
+   size_t c;
+   c = ctx->c >> 3;
+
+   ctx->t[0] += ctx->c;
+   if ( ctx->t[0] < ctx->c )
+      ctx->t[1]++;
+
+   while ( ctx->c < 128 )
+   {
+      ctx->b[c++] = m512_zero;
+      ctx->c += 8;
+   }
+
+   blake2b_8way_compress( ctx, 1 );           // final block flag = 1
+
+   casti_m512i( out, 0 ) = ctx->h[0];
+   casti_m512i( out, 1 ) = ctx->h[1];
+   casti_m512i( out, 2 ) = ctx->h[2];
+   casti_m512i( out, 3 ) = ctx->h[3];
+}
+
+#endif   // AVX512
+
+// AVX2
+
 // G Mixing function.

 #define B2B_G(a, b, c, d, x, y) \
@@ -61,21 +235,6 @@ static const uint64_t blake2b_iv[8] = {

 static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
 {
-	const uint8_t sigma[12][16] = {
-		{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
-		{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
-		{ 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
-		{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
-		{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
-		{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
-		{ 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
-		{ 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
-		{ 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
-		{ 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
-		{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
-		{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
-	};
-	int i;
 	__m256i v[16], m[16];

   v[ 0] = ctx->h[0];
@@ -118,7 +277,7 @@ static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
   m[14] = ctx->b[14];
   m[15] = ctx->b[15];
   
-	for ( i = 0; i < 12; i++ )
+	for ( int i = 0; i < 12; i++ )
   { 
 		B2B_G( 0, 4,  8, 12, m[ sigma[i][ 0] ], m[ sigma[i][ 1] ] );
 		B2B_G( 1, 5,  9, 13, m[ sigma[i][ 2] ], m[ sigma[i][ 3] ] );
@@ -212,4 +371,4 @@ void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out )
   casti_m256i( out, 3 ) = ctx->h[3];
 }

-#endif
+#endif  // AVX2
--- a/algo/blake/blake2b-hash-4way.h
+++ b/algo/blake/blake2b-hash-4way.h
@@ -2,8 +2,6 @@
 #ifndef __BLAKE2B_HASH_4WAY_H__
 #define __BLAKE2B_HASH_4WAY_H__

-#if defined(__AVX2__)
-
 #include "simd-utils.h"
 #include <stddef.h>
 #include <stdint.h>
@@ -16,14 +14,34 @@
 #define ALIGN(x) __attribute__((aligned(x)))
 #endif

+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+typedef struct ALIGN( 64 ) {
+   __m512i b[16]; // input buffer
+   __m512i h[8];  // chained state
+   uint64_t t[2];  // total number of bytes
+   size_t c;       // pointer for b[]
+   size_t outlen;  // digest size
+} blake2b_8way_ctx;
+
+int blake2b_8way_init( blake2b_8way_ctx *ctx );
+void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
+                          size_t inlen );
+void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out );
+
+#endif
+
+#if defined(__AVX2__)
+
 // state context
-ALIGN(64) typedef struct {
+typedef struct ALIGN( 64 ) {
 	__m256i b[16]; // input buffer
 	__m256i h[8];  // chained state
 	uint64_t t[2];  // total number of bytes
 	size_t c;       // pointer for b[]
 	size_t outlen;  // digest size
-} blake2b_4way_ctx __attribute__((aligned(64)));
+} blake2b_4way_ctx;

 int blake2b_4way_init( blake2b_4way_ctx *ctx );
 void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
--- a/algo/blake/blake2b.c
+++ b/algo/blake/blake2b.c
@@ -4,6 +4,9 @@
 */

 #include "blake2b-gate.h"
+
+#if !defined(BLAKE2B_8WAY) && !defined(BLAKE2B_4WAY)
+
 #include <string.h>
 #include <stdint.h>
 #include "algo/blake/sph_blake2b.h"
@@ -43,17 +46,14 @@ int scanhash_blake2b( struct work *work, uint32_t max_nonce,

 	do {
 		be32enc(&endiandata[19], n);
-		//blake2b_hash_end(vhashcpu, endiandata);
 		blake2b_hash(vhashcpu, endiandata);

-		if (vhashcpu[7] < Htarg && fulltest(vhashcpu, ptarget)) {
-			work_set_target_ratio(work, vhashcpu);
-			*hashes_done = n - first_nonce + 1;
+		if (vhashcpu[7] <= Htarg && fulltest(vhashcpu, ptarget))
+      {
 			pdata[19] = n;
-			return 1;
-		}
-		n++;
-
+         submit_solution( work, vhashcpu, mythr );
+      }
+      n++;
 	} while (n < max_nonce && !work_restart[thr_id].restart);
 	*hashes_done = n - first_nonce + 1;
 	pdata[19] = n;
@@ -61,3 +61,4 @@ int scanhash_blake2b( struct work *work, uint32_t max_nonce,
 	return 0;
 }

+#endif
--- a/algo/blake/blake2s-4way.c
+++ b/algo/blake/blake2s-4way.c
@@ -3,22 +3,72 @@
 #include <string.h>
 #include <stdint.h>

-#if defined(BLAKE2S_8WAY)
+#if defined(BLAKE2S_16WAY)
+
+static __thread blake2s_16way_state blake2s_16w_ctx;
+
+void blake2s_16way_hash( void *output, const void *input )
+{
+   blake2s_16way_state ctx;
+   memcpy( &ctx, &blake2s_16w_ctx, sizeof ctx );
+   blake2s_16way_update( &ctx, input + (64<<4), 16 );
+   blake2s_16way_final( &ctx, output, BLAKE2S_OUTBYTES );
+}
+
+int scanhash_blake2s_16way( struct work *work, uint32_t max_nonce,
+                            uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t vdata[20*16] __attribute__ ((aligned (128)));
+   uint32_t hash[8*16] __attribute__ ((aligned (64)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   uint32_t *hash7 = &(hash[7<<4]);
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t Htarg = ptarget[7];
+   const uint32_t first_nonce = pdata[19];
+   __m512i  *noncev = (__m512i*)vdata + 19;   // aligned
+   uint32_t n = first_nonce;
+   int thr_id = mythr->id;  
+
+   mm512_bswap32_intrlv80_16x32( vdata, pdata );
+   blake2s_16way_init( &blake2s_16w_ctx, BLAKE2S_OUTBYTES );
+   blake2s_16way_update( &blake2s_16w_ctx, vdata, 64 );
+
+   do {
+      *noncev = mm512_bswap_32( _mm512_set_epi32(
+	                  n+15, n+14, n+13, n+12, n+11, n+10, n+ 9, n+ 8,
+	                  n+ 7, n+ 6, n+ 5, n+ 4, n+ 3, n+ 2, n+ 1, n ) );
+      pdata[19] = n;
+
+      blake2s_16way_hash( hash, vdata );
+
+      for ( int lane = 0; lane < 16; lane++ )
+      if ( unlikely( hash7[lane] <= Htarg ) )
+      {
+         extr_lane_16x32( lane_hash, hash, lane, 256 );
+         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
+         {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+         }
+      }
+      n += 16;
+   } while ( (n < max_nonce-16) && !work_restart[thr_id].restart );
+
+   *hashes_done = n - first_nonce + 1;
+   return 0;
+}
+
+#elif defined(BLAKE2S_8WAY)

 static __thread blake2s_8way_state blake2s_8w_ctx;

 void blake2s_8way_hash( void *output, const void *input )
 {
-   uint32_t vhash[8*8] __attribute__ ((aligned (64)));
   blake2s_8way_state ctx;
   memcpy( &ctx, &blake2s_8w_ctx, sizeof ctx );
-
   blake2s_8way_update( &ctx, input + (64<<3), 16 );
-   blake2s_8way_final( &ctx, vhash, BLAKE2S_OUTBYTES );
-
-   dintrlv_8x32( output,     output+ 32, output+ 64, output+ 96,
-                 output+128, output+160, output+192, output+224,
-                 vhash, 256 );
+   blake2s_8way_final( &ctx, output, BLAKE2S_OUTBYTES );
 }

 int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
@@ -26,13 +76,15 @@ int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
 {
   uint32_t vdata[20*8] __attribute__ ((aligned (64)));
   uint32_t hash[8*8] __attribute__ ((aligned (32)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[7<<3]);
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   int thr_id = mythr->id; 

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
   blake2s_8way_init( &blake2s_8w_ctx, BLAKE2S_OUTBYTES );
@@ -45,16 +97,17 @@ int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,

      blake2s_8way_hash( hash, vdata );

-
-      for ( int i = 0; i < 8; i++ )
-      if (  (hash+(i<<3))[7] <= Htarg )
-      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[lane] <= Htarg ) )
      {
-          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+         extr_lane_8x32( lane_hash, hash, lane, 256 );
+         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
+         {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+         }
      }
      n += 8;
-
   } while ( (n < max_nonce) && !work_restart[thr_id].restart );

   *hashes_done = n - first_nonce + 1;
@@ -67,15 +120,10 @@ static __thread blake2s_4way_state blake2s_4w_ctx;

 void blake2s_4way_hash( void *output, const void *input )
 {
-   uint32_t vhash[8*4] __attribute__ ((aligned (64)));
   blake2s_4way_state ctx;
   memcpy( &ctx, &blake2s_4w_ctx, sizeof ctx );
-
   blake2s_4way_update( &ctx, input + (64<<2), 16 );
-   blake2s_4way_final( &ctx, vhash, BLAKE2S_OUTBYTES );
-
-   dintrlv_4x32( output, output+32, output+64, output+96,
-		            vhash, 256 );
+   blake2s_4way_final( &ctx, output, BLAKE2S_OUTBYTES );
 }

 int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,
@@ -83,13 +131,15 @@ int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,
 {
   uint32_t vdata[20*4] __attribute__ ((aligned (64)));
   uint32_t hash[8*4] __attribute__ ((aligned (32)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[7<<2]);
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   int thr_id = mythr->id; 

   mm128_bswap32_intrlv80_4x32( vdata, pdata );
   blake2s_4way_init( &blake2s_4w_ctx, BLAKE2S_OUTBYTES );
@@ -101,15 +151,16 @@ int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,

      blake2s_4way_hash( hash, vdata );

-      for ( int i = 0; i < 4; i++ )
-      if ( (hash+(i<<3))[7] <= Htarg )
-      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
+      for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
      {
-          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+         extr_lane_4x32( lane_hash, hash, lane, 256 );
+         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+         {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+              }
      }
      n += 4;
-
   } while ( (n < max_nonce) && !work_restart[thr_id].restart );

   *hashes_done = n - first_nonce + 1;
--- a/algo/blake/blake2s-gate.c
+++ b/algo/blake/blake2s-gate.c
@@ -1,15 +1,12 @@
 #include "blake2s-gate.h"

-
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blake2s_get_max64 ()
-{
-   return 0x7ffffLL;
-}
-
 bool register_blake2s_algo( algo_gate_t* gate )
 {
-#if defined(BLAKE2S_8WAY)
+#if defined(BLAKE2S_16WAY)
+  gate->scanhash  = (void*)&scanhash_blake2s_16way;
+  gate->hash      = (void*)&blake2s_16way_hash;
+#elif defined(BLAKE2S_8WAY)
+//#if defined(BLAKE2S_8WAY)
  gate->scanhash  = (void*)&scanhash_blake2s_8way;
  gate->hash      = (void*)&blake2s_8way_hash;
 #elif defined(BLAKE2S_4WAY)
@@ -19,8 +16,7 @@ bool register_blake2s_algo( algo_gate_t* gate )
  gate->scanhash  = (void*)&scanhash_blake2s;
  gate->hash      = (void*)&blake2s_hash;
 #endif
-  gate->get_max64 = (void*)&blake2s_get_max64;
-  gate->optimizations = SSE2_OPT | AVX2_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
  return true;
 };

--- a/algo/blake/blake2s-gate.h
+++ b/algo/blake/blake2s-gate.h
@@ -8,13 +8,26 @@
 #if defined(__SSE2__)
  #define BLAKE2S_4WAY
 #endif
+
 #if defined(__AVX2__)
  #define BLAKE2S_8WAY
 #endif

+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+  #define BLAKE2S_16WAY
+#endif
+
 bool register_blake2s_algo( algo_gate_t* gate );

-#if defined(BLAKE2S_8WAY)
+#if defined(BLAKE2S_16WAY)
+
+void blake2s_16way_hash( void *state, const void *input );
+int scanhash_blake2s_16way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+
+#elif defined (BLAKE2S_8WAY)
+
+//#if defined(BLAKE2S_8WAY)

 void blake2s_8way_hash( void *state, const void *input );
 int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
--- a/algo/blake/blake2s-hash-4way.c
+++ b/algo/blake/blake2s-hash-4way.c
@@ -20,12 +20,13 @@
 //#if defined(__SSE4_2__)
 #if defined(__SSE2__)

-
+/*
 static const uint32_t blake2s_IV[8] =
 {
 	0x6A09E667UL, 0xBB67AE85UL, 0x3C6EF372UL, 0xA54FF53AUL,
 	0x510E527FUL, 0x9B05688CUL, 0x1F83D9ABUL, 0x5BE0CD19UL
 };
+*/

 static const uint8_t blake2s_sigma[10][16] =
 {
@@ -41,6 +42,7 @@ static const uint8_t blake2s_sigma[10][16] =
 	{ 10,  2,  8,  4,  7,  6,  1,  5, 15, 11,  9, 14,  3, 12, 13 , 0 } ,
 };

+
 // define a constant for initial param.

 int blake2s_4way_init( blake2s_4way_state *S, const uint8_t outlen )
@@ -88,41 +90,45 @@ int blake2s_4way_compress( blake2s_4way_state *S, const __m128i* block )
   memcpy_128( m, block, 16 );
   memcpy_128( v, S->h, 8 );

-   v[ 8] = _mm_set1_epi32( blake2s_IV[0] );
-   v[ 9] = _mm_set1_epi32( blake2s_IV[1] );
-   v[10] = _mm_set1_epi32( blake2s_IV[2] );
-   v[11] = _mm_set1_epi32( blake2s_IV[3] );
+   v[ 8] = m128_const1_64( 0x6A09E6676A09E667ULL );
+   v[ 9] = m128_const1_64( 0xBB67AE85BB67AE85ULL );
+   v[10] = m128_const1_64( 0x3C6EF3723C6EF372ULL );
+   v[11] = m128_const1_64( 0xA54FF53AA54FF53AULL );
   v[12] = _mm_xor_si128( _mm_set1_epi32( S->t[0] ),
-                          _mm_set1_epi32( blake2s_IV[4] ) );
+                          m128_const1_64( 0x510E527F510E527FULL ) );
   v[13] = _mm_xor_si128( _mm_set1_epi32( S->t[1] ),
-                          _mm_set1_epi32( blake2s_IV[5] ) );
+                          m128_const1_64( 0x9B05688C9B05688CULL ) );
   v[14] = _mm_xor_si128( _mm_set1_epi32( S->f[0] ),
-                          _mm_set1_epi32( blake2s_IV[6] ) );
+                          m128_const1_64( 0x1F83D9AB1F83D9ABULL ) );
   v[15] = _mm_xor_si128( _mm_set1_epi32( S->f[1] ),
-                          _mm_set1_epi32( blake2s_IV[7] ) );
+                          m128_const1_64( 0x5BE0CD195BE0CD19ULL ) );

-#define G4W(r,i,a,b,c,d) \
+#define G4W( sigma0, sigma1, a, b, c, d ) \
 do { \
-   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ blake2s_sigma[r][2*i+0] ] ); \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s0 ] ); \
   d = mm128_ror_32( _mm_xor_si128( d, a ), 16 ); \
   c = _mm_add_epi32( c, d ); \
   b = mm128_ror_32( _mm_xor_si128( b, c ), 12 ); \
-   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ blake2s_sigma[r][2*i+1] ] ); \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s1 ] ); \
   d = mm128_ror_32( _mm_xor_si128( d, a ),  8 ); \
   c = _mm_add_epi32( c, d ); \
   b = mm128_ror_32( _mm_xor_si128( b, c ),  7 ); \
 } while(0)

+
 #define ROUND4W(r)  \
 do { \
-   G4W( r, 0, v[ 0], v[ 4], v[ 8], v[12] ); \
-   G4W( r, 1, v[ 1], v[ 5], v[ 9], v[13] ); \
-   G4W( r, 2, v[ 2], v[ 6], v[10], v[14] ); \
-   G4W( r, 3, v[ 3], v[ 7], v[11], v[15] ); \
-   G4W( r, 4, v[ 0], v[ 5], v[10], v[15] ); \
-   G4W( r, 5, v[ 1], v[ 6], v[11], v[12] ); \
-   G4W( r, 6, v[ 2], v[ 7], v[ 8], v[13] ); \
-   G4W( r, 7, v[ 3], v[ 4], v[ 9], v[14] ); \
+   uint8_t *sigma = (uint8_t*)&blake2s_sigma[r]; \
+   G4W( sigma[ 0], sigma[ 1], v[ 0], v[ 4], v[ 8], v[12] ); \
+   G4W( sigma[ 2], sigma[ 3], v[ 1], v[ 5], v[ 9], v[13] ); \
+   G4W( sigma[ 4], sigma[ 5], v[ 2], v[ 6], v[10], v[14] ); \
+   G4W( sigma[ 6], sigma[ 7], v[ 3], v[ 7], v[11], v[15] ); \
+   G4W( sigma[ 8], sigma[ 9], v[ 0], v[ 5], v[10], v[15] ); \
+   G4W( sigma[10], sigma[11], v[ 1], v[ 6], v[11], v[12] ); \
+   G4W( sigma[12], sigma[13], v[ 2], v[ 7], v[ 8], v[13] ); \
+   G4W( sigma[14], sigma[15], v[ 3], v[ 4], v[ 9], v[14] ); \
 } while(0)

   ROUND4W( 0 );
@@ -144,26 +150,47 @@ do { \
   return 0;
 }

+// There is a problem that can't be resolved internally.
+// If the last block is a full 64 bytes it should not be compressed in
+// update but left for final. However, when streaming, it isn't known
+// which block is last. There may be a subsequent call to update to add
+// more data.
+//
+// The reference code handled this by juggling 2 blocks at a time at
+// a significant performance penalty.
+//
+// Instead a new function is introduced called full_blocks which combines
+// update and final and is to be used in non-streaming mode where the data
+// is a multiple of 64 bytes.
+// 
+// Supported:
+//    64 + 16 bytes  (blake2s with midstate optimization)
+//    80 bytes       (blake2s without midstate optimization)
+//    Any multiple of 64 bytes in one shot (x25x)
+//
+// Unsupported:
+//    Stream of full 64 byte blocks one at a time.   
+
+// use only when streaming more data or final block not full.
 int blake2s_4way_update( blake2s_4way_state *S, const void *in,
                         uint64_t inlen )
 {
-  __m128i *input = (__m128i*)in;
-  __m128i *buf = (__m128i*)S->buf;
-  const int bsize = BLAKE2S_BLOCKBYTES;
+   __m128i *input = (__m128i*)in;
+   __m128i *buf = (__m128i*)S->buf;

   while( inlen > 0 )
   {
      size_t left = S->buflen;
-      if( inlen >= bsize - left )
+      if( inlen >= BLAKE2S_BLOCKBYTES - left )
      {
-         memcpy_128( buf + (left>>2), input, (bsize - left) >> 2 );
-         S->buflen += bsize - left;
+         memcpy_128( buf + (left>>2), input, (BLAKE2S_BLOCKBYTES - left) >> 2 );
+         S->buflen += BLAKE2S_BLOCKBYTES - left;
         S->t[0] += BLAKE2S_BLOCKBYTES;
         S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
         blake2s_4way_compress( S, buf ); 
         S->buflen = 0;
-         input += ( bsize >> 2 );
-         inlen -= bsize;
+         input += ( BLAKE2S_BLOCKBYTES >> 2 );
+         inlen -= BLAKE2S_BLOCKBYTES;
      }
      else
      {
@@ -195,8 +222,45 @@ int blake2s_4way_final( blake2s_4way_state *S, void *out, uint8_t outlen )
   return 0;
 }

+// Update and final when inlen is a multiple of 64 bytes
+int blake2s_4way_full_blocks( blake2s_4way_state *S, void *out,
+                              const void *input, uint64_t inlen )
+{
+    __m128i *in = (__m128i*)input;
+    __m128i *buf = (__m128i*)S->buf;
+
+    while( inlen > BLAKE2S_BLOCKBYTES )
+    {
+       memcpy_128( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+       S->buflen = BLAKE2S_BLOCKBYTES;
+       inlen -= BLAKE2S_BLOCKBYTES;
+       S->t[0] += BLAKE2S_BLOCKBYTES;
+       S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
+       blake2s_4way_compress( S, buf );
+       S->buflen = 0;
+       in += ( BLAKE2S_BLOCKBYTES >> 2 );
+    }
+
+    // last block
+    memcpy_128( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+    S->buflen = BLAKE2S_BLOCKBYTES;
+    S->t[0] += S->buflen;
+    S->t[1] += ( S->t[0] < S->buflen );
+    if ( S->last_node )  S->f[1] = ~0U;
+    S->f[0] = ~0U;
+    blake2s_4way_compress( S, buf );
+
+    for ( int i = 0; i < 8; ++i )
+      casti_m128i( out, i ) = S->h[ i ];
+    return 0;
+}
+
 #if defined(__AVX2__)

+// The commented code below is slower on Intel but faster on
+// Zen1 AVX2. It's also faster than Zen1 AVX.
+// Ryzen gen2 is unknown at this time.
+ 
 int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
 {
   __m256i m[16];
@@ -205,6 +269,23 @@ int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
   memcpy_256( m, block, 16 );
   memcpy_256( v, S->h, 8 );

+   v[ 8] = m256_const1_64( 0x6A09E6676A09E667ULL );
+   v[ 9] = m256_const1_64( 0xBB67AE85BB67AE85ULL );
+   v[10] = m256_const1_64( 0x3C6EF3723C6EF372ULL );
+   v[11] = m256_const1_64( 0xA54FF53AA54FF53AULL );
+   v[12] = _mm256_xor_si256( _mm256_set1_epi32( S->t[0] ),
+                          m256_const1_64( 0x510E527F510E527FULL ) );
+
+   v[13] = _mm256_xor_si256( _mm256_set1_epi32( S->t[1] ),
+                          m256_const1_64( 0x9B05688C9B05688CULL ) );
+
+   v[14] = _mm256_xor_si256( _mm256_set1_epi32( S->f[0] ),
+                          m256_const1_64( 0x1F83D9AB1F83D9ABULL ) );
+
+   v[15] = _mm256_xor_si256( _mm256_set1_epi32( S->f[1] ),
+                          m256_const1_64( 0x5BE0CD195BE0CD19ULL ) );
+
+/*
   v[ 8] = _mm256_set1_epi32( blake2s_IV[0] );
   v[ 9] = _mm256_set1_epi32( blake2s_IV[1] );
   v[10] = _mm256_set1_epi32( blake2s_IV[2] );
@@ -218,6 +299,7 @@ int blake2s_8way_compress( blake2s_8way_state *S, const __m256i *block )
   v[15] = _mm256_xor_si256( _mm256_set1_epi32( S->f[1] ),
                             _mm256_set1_epi32( blake2s_IV[7] ) );

+
 #define G8W(r,i,a,b,c,d) \
 do { \
   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), \
@@ -231,7 +313,36 @@ do { \
   c = _mm256_add_epi32( c, d ); \
   b = mm256_ror_32( _mm256_xor_si256( b, c ),  7 ); \
 } while(0)
+*/

+#define G8W( sigma0, sigma1, a, b, c, d) \
+do { \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), m[ s0 ] ); \
+   d = mm256_ror_32( _mm256_xor_si256( d, a ), 16 ); \
+   c = _mm256_add_epi32( c, d ); \
+   b = mm256_ror_32( _mm256_xor_si256( b, c ), 12 ); \
+   a = _mm256_add_epi32( _mm256_add_epi32( a, b ), m[ s1 ] ); \
+   d = mm256_ror_32( _mm256_xor_si256( d, a ),  8 ); \
+   c = _mm256_add_epi32( c, d ); \
+   b = mm256_ror_32( _mm256_xor_si256( b, c ),  7 ); \
+} while(0)
+
+#define ROUND8W(r)  \
+do { \
+   uint8_t *sigma = (uint8_t*)&blake2s_sigma[r]; \
+   G8W( sigma[ 0], sigma[ 1], v[ 0], v[ 4], v[ 8], v[12] ); \
+   G8W( sigma[ 2], sigma[ 3], v[ 1], v[ 5], v[ 9], v[13] ); \
+   G8W( sigma[ 4], sigma[ 5], v[ 2], v[ 6], v[10], v[14] ); \
+   G8W( sigma[ 6], sigma[ 7], v[ 3], v[ 7], v[11], v[15] ); \
+   G8W( sigma[ 8], sigma[ 9], v[ 0], v[ 5], v[10], v[15] ); \
+   G8W( sigma[10], sigma[11], v[ 1], v[ 6], v[11], v[12] ); \
+   G8W( sigma[12], sigma[13], v[ 2], v[ 7], v[ 8], v[13] ); \
+   G8W( sigma[14], sigma[15], v[ 3], v[ 4], v[ 9], v[14] ); \
+} while(0)
+
+/*
 #define ROUND8W(r)  \
 do { \
   G8W( r, 0, v[ 0], v[ 4], v[ 8], v[12] ); \
@@ -243,6 +354,7 @@ do { \
   G8W( r, 6, v[ 2], v[ 7], v[ 8], v[13] ); \
   G8W( r, 7, v[ 3], v[ 4], v[ 9], v[14] ); \
 } while(0)
+*/

   ROUND8W( 0 );
   ROUND8W( 1 );
@@ -351,9 +463,203 @@ int blake2s_8way_final( blake2s_8way_state *S, void *out, uint8_t outlen )
   return 0;
 }

+// Update and final when inlen is a multiple of 64 bytes
+int blake2s_8way_full_blocks( blake2s_8way_state *S, void *out,
+                              const void *input, uint64_t inlen )
+{
+    __m256i *in = (__m256i*)input;
+    __m256i *buf = (__m256i*)S->buf;
+
+    while( inlen > BLAKE2S_BLOCKBYTES )
+    {
+       memcpy_256( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+       S->buflen = BLAKE2S_BLOCKBYTES;
+       inlen -= BLAKE2S_BLOCKBYTES;
+       S->t[0] += BLAKE2S_BLOCKBYTES;
+       S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
+       blake2s_8way_compress( S, buf );
+       S->buflen = 0;
+       in += ( BLAKE2S_BLOCKBYTES >> 2 );
+    }
+
+    // last block
+    memcpy_256( buf, in, BLAKE2S_BLOCKBYTES >> 2 );
+    S->buflen = BLAKE2S_BLOCKBYTES;
+    S->t[0] += S->buflen;
+    S->t[1] += ( S->t[0] < S->buflen );
+    if ( S->last_node )  S->f[1] = ~0U;
+    S->f[0] = ~0U;
+    blake2s_8way_compress( S, buf );
+
+    for ( int i = 0; i < 8; ++i )
+      casti_m256i( out, i ) = S->h[ i ];
+    return 0;
+}

 #endif // __AVX2__

+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+// Blake2s-256 16 way
+
+int blake2s_16way_compress( blake2s_16way_state *S, const __m512i *block )
+{
+   __m512i m[16];
+   __m512i v[16];
+
+   memcpy_512( m, block, 16 );
+   memcpy_512( v, S->h, 8 );
+
+   v[ 8] = m512_const1_64( 0x6A09E6676A09E667ULL );
+   v[ 9] = m512_const1_64( 0xBB67AE85BB67AE85ULL );
+   v[10] = m512_const1_64( 0x3C6EF3723C6EF372ULL );
+   v[11] = m512_const1_64( 0xA54FF53AA54FF53AULL );
+   v[12] = _mm512_xor_si512( _mm512_set1_epi32( S->t[0] ),
+                          m512_const1_64( 0x510E527F510E527FULL ) );
+
+   v[13] = _mm512_xor_si512( _mm512_set1_epi32( S->t[1] ),
+                          m512_const1_64( 0x9B05688C9B05688CULL ) );
+
+   v[14] = _mm512_xor_si512( _mm512_set1_epi32( S->f[0] ),
+                          m512_const1_64( 0x1F83D9AB1F83D9ABULL ) );
+
+   v[15] = _mm512_xor_si512( _mm512_set1_epi32( S->f[1] ),
+                          m512_const1_64( 0x5BE0CD195BE0CD19ULL ) );
+
+
+#define G16W( sigma0, sigma1, a, b, c, d) \
+do { \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm512_add_epi32( _mm512_add_epi32( a, b ), m[ s0 ] ); \
+   d = mm512_ror_32( _mm512_xor_si512( d, a ), 16 ); \
+   c = _mm512_add_epi32( c, d ); \
+   b = mm512_ror_32( _mm512_xor_si512( b, c ), 12 ); \
+   a = _mm512_add_epi32( _mm512_add_epi32( a, b ), m[ s1 ] ); \
+   d = mm512_ror_32( _mm512_xor_si512( d, a ),  8 ); \
+   c = _mm512_add_epi32( c, d ); \
+   b = mm512_ror_32( _mm512_xor_si512( b, c ),  7 ); \
+} while(0)
+
+#define ROUND16W(r)  \
+do { \
+   uint8_t *sigma = (uint8_t*)&blake2s_sigma[r]; \
+   G16W( sigma[ 0], sigma[ 1], v[ 0], v[ 4], v[ 8], v[12] ); \
+   G16W( sigma[ 2], sigma[ 3], v[ 1], v[ 5], v[ 9], v[13] ); \
+   G16W( sigma[ 4], sigma[ 5], v[ 2], v[ 6], v[10], v[14] ); \
+   G16W( sigma[ 6], sigma[ 7], v[ 3], v[ 7], v[11], v[15] ); \
+   G16W( sigma[ 8], sigma[ 9], v[ 0], v[ 5], v[10], v[15] ); \
+   G16W( sigma[10], sigma[11], v[ 1], v[ 6], v[11], v[12] ); \
+   G16W( sigma[12], sigma[13], v[ 2], v[ 7], v[ 8], v[13] ); \
+   G16W( sigma[14], sigma[15], v[ 3], v[ 4], v[ 9], v[14] ); \
+} while(0)
+
+   ROUND16W( 0 );
+   ROUND16W( 1 );
+   ROUND16W( 2 );
+   ROUND16W( 3 );
+   ROUND16W( 4 );
+   ROUND16W( 5 );
+   ROUND16W( 6 );
+   ROUND16W( 7 );
+   ROUND16W( 8 );
+   ROUND16W( 9 );
+
+   for( size_t i = 0; i < 8; ++i )
+      S->h[i] = _mm512_xor_si512( _mm512_xor_si512( S->h[i], v[i] ), v[i + 8] );
+
+#undef G16W
+#undef ROUND16W
+   return 0;
+}
+
+int blake2s_16way_init( blake2s_16way_state *S, const uint8_t outlen )
+{
+   blake2s_nway_param P[1];
+
+   P->digest_length = outlen;
+   P->key_length    = 0;
+   P->fanout        = 1;
+   P->depth         = 1;
+   P->leaf_length   = 0;
+   *((uint64_t*)(P->node_offset)) = 0;
+   P->node_depth    = 0;
+   P->inner_length  = 0;
+   memset( P->salt,     0, sizeof( P->salt ) );
+   memset( P->personal, 0, sizeof( P->personal ) );
+
+   memset( S, 0, sizeof( blake2s_16way_state ) );
+   S->h[0] = m512_const1_64( 0x6A09E6676A09E667ULL );
+   S->h[1] = m512_const1_64( 0xBB67AE85BB67AE85ULL );
+   S->h[2] = m512_const1_64( 0x3C6EF3723C6EF372ULL );
+   S->h[3] = m512_const1_64( 0xA54FF53AA54FF53AULL );
+   S->h[4] = m512_const1_64( 0x510E527F510E527FULL );
+   S->h[5] = m512_const1_64( 0x9B05688C9B05688CULL );
+   S->h[6] = m512_const1_64( 0x1F83D9AB1F83D9ABULL );
+   S->h[7] = m512_const1_64( 0x5BE0CD195BE0CD19ULL );
+
+   uint32_t *p = ( uint32_t * )( P );
+
+   /* IV XOR ParamBlock */
+   for ( size_t i = 0; i < 8; ++i )
+      S->h[i] = _mm512_xor_si512( S->h[i], _mm512_set1_epi32( p[i] ) );
+   return 0;
+}
+
+int blake2s_16way_update( blake2s_16way_state *S, const void *in,
+                         uint64_t inlen )
+{
+  __m512i *input = (__m512i*)in;
+  __m512i *buf = (__m512i*)S->buf;
+  const int bsize = BLAKE2S_BLOCKBYTES;
+
+   while( inlen > 0 )
+   {
+      size_t left = S->buflen;
+      if( inlen >= bsize - left )
+      {
+         memcpy_512( buf + (left>>2), input, (bsize - left) >> 2 );
+         S->buflen += bsize - left;
+         S->t[0] += BLAKE2S_BLOCKBYTES;
+         S->t[1] += ( S->t[0] < BLAKE2S_BLOCKBYTES );
+         blake2s_16way_compress( S, buf );
+         S->buflen = 0;
+         input += ( bsize >> 2 );
+         inlen -= bsize;
+      }
+      else
+      {
+          memcpy_512( buf + ( left>>2 ), input, inlen>>2 );
+          S->buflen += (size_t) inlen;
+          input += ( inlen>>2 );
+          inlen -= inlen;
+      }
+   }
+   return 0;
+}
+
+int blake2s_16way_final( blake2s_16way_state *S, void *out, uint8_t outlen )
+{
+   __m512i *buf = (__m512i*)S->buf;
+
+   S->t[0] += S->buflen;
+   S->t[1] += ( S->t[0] < S->buflen );
+   if ( S->last_node )
+      S->f[1] = ~0U;
+   S->f[0] = ~0U;
+
+   memset_zero_512( buf + ( S->buflen>>2 ),
+                    ( BLAKE2S_BLOCKBYTES - S->buflen ) >> 2 );
+   blake2s_16way_compress( S, buf );
+
+   for ( int i = 0; i < 8; ++i )
+      casti_m512i( out, i ) = S->h[ i ];
+   return 0;
+}
+
+#endif   // AVX512
+
+
 #if 0
 int blake2s( uint8_t *out, const void *in, const void *key, const uint8_t outlen, const uint64_t inlen, uint8_t keylen )
 {
--- a/algo/blake/blake2s-hash-4way.h
+++ b/algo/blake/blake2s-hash-4way.h
@@ -14,7 +14,6 @@
 #ifndef __BLAKE2S_HASH_4WAY_H__
 #define __BLAKE2S_HASH_4WAY_H__ 1

-//#if defined(__SSE4_2__)
 #if defined(__SSE2__)

 #include "simd-utils.h"
@@ -61,7 +60,7 @@ typedef struct __blake2s_nway_param
 } blake2s_nway_param;
 #pragma pack(pop)

-ALIGN( 64 ) typedef struct __blake2s_4way_state
+typedef struct ALIGN( 64 ) __blake2s_4way_state
 {
   __m128i h[8];
   uint8_t  buf[ BLAKE2S_BLOCKBYTES * 4 ];
@@ -75,10 +74,13 @@ int blake2s_4way_init( blake2s_4way_state *S, const uint8_t outlen );
 int blake2s_4way_update( blake2s_4way_state *S, const void *in,
                         uint64_t inlen );
 int blake2s_4way_final( blake2s_4way_state *S, void *out, uint8_t outlen );
+int blake2s_4way_full_blocks( blake2s_4way_state *S, void *out,
+                              const void *input, uint64_t inlen );
+

 #if defined(__AVX2__)

-ALIGN( 64 ) typedef struct __blake2s_8way_state
+typedef struct ALIGN( 64 ) __blake2s_8way_state
 {
   __m256i h[8];
   uint8_t  buf[ BLAKE2S_BLOCKBYTES * 8 ];
@@ -92,6 +94,27 @@ int blake2s_8way_init( blake2s_8way_state *S, const uint8_t outlen );
 int blake2s_8way_update( blake2s_8way_state *S, const void *in,
                         uint64_t inlen );
 int blake2s_8way_final( blake2s_8way_state *S, void *out, uint8_t outlen );
+int blake2s_8way_full_blocks( blake2s_8way_state *S, void *out,
+                              const void *input, uint64_t inlen );
+
+#endif
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+typedef struct ALIGN( 64 ) __blake2s_16way_state
+{
+   __m512i h[8];
+   uint8_t  buf[ BLAKE2S_BLOCKBYTES * 16 ];
+   uint32_t t[2];
+   uint32_t f[2];
+   size_t   buflen;
+   uint8_t  last_node;
+} blake2s_16way_state ;
+
+int blake2s_16way_init( blake2s_16way_state *S, const uint8_t outlen );
+int blake2s_16way_update( blake2s_16way_state *S, const void *in,
+                         uint64_t inlen );
+int blake2s_16way_final( blake2s_16way_state *S, void *out, uint8_t outlen );

 #endif

@@ -108,6 +131,6 @@ int blake2s_8way_final( blake2s_8way_state *S, void *out, uint8_t outlen );
 }
 #endif

-#endif  // __SSE4_2__
+#endif  // __SSE2__

 #endif
--- a/algo/blake/blake2s.c
+++ b/algo/blake/blake2s.c
@@ -1,5 +1,7 @@
 #include "blake2s-gate.h"

+#if  !defined(BLAKE2S_16WAY) && !defined(BLAKE2S_8WAY) && !defined(BLAKE2S)
+
 #include <string.h>
 #include <stdint.h>

@@ -56,7 +58,7 @@ int scanhash_blake2s( struct work *work,
 	do {
 		be32enc(&endiandata[19], n);
 		blake2s_hash( hash64, endiandata );
-		if (hash64[7] < Htarg && fulltest(hash64, ptarget)) {
+		if (hash64[7] <= Htarg && fulltest(hash64, ptarget)) {
 			*hashes_done = n - first_nonce + 1;
 			pdata[19] = n;
 			return true;
@@ -70,18 +72,4 @@ int scanhash_blake2s( struct work *work,

 	return 0;
 }
-/*
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blake2s_get_max64 ()
-{
-   return 0x7ffffLL;
-}
-
-bool register_blake2s_algo( algo_gate_t* gate )
-{
-  gate->scanhash  = (void*)&scanhash_blake2s;
-  gate->hash      = (void*)&blake2s_hash;
-  gate->get_max64 = (void*)&blake2s_get_max64;
-  return true;
-};
-*/
+#endif
--- a/algo/blake/blake512-hash-4way.c
+++ b/algo/blake/blake512-hash-4way.c
--- a/algo/blake/blakecoin-4way.c
+++ b/algo/blake/blakecoin-4way.c
@@ -14,7 +14,7 @@ void blakecoin_4way_hash(void *state, const void *input)
     blake256r8_4way_context ctx;

     memcpy( &ctx, &blakecoin_4w_ctx, sizeof ctx );
-     blake256r8_4way( &ctx, input + (64<<2), 16 );
+     blake256r8_4way_update( &ctx, input + (64<<2), 16 );
     blake256r8_4way_close( &ctx, vhash );

     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
@@ -37,7 +37,7 @@ int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,

   mm128_bswap32_intrlv80_4x32( vdata, pdata );
   blake256r8_4way_init( &blakecoin_4w_ctx );
-   blake256r8_4way( &blakecoin_4w_ctx, vdata, 64 );
+   blake256r8_4way_update( &blakecoin_4w_ctx, vdata, 64 );

   do {
      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
@@ -49,7 +49,7 @@ int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,
           && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;

@@ -71,7 +71,7 @@ void blakecoin_8way_hash( void *state, const void *input )
     blake256r8_8way_context ctx;

     memcpy( &ctx, &blakecoin_8w_ctx, sizeof ctx );
-     blake256r8_8way( &ctx, input + (64<<3), 16 );
+     blake256r8_8way_update( &ctx, input + (64<<3), 16 );
     blake256r8_8way_close( &ctx, vhash );

     dintrlv_8x32( state,     state+ 32, state+ 64, state+ 96, state+128,
@@ -95,7 +95,7 @@ int scanhash_blakecoin_8way( struct work *work, uint32_t max_nonce,

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
   blake256r8_8way_init( &blakecoin_8w_ctx );
-   blake256r8_8way( &blakecoin_8w_ctx, vdata, 64 );
+   blake256r8_8way_update( &blakecoin_8w_ctx, vdata, 64 );

   do {
      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
@@ -108,7 +108,7 @@ int scanhash_blakecoin_8way( struct work *work, uint32_t max_nonce,
          && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 8;
   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
--- a/algo/blake/blakecoin-gate.c
+++ b/algo/blake/blakecoin-gate.c
@@ -1,13 +1,6 @@
 #include "blakecoin-gate.h"
 #include <memory.h>

-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blakecoin_get_max64 ()
-{
-  return 0x7ffffLL;
-//  return 0x3fffffLL;
-}
-
 // vanilla uses default gen merkle root, otherwise identical to blakecoin
 bool register_vanilla_algo( algo_gate_t* gate )
 {
@@ -23,7 +16,6 @@ bool register_vanilla_algo( algo_gate_t* gate )
  gate->hash     = (void*)&blakecoinhash;
 #endif
  gate->optimizations = SSE42_OPT | AVX2_OPT;
-  gate->get_max64 = (void*)&blakecoin_get_max64;
  return true;
 }

--- a/algo/blake/blakecoin.c
+++ b/algo/blake/blakecoin.c
@@ -1,4 +1,7 @@
 #include "blakecoin-gate.h"
+
+#if !defined(BLAKECOIN_8WAY) && !defined(BLAKECOIN_4WAY)
+
 #define BLAKE32_ROUNDS 8
 #include "sph_blake.h"

@@ -93,33 +96,4 @@ int scanhash_blakecoin( struct work *work, uint32_t max_nonce,
 	return 0;
 }

-/*
-void blakecoin_gen_merkle_root ( char* merkle_root, struct stratum_ctx* sctx )
-{
- SHA256( sctx->job.coinbase, (int)sctx->job.coinbase_size, merkle_root );
-}
-*/
-/*
-// changed to get_max64_0x3fffffLL in cpuminer-multi-decred
-int64_t blakecoin_get_max64 ()
-{
-  return 0x7ffffLL;
-}
-
-// vanilla uses default gen merkle root, otherwise identical to blakecoin
-bool register_vanilla_algo( algo_gate_t* gate )
-{
-    gate->scanhash = (void*)&scanhash_blakecoin;
-    gate->hash     = (void*)&blakecoinhash;
-    gate->get_max64 = (void*)&blakecoin_get_max64;
-    blakecoin_init( &blake_init_ctx );
-    return true;
-}
-
-bool register_blakecoin_algo( algo_gate_t* gate )
-{
-  register_vanilla_algo( gate );
-  gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
-  return true;
-}
-*/
+#endif
--- a/algo/blake/decred-4way.c
+++ b/algo/blake/decred-4way.c
@@ -21,7 +21,7 @@ void decred_hash_4way( void *state, const void *input )
     blake256_4way_context ctx __attribute__ ((aligned (64)));

     memcpy( &ctx, &blake_mid, sizeof(blake_mid) );
-     blake256_4way( &ctx, tail, tail_len );
+     blake256_4way_update( &ctx, tail, tail_len );
     blake256_4way_close( &ctx, vhash );
     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
 }
@@ -46,7 +46,7 @@ int scanhash_decred_4way( struct work *work, uint32_t max_nonce,
   mm128_intrlv_4x32x( vdata, edata, edata, edata, edata, 180*8 );

   blake256_4way_init( &blake_mid );
-   blake256_4way( &blake_mid, vdata, DECRED_MIDSTATE_LEN );
+   blake256_4way_update( &blake_mid, vdata, DECRED_MIDSTATE_LEN );

   uint32_t *noncep = vdata + DECRED_NONCE_INDEX * 4;
   do {
@@ -62,7 +62,7 @@ int scanhash_decred_4way( struct work *work, uint32_t max_nonce,
      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
      {
          pdata[DECRED_NONCE_INDEX] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;
  } while ( (n < max_nonce) && !work_restart[thr_id].restart );
--- a/algo/blake/decred-gate.c
+++ b/algo/blake/decred-gate.c
@@ -38,7 +38,7 @@ void decred_decode_extradata( struct work* work, uint64_t* net_blocks )
   if (!have_longpoll && work->height > *net_blocks + 1)
   {
      char netinfo[64] = { 0 };
-      if (opt_showdiff && net_diff > 0.)
+      if ( net_diff > 0. )
      {
         if (net_diff != work->targetdiff)
            sprintf(netinfo, ", diff %.3f, target %.1f", net_diff,
@@ -78,7 +78,6 @@ void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
   uint32_t extraheader[32] = { 0 };
   int headersize = 0;
   uint32_t* extradata = (uint32_t*) sctx->xnonce1;
-   size_t t;
   int i;

   // getwork over stratum, getwork merkle + header passed in coinb1
@@ -87,9 +86,6 @@ void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
                  sizeof(extraheader) );
   memcpy( extraheader, &sctx->job.coinbase[32], headersize );

-   // Increment extranonce2 
-   for ( t = 0; t < sctx->xnonce2_size && !( ++sctx->job.xnonce2[t] ); t++ );
-
   // Assemble block header 
   memset( g_work->data, 0, sizeof(g_work->data) );
   g_work->data[0] = le32dec( sctx->job.version );
@@ -116,7 +112,7 @@ void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
   // block header suffix from coinb2 (stake version)
   memcpy( &g_work->data[44],
           &sctx->job.coinbase[ sctx->job.coinbase_size-4 ], 4 );
-   sctx->bloc_height = g_work->data[32];
+   sctx->block_height = g_work->data[32];
   //applog_hex(work->data, 180);
   //applog_hex(&work->data[36], 36);
 }
@@ -153,8 +149,7 @@ bool register_decred_algo( algo_gate_t* gate )
  gate->hash      = (void*)&decred_hash;
 #endif
  gate->optimizations = AVX2_OPT;
-  gate->get_nonceptr          = (void*)&decred_get_nonceptr;
-  gate->get_max64             = (void*)&get_max64_0x3fffffLL;
+//  gate->get_nonceptr          = (void*)&decred_get_nonceptr;
  gate->decode_extra_data     = (void*)&decred_decode_extradata;
  gate->build_stratum_request = (void*)&decred_be_build_stratum_request;
  gate->work_decode           = (void*)&std_be_work_decode;
--- a/algo/blake/decred.c
+++ b/algo/blake/decred.c
@@ -1,4 +1,7 @@
 #include "decred-gate.h"
+
+#if !defined(DECRED_8WAY) && !defined(DECRED_4WAY)
+
 #include "sph_blake.h"

 #include <string.h>
@@ -77,25 +80,15 @@ int scanhash_decred( struct work *work, uint32_t max_nonce,
                be32enc(&endiandata[k], pdata[k]);
 #endif

-#ifdef DEBUG_ALGO
-        if (!thr_id) applog(LOG_DEBUG,"[%d] Target=%08x %08x", thr_id, ptarget[6], ptarget[7]);
-#endif
-
        do {
                //be32enc(&endiandata[DCR_NONCE_OFT32], n);
                endiandata[DECRED_NONCE_INDEX] = n;
                decred_hash(hash32, endiandata);

-                if (hash32[7] <= HTarget && fulltest(hash32, ptarget)) {
-                        work_set_target_ratio(work, hash32);
-                        *hashes_done = n - first_nonce + 1;
-#ifdef DEBUG_ALGO
-                        applog(LOG_BLUE, "Nonce : %08x %08x", n, swab32(n));
-                        applog_hash(ptarget);
-                        applog_compare_hash(hash32, ptarget);
-#endif
-                        pdata[DECRED_NONCE_INDEX] = n;
-                        return 1;
+                if (hash32[7] <= HTarget && fulltest(hash32, ptarget))
+                {
+                   pdata[DECRED_NONCE_INDEX] = n;
+                   submit_solution( work, hash32, mythr );
                }

                n++;
@@ -143,7 +136,7 @@ void decred_decode_extradata( struct work* work, uint64_t* net_blocks )
   if (!have_longpoll && work->height > *net_blocks + 1)
   {
      char netinfo[64] = { 0 };
-      if (opt_showdiff && net_diff > 0.)
+      if (net_diff > 0.)
      {
         if (net_diff != work->targetdiff)
 	    sprintf(netinfo, ", diff %.3f, target %.1f", net_diff,
@@ -269,7 +262,6 @@ bool register_decred_algo( algo_gate_t* gate )
  gate->scanhash              = (void*)&scanhash_decred;
  gate->hash                  = (void*)&decred_hash;
  gate->get_nonceptr          = (void*)&decred_get_nonceptr;
-  gate->get_max64             = (void*)&get_max64_0x3fffffLL;
  gate->decode_extra_data     = (void*)&decred_decode_extradata;
  gate->build_stratum_request = (void*)&decred_be_build_stratum_request;
  gate->work_decode           = (void*)&std_be_work_decode;
@@ -286,3 +278,5 @@ bool register_decred_algo( algo_gate_t* gate )
  return true;
 }
 */
+
+#endif
--- a/algo/blake/pentablake-4way.c
+++ b/algo/blake/pentablake-4way.c
@@ -22,23 +22,23 @@ extern void pentablakehash_4way( void *output, const void *input )


     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, input, 80 );
+     blake512_4way_update( &ctx, input, 80 );
     blake512_4way_close( &ctx, vhash );

     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, vhash, 64 );
+     blake512_4way_update( &ctx, vhash, 64 );
     blake512_4way_close( &ctx, vhash );

     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, vhash, 64 );
+     blake512_4way_update( &ctx, vhash, 64 );
     blake512_4way_close( &ctx, vhash );

     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, vhash, 64 );
+     blake512_4way_update( &ctx, vhash, 64 );
     blake512_4way_close( &ctx, vhash );

     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, vhash, 64 );
+     blake512_4way_update( &ctx, vhash, 64 );
     blake512_4way_close( &ctx, vhash );

     memcpy( output,    hash0, 32 );
@@ -105,7 +105,7 @@ int scanhash_pentablake_4way( struct work *work,
                  && fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
              {
                 pdata[19] = n + i;
-                 submit_lane_solution( work, hash+(i<<3), mythr, i );
+                 submit_solution( work, hash+(i<<3), mythr );
              }
              n += 4;

--- a/algo/blake/pentablake-gate.c
+++ b/algo/blake/pentablake-gate.c
@@ -10,7 +10,6 @@ bool register_pentablake_algo( algo_gate_t* gate )
    gate->hash      = (void*)&pentablakehash;
 #endif
    gate->optimizations = AVX2_OPT;
-    gate->get_max64 = (void*)&get_max64_0x3ffff;
    return true;
 };

--- a/algo/blake/pentablake.c
+++ b/algo/blake/pentablake.c
@@ -1,4 +1,7 @@
 #include "pentablake-gate.h"
+
+#if !defined(PENTABLAKE_8WAY) && !defined(PENTABLAKE_4WAY)
+
 #include <stdlib.h>
 #include <stdint.h>
 #include <string.h>
@@ -111,3 +114,4 @@ int scanhash_pentablake( struct work *work, uint32_t max_nonce,
 	return 0;
 } 

+#endif
--- a/algo/blake/sph-blake2s.c
+++ b/algo/blake/sph-blake2s.c
@@ -323,7 +323,7 @@ int blake2s_final( blake2s_state *S, uint8_t *out, uint8_t outlen )

 int blake2s( uint8_t *out, const void *in, const void *key, const uint8_t outlen, const uint64_t inlen, uint8_t keylen )
 {
-	blake2s_state S[1];
+	blake2s_state S;

 	/* Verify parameters */
 	if ( NULL == in ) return -1;
@@ -334,15 +334,15 @@ int blake2s( uint8_t *out, const void *in, const void *key, const uint8_t outlen

 	if( keylen > 0 )
 	{
-		if( blake2s_init_key( S, outlen, key, keylen ) < 0 ) return -1;
+		if( blake2s_init_key( &S, outlen, key, keylen ) < 0 ) return -1;
 	}
 	else
 	{
-		if( blake2s_init( S, outlen ) < 0 ) return -1;
+		if( blake2s_init( &S, outlen ) < 0 ) return -1;
 	}

-	blake2s_update( S, ( uint8_t * )in, inlen );
-	blake2s_final( S, out, outlen );
+	blake2s_update( &S, ( uint8_t * )in, inlen );
+	blake2s_final( &S, out, outlen );
 	return 0;
 }

--- a/algo/blake/sph-blake2s.h
+++ b/algo/blake/sph-blake2s.h
@@ -116,7 +116,7 @@ extern "C" {
 		uint8_t  personal[BLAKE2S_PERSONALBYTES];  // 32
 	} blake2s_param;

-	ALIGN( 64 ) typedef struct __blake2s_state
+	typedef struct ALIGN( 64 ) __blake2s_state
 	{
 		uint32_t h[8];
 		uint32_t t[2];
--- a/algo/blake/sph_blake2b.h
+++ b/algo/blake/sph_blake2b.h
@@ -18,7 +18,7 @@
 #endif

 // state context
-ALIGN(64) typedef struct {
+typedef ALIGN(64) struct {
 	uint8_t b[128]; // input buffer
 	uint64_t h[8];  // chained state
 	uint64_t t[2];  // total number of bytes
--- a/algo/blake/sse2/blake.c
+++ b/algo/blake/sse2/blake.c
@@ -1,476 +0,0 @@
-/* $Id: blake.c 252 2011-06-07 17:55:14Z tp $ */
-/*
- * BLAKE implementation.
- *
- * ==========================(LICENSE BEGIN)============================
- *
- * Copyright (c) 2007-2010  Projet RNRT SAPHIR
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
- * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * ===========================(LICENSE END)=============================
- *
- * @author   Thomas Pornin <thomas.pornin@cryptolog.com>
- */
-#include <stddef.h>
-#include <string.h>
-#include <limits.h>
-
-#include "../sph_blake.h"
-
-#ifdef __cplusplus
-extern "C"{
-#endif
-
-#ifdef _MSC_VER
-#pragma warning (disable: 4146)
-#endif
-
-static const sph_u64 blkIV512[8] = {
-	SPH_C64(0x6A09E667F3BCC908), SPH_C64(0xBB67AE8584CAA73B),
-	SPH_C64(0x3C6EF372FE94F82B), SPH_C64(0xA54FF53A5F1D36F1),
-	SPH_C64(0x510E527FADE682D1), SPH_C64(0x9B05688C2B3E6C1F),
-	SPH_C64(0x1F83D9ABFB41BD6B), SPH_C64(0x5BE0CD19137E2179)
-};
-
-#define Z00   0
-#define Z01   1
-#define Z02   2
-#define Z03   3
-#define Z04   4
-#define Z05   5
-#define Z06   6
-#define Z07   7
-#define Z08   8
-#define Z09   9
-#define Z0A   A
-#define Z0B   B
-#define Z0C   C
-#define Z0D   D
-#define Z0E   E
-#define Z0F   F
-
-#define Z10   E
-#define Z11   A
-#define Z12   4
-#define Z13   8
-#define Z14   9
-#define Z15   F
-#define Z16   D
-#define Z17   6
-#define Z18   1
-#define Z19   C
-#define Z1A   0
-#define Z1B   2
-#define Z1C   B
-#define Z1D   7
-#define Z1E   5
-#define Z1F   3
-
-#define Z20   B
-#define Z21   8
-#define Z22   C
-#define Z23   0
-#define Z24   5
-#define Z25   2
-#define Z26   F
-#define Z27   D
-#define Z28   A
-#define Z29   E
-#define Z2A   3
-#define Z2B   6
-#define Z2C   7
-#define Z2D   1
-#define Z2E   9
-#define Z2F   4
-
-#define Z30   7
-#define Z31   9
-#define Z32   3
-#define Z33   1
-#define Z34   D
-#define Z35   C
-#define Z36   B
-#define Z37   E
-#define Z38   2
-#define Z39   6
-#define Z3A   5
-#define Z3B   A
-#define Z3C   4
-#define Z3D   0
-#define Z3E   F
-#define Z3F   8
-
-#define Z40   9
-#define Z41   0
-#define Z42   5
-#define Z43   7
-#define Z44   2
-#define Z45   4
-#define Z46   A
-#define Z47   F
-#define Z48   E
-#define Z49   1
-#define Z4A   B
-#define Z4B   C
-#define Z4C   6
-#define Z4D   8
-#define Z4E   3
-#define Z4F   D
-
-#define Z50   2
-#define Z51   C
-#define Z52   6
-#define Z53   A
-#define Z54   0
-#define Z55   B
-#define Z56   8
-#define Z57   3
-#define Z58   4
-#define Z59   D
-#define Z5A   7
-#define Z5B   5
-#define Z5C   F
-#define Z5D   E
-#define Z5E   1
-#define Z5F   9
-
-#define Z60   C
-#define Z61   5
-#define Z62   1
-#define Z63   F
-#define Z64   E
-#define Z65   D
-#define Z66   4
-#define Z67   A
-#define Z68   0
-#define Z69   7
-#define Z6A   6
-#define Z6B   3
-#define Z6C   9
-#define Z6D   2
-#define Z6E   8
-#define Z6F   B
-
-#define Z70   D
-#define Z71   B
-#define Z72   7
-#define Z73   E
-#define Z74   C
-#define Z75   1
-#define Z76   3
-#define Z77   9
-#define Z78   5
-#define Z79   0
-#define Z7A   F
-#define Z7B   4
-#define Z7C   8
-#define Z7D   6
-#define Z7E   2
-#define Z7F   A
-
-#define Z80   6
-#define Z81   F
-#define Z82   E
-#define Z83   9
-#define Z84   B
-#define Z85   3
-#define Z86   0
-#define Z87   8
-#define Z88   C
-#define Z89   2
-#define Z8A   D
-#define Z8B   7
-#define Z8C   1
-#define Z8D   4
-#define Z8E   A
-#define Z8F   5
-
-#define Z90   A
-#define Z91   2
-#define Z92   8
-#define Z93   4
-#define Z94   7
-#define Z95   6
-#define Z96   1
-#define Z97   5
-#define Z98   F
-#define Z99   B
-#define Z9A   9
-#define Z9B   E
-#define Z9C   3
-#define Z9D   C
-#define Z9E   D
-#define Z9F   0
-
-#define Mx(r, i)    Mx_(Z ## r ## i)
-#define Mx_(n)      Mx__(n)
-#define Mx__(n)     M ## n
-
-#define CSx(r, i)   CSx_(Z ## r ## i)
-#define CSx_(n)     CSx__(n)
-#define CSx__(n)    CS ## n
-
-#define CS0   SPH_C32(0x243F6A88)
-#define CS1   SPH_C32(0x85A308D3)
-#define CS2   SPH_C32(0x13198A2E)
-#define CS3   SPH_C32(0x03707344)
-#define CS4   SPH_C32(0xA4093822)
-#define CS5   SPH_C32(0x299F31D0)
-#define CS6   SPH_C32(0x082EFA98)
-#define CS7   SPH_C32(0xEC4E6C89)
-#define CS8   SPH_C32(0x452821E6)
-#define CS9   SPH_C32(0x38D01377)
-#define CSA   SPH_C32(0xBE5466CF)
-#define CSB   SPH_C32(0x34E90C6C)
-#define CSC   SPH_C32(0xC0AC29B7)
-#define CSD   SPH_C32(0xC97C50DD)
-#define CSE   SPH_C32(0x3F84D5B5)
-#define CSF   SPH_C32(0xB5470917)
-
-
-
-#define CBx(r, i)   CBx_(Z ## r ## i)
-#define CBx_(n)     CBx__(n)
-#define CBx__(n)    CB ## n
-
-#define CB0   SPH_C64(0x243F6A8885A308D3)
-#define CB1   SPH_C64(0x13198A2E03707344)
-#define CB2   SPH_C64(0xA4093822299F31D0)
-#define CB3   SPH_C64(0x082EFA98EC4E6C89)
-#define CB4   SPH_C64(0x452821E638D01377)
-#define CB5   SPH_C64(0xBE5466CF34E90C6C)
-#define CB6   SPH_C64(0xC0AC29B7C97C50DD)
-#define CB7   SPH_C64(0x3F84D5B5B5470917)
-#define CB8   SPH_C64(0x9216D5D98979FB1B)
-#define CB9   SPH_C64(0xD1310BA698DFB5AC)
-#define CBA   SPH_C64(0x2FFD72DBD01ADFB7)
-#define CBB   SPH_C64(0xB8E1AFED6A267E96)
-#define CBC   SPH_C64(0xBA7C9045F12C7F99)
-#define CBD   SPH_C64(0x24A19947B3916CF7)
-#define CBE   SPH_C64(0x0801F2E2858EFC16)
-#define CBF   SPH_C64(0x636920D871574E69)
-
-
-#define GS(m0, m1, c0, c1, a, b, c, d)   do { \
-		a = SPH_T32(a + b + (m0 ^ c1)); \
-		d = SPH_ROTR32(d ^ a, 16); \
-		c = SPH_T32(c + d); \
-		b = SPH_ROTR32(b ^ c, 12); \
-		a = SPH_T32(a + b + (m1 ^ c0)); \
-		d = SPH_ROTR32(d ^ a, 8); \
-		c = SPH_T32(c + d); \
-		b = SPH_ROTR32(b ^ c, 7); \
-	} while (0)
-
-#define ROUND_S(r)   do { \
-		GS(Mx(r, 0), Mx(r, 1), CSx(r, 0), CSx(r, 1), V0, V4, V8, VC); \
-		GS(Mx(r, 2), Mx(r, 3), CSx(r, 2), CSx(r, 3), V1, V5, V9, VD); \
-		GS(Mx(r, 4), Mx(r, 5), CSx(r, 4), CSx(r, 5), V2, V6, VA, VE); \
-		GS(Mx(r, 6), Mx(r, 7), CSx(r, 6), CSx(r, 7), V3, V7, VB, VF); \
-		GS(Mx(r, 8), Mx(r, 9), CSx(r, 8), CSx(r, 9), V0, V5, VA, VF); \
-		GS(Mx(r, A), Mx(r, B), CSx(r, A), CSx(r, B), V1, V6, VB, VC); \
-		GS(Mx(r, C), Mx(r, D), CSx(r, C), CSx(r, D), V2, V7, V8, VD); \
-		GS(Mx(r, E), Mx(r, F), CSx(r, E), CSx(r, F), V3, V4, V9, VE); \
-	} while (0)
-
-
-
-#define GB(m0, m1, c0, c1, a, b, c, d)   do { \
-		a = SPH_T64(a + b + (m0 ^ c1)); \
-		d = SPH_ROTR64(d ^ a, 32); \
-		c = SPH_T64(c + d); \
-		b = SPH_ROTR64(b ^ c, 25); \
-		a = SPH_T64(a + b + (m1 ^ c0)); \
-		d = SPH_ROTR64(d ^ a, 16); \
-		c = SPH_T64(c + d); \
-		b = SPH_ROTR64(b ^ c, 11); \
-	} while (0)
-
-#define ROUND_B(r)   do { \
-		GB(Mx(r, 0), Mx(r, 1), CBx(r, 0), CBx(r, 1), V0, V4, V8, VC); \
-		GB(Mx(r, 2), Mx(r, 3), CBx(r, 2), CBx(r, 3), V1, V5, V9, VD); \
-		GB(Mx(r, 4), Mx(r, 5), CBx(r, 4), CBx(r, 5), V2, V6, VA, VE); \
-		GB(Mx(r, 6), Mx(r, 7), CBx(r, 6), CBx(r, 7), V3, V7, VB, VF); \
-		GB(Mx(r, 8), Mx(r, 9), CBx(r, 8), CBx(r, 9), V0, V5, VA, VF); \
-		GB(Mx(r, A), Mx(r, B), CBx(r, A), CBx(r, B), V1, V6, VB, VC); \
-		GB(Mx(r, C), Mx(r, D), CBx(r, C), CBx(r, D), V2, V7, V8, VD); \
-		GB(Mx(r, E), Mx(r, F), CBx(r, E), CBx(r, F), V3, V4, V9, VE); \
-	} while (0)
-
-
-#define COMPRESS64   do { \
-                int b=0; \
-		sph_u64 M0, M1, M2, M3, M4, M5, M6, M7; \
-		sph_u64 M8, M9, MA, MB, MC, MD, ME, MF; \
-		sph_u64 V0, V1, V2, V3, V4, V5, V6, V7; \
-		sph_u64 V8, V9, VA, VB, VC, VD, VE, VF; \
-		V0 = blkH0, \
-		V1 = blkH1, \
-		V2 = blkH2, \
-		V3 = blkH3, \
-		V4 = blkH4, \
-		V5 = blkH5, \
-		V6 = blkH6, \
-		V7 = blkH7; \
-		V8 = blkS0 ^ CB0, \
-		V9 = blkS1 ^ CB1, \
-		VA = blkS2 ^ CB2, \
-		VB = blkS3 ^ CB3, \
-		VC = hashctA ^ CB4, \
-		VD = hashctA ^ CB5, \
-		VE = hashctB ^ CB6, \
-		VF = hashctB ^ CB7; \
-		M0 = sph_dec64be_aligned(buf +   0), \
-		M1 = sph_dec64be_aligned(buf +   8), \
-		M2 = sph_dec64be_aligned(buf +  16), \
-		M3 = sph_dec64be_aligned(buf +  24), \
-		M4 = sph_dec64be_aligned(buf +  32), \
-		M5 = sph_dec64be_aligned(buf +  40), \
-		M6 = sph_dec64be_aligned(buf +  48), \
-		M7 = sph_dec64be_aligned(buf +  56), \
-		M8 = sph_dec64be_aligned(buf +  64), \
-		M9 = sph_dec64be_aligned(buf +  72), \
-		MA = sph_dec64be_aligned(buf +  80), \
-		MB = sph_dec64be_aligned(buf +  88), \
-		MC = sph_dec64be_aligned(buf +  96), \
-		MD = sph_dec64be_aligned(buf + 104), \
-		ME = sph_dec64be_aligned(buf + 112), \
-		MF = sph_dec64be_aligned(buf + 120); \
-                /* loop once and a half */ \
-                /* save some space */ \
-                for (;;) { \
-		    ROUND_B(0); \
-		    ROUND_B(1); \
-		    ROUND_B(2); \
-		    ROUND_B(3); \
-		    ROUND_B(4); \
-		    ROUND_B(5); \
-                    if (b)  break; \
-                    b = 1; \
-		    ROUND_B(6); \
-		    ROUND_B(7); \
-		    ROUND_B(8); \
-		    ROUND_B(9); \
-                }; \
-		blkH0 ^= blkS0 ^ V0 ^ V8, \
-		blkH1 ^= blkS1 ^ V1 ^ V9, \
-		blkH2 ^= blkS2 ^ V2 ^ VA, \
-		blkH3 ^= blkS3 ^ V3 ^ VB, \
-		blkH4 ^= blkS0 ^ V4 ^ VC, \
-		blkH5 ^= blkS1 ^ V5 ^ VD, \
-		blkH6 ^= blkS2 ^ V6 ^ VE, \
-		blkH7 ^= blkS3 ^ V7 ^ VF; \
-	} while (0)
-/*
-*/
-#define DECL_BLK \
-	sph_u64 blkH0; \
-	sph_u64 blkH1; \
-	sph_u64 blkH2; \
-	sph_u64 blkH3; \
-	sph_u64 blkH4; \
-	sph_u64 blkH5; \
-	sph_u64 blkH6; \
-	sph_u64 blkH7; \
-	sph_u64 blkS0; \
-	sph_u64 blkS1; \
-	sph_u64 blkS2; \
-	sph_u64 blkS3; \
-
-/* load initial constants */
-#define BLK_I \
-do { \
-    blkH0 = SPH_C64(0x6A09E667F3BCC908); \
-    blkH1 = SPH_C64(0xBB67AE8584CAA73B); \
-    blkH2 = SPH_C64(0x3C6EF372FE94F82B); \
-    blkH3 = SPH_C64(0xA54FF53A5F1D36F1); \
-    blkH4 = SPH_C64(0x510E527FADE682D1); \
-    blkH5 = SPH_C64(0x9B05688C2B3E6C1F); \
-    blkH6 = SPH_C64(0x1F83D9ABFB41BD6B); \
-    blkH7 = SPH_C64(0x5BE0CD19137E2179); \
-    blkS0 = 0; \
-    blkS1 = 0; \
-    blkS2 = 0; \
-    blkS3 = 0; \
-    hashctB = SPH_T64(0- 1); \
-} while (0)
-
-/* copy in 80 for initial hash */
-#define BLK_W \
-do { \
-    memcpy(hashbuf, input, 80); \
-    hashctA = SPH_C64(0xFFFFFFFFFFFFFC00) + 80*8; \
-    hashptr = 80; \
-} while (0)
-
-/* copy in 64 for looped hash */
-#define BLK_U \
-do { \
-    memcpy(hashbuf, hash , 64); \
-    hashctA = SPH_C64(0xFFFFFFFFFFFFFC00) + 64*8; \
-    hashptr = 64; \
-} while (0)
-
-/* blake compress function */
-/* hash = blake512(loaded) */
-#define BLK_C \
-do { \
-    \
-    union { \
-        unsigned char buf[128]; \
-        sph_u64 dummy; \
-    } u; \
-    size_t ptr; \
-    unsigned bit_len; \
- \
-    ptr = hashptr; \
-    bit_len = ((unsigned)ptr << 3) + 0; \
-    u.buf[ptr] = ((0 & -(0x80)) | (0x80)) & 0xFF; \
-    memset(u.buf + ptr + 1, 0, 111 - ptr); \
-    u.buf[111] |= 1; \
-    sph_enc64be_aligned(u.buf + 112, 0); \
-    sph_enc64be_aligned(u.buf + 120, bit_len); \
-    do { \
-    const void *data = u.buf + ptr; \
-    unsigned char *buf; \
-    buf = hashbuf; \
-    size_t clen; \
-    clen = (sizeof(char)*128) - hashptr; \
-    memcpy(buf + hashptr, data, clen); \
-    hashctA = SPH_T64(hashctA + 1024); \
-    hashctB = SPH_T64(hashctB + 1); \
-    COMPRESS64; \
-    } while (0); \
-    /* end blake64(sc, u.buf + ptr, 128 - ptr); */ \
-    sph_enc64be((unsigned char*)(hash) + (0 << 3), blkH0), \
-    sph_enc64be((unsigned char*)(hash) + (1 << 3), blkH1); \
-    sph_enc64be((unsigned char*)(hash) + (2 << 3), blkH2), \
-    sph_enc64be((unsigned char*)(hash) + (3 << 3), blkH3); \
-    sph_enc64be((unsigned char*)(hash) + (4 << 3), blkH4), \
-    sph_enc64be((unsigned char*)(hash) + (5 << 3), blkH5); \
-    sph_enc64be((unsigned char*)(hash) + (6 << 3), blkH6), \
-    sph_enc64be((unsigned char*)(hash) + (7 << 3), blkH7); \
-} while (0) 
-
-
-#ifdef __cplusplus
-}
-#endif
--- a/algo/blake/sse2/blake/sse41/api.h
+++ b/algo/blake/sse2/blake/sse41/api.h
@@ -1,2 +0,0 @@
-#define CRYPTO_BYTES 64
-
--- a/algo/blake/sse2/blake/sse41/architectures
+++ b/algo/blake/sse2/blake/sse41/architectures
@@ -1,2 +0,0 @@
-amd64
-x86
--- a/algo/blake/sse2/blake/sse41/config.h
+++ b/algo/blake/sse2/blake/sse41/config.h
@@ -1,8 +0,0 @@
-#ifndef __BLAKE512_CONFIG_H__
-#define __BLAKE512_CONFIG_H__
-
-#define AVOID_BRANCHING 1
-//#define HAVE_XOP 1
-
-#endif
-
--- a/algo/blake/sse2/blake/sse41/hash.c
+++ b/algo/blake/sse2/blake/sse41/hash.c
@@ -1,287 +0,0 @@
-
-#include "hash.h"
-/*
-#ifndef NOT_SUPERCOP
-
-#include "crypto_hash.h"
-#include "crypto_uint64.h"
-#include "crypto_uint32.h"
-#include "crypto_uint8.h"
-
-typedef crypto_uint64 u64;
-typedef crypto_uint32 u32;
-typedef crypto_uint8 u8; 
-
-#else
-
-typedef unsigned long long u64; 
-typedef unsigned int u32; 
-typedef unsigned char u8; 
-
-#endif
-*/
-#define U8TO32(p) \
-  (((u32)((p)[0]) << 24) | ((u32)((p)[1]) << 16) | \
-   ((u32)((p)[2]) <<  8) | ((u32)((p)[3])      ))
-#define U8TO64(p) \
-  (((u64)U8TO32(p) << 32) | (u64)U8TO32((p) + 4))
-#define U32TO8(p, v) \
-    (p)[0] = (u8)((v) >> 24); (p)[1] = (u8)((v) >> 16); \
-    (p)[2] = (u8)((v) >>  8); (p)[3] = (u8)((v)      ); 
-#define U64TO8(p, v) \
-    U32TO8((p),     (u32)((v) >> 32));	\
-    U32TO8((p) + 4, (u32)((v)      )); 
-/*
-typedef struct  
-{ 
-	__m128i h[4];
-  u64 s[4], t[2];
-  u32 buflen, nullt;
-  u8 buf[128];
-} state __attribute__ ((aligned (64)));
-*/
-static const u8 padding[129] =
-{ 
-	0x80,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-};
-
-static inline int blake512_compress( hashState_blake * state, const u8 * datablock ) 
-{
-
-  __m128i row1l,row1h;
-  __m128i row2l,row2h;
-  __m128i row3l,row3h;
-  __m128i row4l,row4h;
-
-  const __m128i r16 = _mm_setr_epi8(2,3,4,5,6,7,0,1,10,11,12,13,14,15,8,9);
-  const __m128i u8to64 = _mm_set_epi8(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);
-
-  __m128i m0, m1, m2, m3, m4, m5, m6, m7;
-  __m128i t0, t1, t2, t3, t4, t5, t6, t7;
-  __m128i b0, b1, b2, b3;
-
-  m0 = _mm_loadu_si128((__m128i*)(datablock +   0));
-  m1 = _mm_loadu_si128((__m128i*)(datablock +  16));
-  m2 = _mm_loadu_si128((__m128i*)(datablock +  32));
-  m3 = _mm_loadu_si128((__m128i*)(datablock +  48));
-  m4 = _mm_loadu_si128((__m128i*)(datablock +  64));
-  m5 = _mm_loadu_si128((__m128i*)(datablock +  80));
-  m6 = _mm_loadu_si128((__m128i*)(datablock +  96));
-  m7 = _mm_loadu_si128((__m128i*)(datablock + 112));
-
-  m0 = BSWAP64(m0);
-  m1 = BSWAP64(m1);
-  m2 = BSWAP64(m2);
-  m3 = BSWAP64(m3);
-  m4 = BSWAP64(m4);
-  m5 = BSWAP64(m5);
-  m6 = BSWAP64(m6);
-  m7 = BSWAP64(m7);
-
-  row1l = state->h[0];
-  row1h = state->h[1];
-  row2l = state->h[2];
-  row2h = state->h[3];
-  row3l = _mm_set_epi64x(0x13198A2E03707344ULL, 0x243F6A8885A308D3ULL);
-  row3h = _mm_set_epi64x(0x082EFA98EC4E6C89ULL, 0xA4093822299F31D0ULL);
-
-  row4l = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0x452821E638D01377ULL);
-  row4h = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0xC0AC29B7C97C50DDULL);
-
-#ifdef AVOID_BRANCHING
-  do
-  {
-    const __m128i mask = _mm_cmpeq_epi32(_mm_setzero_si128(), _mm_set1_epi32(state->nullt));
-    const __m128i xor1 = _mm_and_si128(_mm_set1_epi64x(state->t[0]), mask);
-    const __m128i xor2 = _mm_and_si128(_mm_set1_epi64x(state->t[1]), mask);
-    row4l = _mm_xor_si128(row4l, xor1);
-    row4h = _mm_xor_si128(row4h, xor2);
-  } while(0);
-#else
-  if(!state->nullt)
-  {
-  	row4l = _mm_xor_si128(row4l, _mm_set1_epi64x(state->t[0]));
-  	row4h = _mm_xor_si128(row4h, _mm_set1_epi64x(state->t[1]));
-  }
-#endif
-
-  ROUND( 0);
-  ROUND( 1);
-  ROUND( 2);
-  ROUND( 3);
-  ROUND( 4);
-  ROUND( 5);
-  ROUND( 6);
-  ROUND( 7);
-  ROUND( 8);
-  ROUND( 9);
-  ROUND(10);
-  ROUND(11);
-  ROUND(12);
-  ROUND(13);
-  ROUND(14);
-  ROUND(15);
-
-  row1l = _mm_xor_si128(row3l,row1l);
-  row1h = _mm_xor_si128(row3h,row1h);
-
-  state->h[0] = _mm_xor_si128(row1l, state->h[0]);
-  state->h[1] = _mm_xor_si128(row1h, state->h[1]);
-
-  row2l = _mm_xor_si128(row4l,row2l);
-  row2h = _mm_xor_si128(row4h,row2h);
-
-  state->h[2] = _mm_xor_si128(row2l, state->h[2]);
-  state->h[3] = _mm_xor_si128(row2h, state->h[3]);
-  
-  return 0;
-}
-
-static inline void blake512_init( hashState_blake * S, u64 databitlen )
-{
-  memset(S, 0, sizeof(hashState_blake));
-  S->h[0] = _mm_set_epi64x(0xBB67AE8584CAA73BULL, 0x6A09E667F3BCC908ULL);
-  S->h[1] = _mm_set_epi64x(0xA54FF53A5F1D36F1ULL, 0x3C6EF372FE94F82BULL);
-  S->h[2] = _mm_set_epi64x(0x9B05688C2B3E6C1FULL, 0x510E527FADE682D1ULL);
-  S->h[3] = _mm_set_epi64x(0x5BE0CD19137E2179ULL, 0x1F83D9ABFB41BD6BULL);
-  S->buflen = databitlen;
-}
-
-
-static void blake512_update( hashState_blake * S, const u8 * data, u64 datalen ) 
-{
-
-
-  int left = (S->buflen >> 3); 
-  int fill = 128 - left;
-
-  if( left && ( ((datalen >> 3) & 0x7F) >= fill ) ) {
-    memcpy( (void *) (S->buf + left), (void *) data, fill );
-    S->t[0] += 1024;
-    blake512_compress( S, S->buf );
-    data += fill;
-    datalen  -= (fill << 3);       
-    left = 0;
-  }
-
-  while( datalen >= 1024 ) {  
-    S->t[0] += 1024;
-    blake512_compress( S, data );
-    data += 128;
-    datalen  -= 1024;
-  }
-
-  if( datalen > 0 ) {
-    memcpy( (void *) (S->buf + left), (void *) data, ( datalen>>3 ) & 0x7F );
-    S->buflen = (left<<3) + datalen;
-  }
-  else S->buflen=0;
-}
-
-static inline void blake512_final( hashState_blake * S, u8 * digest ) 
-{
-
-  u8 msglen[16], zo=0x01,oo=0x81;
-  u64 lo=S->t[0] + S->buflen, hi = S->t[1];
-  if ( lo < S->buflen ) hi++;
-  U64TO8(  msglen + 0, hi );
-  U64TO8(  msglen + 8, lo );
-
-  if ( S->buflen == 888 ) /* one padding byte */
-  { 
-    S->t[0] -= 8; 
-    blake512_update( S, &oo, 8 );
-  }
-  else 
-  {
-    if ( S->buflen < 888 ) /* enough space to fill the block */
-    { 
-      if ( S->buflen == 0 ) S->nullt=1;
-      S->t[0] -= 888 - S->buflen;
-      blake512_update( S, padding, 888 - S->buflen );
-    }
-    else /* NOT enough space, need 2 compressions */ 
-    { 
-      S->t[0] -= 1024 - S->buflen; 
-      blake512_update( S, padding, 1024 - S->buflen );
-      S->t[0] -= 888;
-      blake512_update( S, padding+1, 888 );
-      S->nullt = 1;
-    }
-    blake512_update( S, &zo, 8 );
-    S->t[0] -= 8;
-  }
-  S->t[0] -= 128;
-  blake512_update( S, msglen, 128 );    
-
-  do
-  {
-    const __m128i u8to64 = _mm_set_epi8(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);
-    _mm_storeu_si128((__m128i*)(digest +  0), BSWAP64(S->h[0]));
-    _mm_storeu_si128((__m128i*)(digest + 16), BSWAP64(S->h[1]));
-    _mm_storeu_si128((__m128i*)(digest + 32), BSWAP64(S->h[2]));
-    _mm_storeu_si128((__m128i*)(digest + 48), BSWAP64(S->h[3]));
-  } while(0);
-}
-
-/*
-int crypto_hash( unsigned char *out, const unsigned char *in, unsigned long long inlen ) 
-{
-
-  hashState_blake S;
-  blake512_init( &S );
-  blake512_update( &S, in, inlen*8 );
-  blake512_final( &S, out );
-  return 0;
-}
-*/
-/*
-#ifdef NOT_SUPERCOP
-
-int main() 
-{
-
-  int i, v;
-  u8 data[144], digest[64];
-  u8 test1[]= {0x97, 0x96, 0x15, 0x87, 0xF6, 0xD9, 0x70, 0xFA, 0xBA, 0x6D, 0x24, 0x78, 0x04, 0x5D, 0xE6, 0xD1, 
-	       0xFA, 0xBD, 0x09, 0xB6, 0x1A, 0xE5, 0x09, 0x32, 0x05, 0x4D, 0x52, 0xBC, 0x29, 0xD3, 0x1B, 0xE4, 
-	       0xFF, 0x91, 0x02, 0xB9, 0xF6, 0x9E, 0x2B, 0xBD, 0xB8, 0x3B, 0xE1, 0x3D, 0x4B, 0x9C, 0x06, 0x09, 
-	       0x1E, 0x5F, 0xA0, 0xB4, 0x8B, 0xD0, 0x81, 0xB6, 0x34, 0x05, 0x8B, 0xE0, 0xEC, 0x49, 0xBE, 0xB3};
-  u8 test2[]= {0x31, 0x37, 0x17, 0xD6, 0x08, 0xE9, 0xCF, 0x75, 0x8D, 0xCB, 0x1E, 0xB0, 0xF0, 0xC3, 0xCF, 0x9F, 
-	       0xC1, 0x50, 0xB2, 0xD5, 0x00, 0xFB, 0x33, 0xF5, 0x1C, 0x52, 0xAF, 0xC9, 0x9D, 0x35, 0x8A, 0x2F, 
-	       0x13, 0x74, 0xB8, 0xA3, 0x8B, 0xBA, 0x79, 0x74, 0xE7, 0xF6, 0xEF, 0x79, 0xCA, 0xB1, 0x6F, 0x22, 
-	       0xCE, 0x1E, 0x64, 0x9D, 0x6E, 0x01, 0xAD, 0x95, 0x89, 0xC2, 0x13, 0x04, 0x5D, 0x54, 0x5D, 0xDE};
-
-  for(i=0; i<144; ++i) data[i]=0;  
-
-  crypto_hash( digest, data, 1 );    
-  v=0;
-  for(i=0; i<64; ++i) {
-    printf("%02X", digest[i]);
-    if ( digest[i] != test1[i]) v=1;
-  }
-  if (v) printf("\nerror\n");
-  else  printf("\nok\n");
-
-  for(i=0; i<144; ++i) data[i]=0;  
-
-  crypto_hash( digest, data, 144 );    
-  v=0;
-  for(i=0; i<64; ++i) {
-    printf("%02X", digest[i]);
-    if ( digest[i] != test2[i]) v=1;
-  }
-  if (v) printf("\nerror\n");
-  else printf("\nok\n");
-
-  return 0;
-}
-
-#endif
-
-*/
-
-
--- a/algo/blake/sse2/blake/sse41/hash.h
+++ b/algo/blake/sse2/blake/sse41/hash.h
@@ -1,74 +0,0 @@
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <x86intrin.h>
-
-#include "config.h"
-#include "rounds.h"
-/*
-#ifndef NOT_SUPERCOP
-
-#include "crypto_hash.h"
-#include "crypto_uint64.h"
-#include "crypto_uint32.h"
-#include "crypto_uint8.h"
-
-typedef crypto_uint64 u64;
-typedef crypto_uint32 u32;
-typedef crypto_uint8 u8; 
-
-#else
-*/
-typedef unsigned long long u64; 
-typedef unsigned int u32; 
-typedef unsigned char u8; 
-
-typedef struct  
-{ 
-	__m128i h[4];
-  u64 s[4], t[2];
-  u32 buflen, nullt;
-  u8 buf[128];
-} hashState_blake __attribute__ ((aligned (64)));
-/*
-#endif
-
-#define U8TO32(p) \
-  (((u32)((p)[0]) << 24) | ((u32)((p)[1]) << 16) | \
-   ((u32)((p)[2]) <<  8) | ((u32)((p)[3])      ))
-#define U8TO64(p) \
-  (((u64)U8TO32(p) << 32) | (u64)U8TO32((p) + 4))
-#define U32TO8(p, v) \
-    (p)[0] = (u8)((v) >> 24); (p)[1] = (u8)((v) >> 16); \
-    (p)[2] = (u8)((v) >>  8); (p)[3] = (u8)((v)      ); 
-#define U64TO8(p, v) \
-    U32TO8((p),     (u32)((v) >> 32));	\
-    U32TO8((p) + 4, (u32)((v)      )); 
-*/
-
-/*
-static const u8 padding[129] =
-{ 
-	0x80,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
-    0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-};
-
-*/
-static inline void blake512_init( hashState_blake * S, u64 datalen );
-
-
-static void blake512_update( hashState_blake * S, const u8 * data, u64 datalen ) ;
-
-static inline void blake512_final( hashState_blake * S, u8 * digest ) ;
-
-
-int crypto_hash( unsigned char *out, const unsigned char *in, unsigned long long inlen ) ;
-
-
-
-
-
-
--- a/algo/blake/sse2/blake/sse41/implementors
+++ b/algo/blake/sse2/blake/sse41/implementors
@@ -1,2 +0,0 @@
-Jean-Philippe Aumasson
-Samuel Neves
--- a/algo/blake/sse2/blake/sse41/rounds.h
+++ b/algo/blake/sse2/blake/sse41/rounds.h
@@ -1,871 +0,0 @@
-
-#ifndef __BLAKE512_ROUNDS_H__
-#define __BLAKE512_ROUNDS_H__
-
-#ifndef HAVE_XOP
-	#define BSWAP64(x) _mm_shuffle_epi8((x), u8to64)
-
-	#define _mm_roti_epi64(x, c) \
-	(-(c) == 32) ? _mm_shuffle_epi32((x), _MM_SHUFFLE(2,3,0,1))  \
-	: (-(c) == 16) ? _mm_shuffle_epi8((x), r16) \
-		: _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_slli_epi64((x), 64-(-c))) 
-#else
-	#define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
-#endif
-
-
-#define LOAD_MSG_0_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m0, m1); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m2, m3); \
-t3 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0xBE5466CF34E90C6CULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_0_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m0, m1); \
-t1 = _mm_set_epi64x(0xA4093822299F31D0ULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m2, m3); \
-t3 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_0_3(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m4, m5); \
-t1 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x24A19947B3916CF7ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_0_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m5); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0x9216D5D98979FB1BULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_1_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m7, m2); \
-t1 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x2FFD72DBD01ADFB7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m4, m6); \
-t3 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x636920D871574E69ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_1_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m5, m4); \
-t1 = _mm_set_epi64x(0x452821E638D01377ULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m3, m7, 8); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0xD1310BA698DFB5ACULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_1_3(b0, b1) \
-do \
-{ \
-t0 = _mm_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
-t1 = _mm_set_epi64x(0xA4093822299F31D0ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m5, m2); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x3F84D5B5B5470917ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_1_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m6, m1); \
-t1 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m3, m1); \
-t3 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xB8E1AFED6A267E96ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_2_1(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m6, m5, 8); \
-t1 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0x9216D5D98979FB1BULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m2, m7); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0xA4093822299F31D0ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_2_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m4, m0); \
-t1 = _mm_set_epi64x(0xBA7C9045F12C7F99ULL, 0xB8E1AFED6A267E96ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m1, m6, 0xF0); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0xBE5466CF34E90C6CULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_2_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m5, m1, 0xF0); \
-t1 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m3, m4); \
-t3 = _mm_set_epi64x(0x452821E638D01377ULL, 0x13198A2E03707344ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_2_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m7, m3); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x2FFD72DBD01ADFB7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m2, m0, 8); \
-t3 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x3F84D5B5B5470917ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_3_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m3, m1); \
-t1 = _mm_set_epi64x(0x13198A2E03707344ULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m6, m5); \
-t3 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_3_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m0); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x3F84D5B5B5470917ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0x24A19947B3916CF7ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_3_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m1, m2, 0xF0); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xC0AC29B7C97C50DDULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m2, m7, 0xF0); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x243F6A8885A308D3ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_3_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m3, m5); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m0, m4); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_4_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m2); \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m1, m5); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_4_2(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m0, m3, 0xF0); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m2, m7, 0xF0); \
-t3 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xA4093822299F31D0ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_4_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m7, m5, 0xF0); \
-t1 = _mm_set_epi64x(0xBA7C9045F12C7F99ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m3, m1, 0xF0); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0x9216D5D98979FB1BULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_4_4(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m6, m0, 8); \
-t1 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m4, m6, 0xF0); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0xC0AC29B7C97C50DDULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_5_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m1, m3); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m0, m4); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0xB8E1AFED6A267E96ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_5_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m6, m5); \
-t1 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m5, m1); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x243F6A8885A308D3ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_5_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m2, m3, 0xF0); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0x24A19947B3916CF7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m7, m0); \
-t3 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x801F2E2858EFC16ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_5_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m6, m2); \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0x452821E638D01377ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m7, m4, 0xF0); \
-t3 = _mm_set_epi64x(0x13198A2E03707344ULL, 0x636920D871574E69ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_6_1(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m6, m0, 0xF0); \
-t1 = _mm_set_epi64x(0x636920D871574E69ULL, 0xBE5466CF34E90C6CULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m7, m2); \
-t3 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0x24A19947B3916CF7ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_6_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m2, m7); \
-t1 = _mm_set_epi64x(0x13198A2E03707344ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m5, m6, 8); \
-t3 = _mm_set_epi64x(0x452821E638D01377ULL, 0x801F2E2858EFC16ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_6_3(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m0, m3); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x3F84D5B5B5470917ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_shuffle_epi32(m4, _MM_SHUFFLE(1,0,3,2)); \
-t3 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0xA4093822299F31D0ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_6_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m3, m1); \
-t1 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m1, m5, 0xF0); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0xD1310BA698DFB5ACULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_7_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m6, m3); \
-t1 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xB8E1AFED6A267E96ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m6, m1, 0xF0); \
-t3 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x13198A2E03707344ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_7_2(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m7, m5, 8); \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0x24A19947B3916CF7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m0, m4); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_7_3(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m2, m7); \
-t1 = _mm_set_epi64x(0x452821E638D01377ULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m4, m1); \
-t3 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xC0AC29B7C97C50DDULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_7_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m0, m2); \
-t1 = _mm_set_epi64x(0x636920D871574E69ULL, 0xBE5466CF34E90C6CULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m3, m5); \
-t3 = _mm_set_epi64x(0xA4093822299F31D0ULL, 0x9216D5D98979FB1BULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_8_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m3, m7); \
-t1 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x636920D871574E69ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m0, m5, 8); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x82EFA98EC4E6C89ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_8_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m7, m4); \
-t1 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xC0AC29B7C97C50DDULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m4, m1, 8); \
-t3 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0xB8E1AFED6A267E96ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_8_3(b0, b1) \
-do \
-{ \
-t0 = m6; \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m5, m0, 8); \
-t3 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_8_4(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m1, m3, 0xF0); \
-t1 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = m2; \
-t3 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0x13198A2E03707344ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_9_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m5, m4); \
-t1 = _mm_set_epi64x(0x452821E638D01377ULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m3, m0); \
-t3 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xC0AC29B7C97C50DDULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_9_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m1, m2); \
-t1 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x2FFD72DBD01ADFB7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m3, m2, 0xF0); \
-t3 = _mm_set_epi64x(0x13198A2E03707344ULL, 0x3F84D5B5B5470917ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_9_3(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m7, m4); \
-t1 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xB8E1AFED6A267E96ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m1, m6); \
-t3 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_9_4(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m7, m5, 8); \
-t1 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x636920D871574E69ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m6, m0); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0x82EFA98EC4E6C89ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_10_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m0, m1); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m2, m3); \
-t3 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0xBE5466CF34E90C6CULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_10_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m0, m1); \
-t1 = _mm_set_epi64x(0xA4093822299F31D0ULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m2, m3); \
-t3 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_10_3(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m4, m5); \
-t1 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x24A19947B3916CF7ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_10_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m5); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0x9216D5D98979FB1BULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_11_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m7, m2); \
-t1 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x2FFD72DBD01ADFB7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m4, m6); \
-t3 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x636920D871574E69ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_11_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m5, m4); \
-t1 = _mm_set_epi64x(0x452821E638D01377ULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m3, m7, 8); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0xD1310BA698DFB5ACULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_11_3(b0, b1) \
-do \
-{ \
-t0 = _mm_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
-t1 = _mm_set_epi64x(0xA4093822299F31D0ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m5, m2); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x3F84D5B5B5470917ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_11_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m6, m1); \
-t1 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m3, m1); \
-t3 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xB8E1AFED6A267E96ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_12_1(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m6, m5, 8); \
-t1 = _mm_set_epi64x(0x243F6A8885A308D3ULL, 0x9216D5D98979FB1BULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m2, m7); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0xA4093822299F31D0ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_12_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m4, m0); \
-t1 = _mm_set_epi64x(0xBA7C9045F12C7F99ULL, 0xB8E1AFED6A267E96ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m1, m6, 0xF0); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0xBE5466CF34E90C6CULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_12_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m5, m1, 0xF0); \
-t1 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m3, m4); \
-t3 = _mm_set_epi64x(0x452821E638D01377ULL, 0x13198A2E03707344ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_12_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m7, m3); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x2FFD72DBD01ADFB7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_alignr_epi8(m2, m0, 8); \
-t3 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x3F84D5B5B5470917ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_13_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m3, m1); \
-t1 = _mm_set_epi64x(0x13198A2E03707344ULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m6, m5); \
-t3 = _mm_set_epi64x(0x801F2E2858EFC16ULL, 0xBA7C9045F12C7F99ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_13_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m0); \
-t1 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0x3F84D5B5B5470917ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m6, m7); \
-t3 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0x24A19947B3916CF7ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_13_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m1, m2, 0xF0); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xC0AC29B7C97C50DDULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m2, m7, 0xF0); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x243F6A8885A308D3ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_13_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m3, m5); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m0, m4); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_14_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m4, m2); \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0x243F6A8885A308D3ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m1, m5); \
-t3 = _mm_set_epi64x(0x636920D871574E69ULL, 0x452821E638D01377ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_14_2(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m0, m3, 0xF0); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0xD1310BA698DFB5ACULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m2, m7, 0xF0); \
-t3 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xA4093822299F31D0ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_14_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m7, m5, 0xF0); \
-t1 = _mm_set_epi64x(0xBA7C9045F12C7F99ULL, 0x13198A2E03707344ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m3, m1, 0xF0); \
-t3 = _mm_set_epi64x(0x24A19947B3916CF7ULL, 0x9216D5D98979FB1BULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_14_4(b0, b1) \
-do \
-{ \
-t0 = _mm_alignr_epi8(m6, m0, 8); \
-t1 = _mm_set_epi64x(0xB8E1AFED6A267E96ULL, 0x801F2E2858EFC16ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m4, m6, 0xF0); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0xC0AC29B7C97C50DDULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_15_1(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m1, m3); \
-t1 = _mm_set_epi64x(0x2FFD72DBD01ADFB7ULL, 0xBA7C9045F12C7F99ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpacklo_epi64(m0, m4); \
-t3 = _mm_set_epi64x(0x82EFA98EC4E6C89ULL, 0xB8E1AFED6A267E96ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_15_2(b0, b1) \
-do \
-{ \
-t0 = _mm_unpacklo_epi64(m6, m5); \
-t1 = _mm_set_epi64x(0xC0AC29B7C97C50DDULL, 0xA4093822299F31D0ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m5, m1); \
-t3 = _mm_set_epi64x(0x9216D5D98979FB1BULL, 0x243F6A8885A308D3ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_15_3(b0, b1) \
-do \
-{ \
-t0 = _mm_blend_epi16(m2, m3, 0xF0); \
-t1 = _mm_set_epi64x(0xBE5466CF34E90C6CULL, 0x24A19947B3916CF7ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_unpackhi_epi64(m7, m0); \
-t3 = _mm_set_epi64x(0xD1310BA698DFB5ACULL, 0x801F2E2858EFC16ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-#define LOAD_MSG_15_4(b0, b1) \
-do \
-{ \
-t0 = _mm_unpackhi_epi64(m6, m2); \
-t1 = _mm_set_epi64x(0x3F84D5B5B5470917ULL, 0x452821E638D01377ULL); \
-b0 = _mm_xor_si128(t0, t1); \
-t2 = _mm_blend_epi16(m7, m4, 0xF0); \
-t3 = _mm_set_epi64x(0x13198A2E03707344ULL, 0x636920D871574E69ULL); \
-b1 = _mm_xor_si128(t2, t3); \
-} while(0) 
-
-
-
-
-
-
-#define G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -32); \
-  row4h = _mm_roti_epi64(row4h, -32); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -25); \
-  row2h = _mm_roti_epi64(row2h, -25); \
-
-#define G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
-  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
-  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
-  \
-  row4l = _mm_xor_si128(row4l, row1l); \
-  row4h = _mm_xor_si128(row4h, row1h); \
-  \
-  row4l = _mm_roti_epi64(row4l, -16); \
-  row4h = _mm_roti_epi64(row4h, -16); \
-  \
-  row3l = _mm_add_epi64(row3l, row4l); \
-  row3h = _mm_add_epi64(row3h, row4h); \
-  \
-  row2l = _mm_xor_si128(row2l, row3l); \
-  row2h = _mm_xor_si128(row2h, row3h); \
-  \
-  row2l = _mm_roti_epi64(row2l, -11); \
-  row2h = _mm_roti_epi64(row2h, -11); \
-
-
-#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-	t0 = _mm_alignr_epi8(row2h, row2l, 8); \
-	t1 = _mm_alignr_epi8(row2l, row2h, 8); \
-	row2l = t0; \
-	row2h = t1; \
-	\
-	t0 = row3l; \
-	row3l = row3h; \
-	row3h = t0;    \
-	\
-	t0 = _mm_alignr_epi8(row4h, row4l, 8); \
-	t1 = _mm_alignr_epi8(row4l, row4h, 8); \
-	row4l = t1; \
-	row4h = t0; 
-
-#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
-	t0 = _mm_alignr_epi8(row2l, row2h, 8); \
-	t1 = _mm_alignr_epi8(row2h, row2l, 8); \
-	row2l = t0; \
-	row2h = t1; \
-	\
-	t0 = row3l; \
-	row3l = row3h; \
-	row3h = t0; \
-	\
-	t0 = _mm_alignr_epi8(row4l, row4h, 8); \
-	t1 = _mm_alignr_epi8(row4h, row4l, 8); \
-	row4l = t1; \
-	row4h = t0; 
-
-#define ROUND(r) \
-  LOAD_MSG_ ##r ##_1(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_2(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h); \
-  LOAD_MSG_ ##r ##_3(b0, b1); \
-  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  LOAD_MSG_ ##r ##_4(b0, b1); \
-  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
-  UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h);
-
-#endif
-
--- a/algo/bmw/bmw-hash-4way.h
+++ b/algo/bmw/bmw-hash-4way.h
@@ -64,7 +64,8 @@ typedef bmw_4way_small_context bmw256_4way_context;

 void bmw256_4way_init( bmw256_4way_context *ctx );

-void bmw256_4way(void *cc, const void *data, size_t len);
+void bmw256_4way_update(void *cc, const void *data, size_t len);
+#define bmw256_4way bmw256_4way_update

 void bmw256_4way_close(void *cc, void *dst);

@@ -78,7 +79,7 @@ void bmw256_4way_addbits_and_close(
 // BMW-256 8 way 32

 typedef struct {
-   __m256i buf[64];
+   __m256i buf[16];
   __m256i H[16];
   size_t ptr;
   uint32_t bit_count;  // assume bit_count fits in 32 bits
@@ -87,11 +88,33 @@ typedef struct {
 typedef bmw_8way_small_context bmw256_8way_context;

 void bmw256_8way_init( bmw256_8way_context *ctx );
-void bmw256_8way( bmw256_8way_context *ctx, const void *data, size_t len );
+void bmw256_8way_update( bmw256_8way_context *ctx, const void *data,
+                         size_t len );
+#define bmw256_8way bmw256_8way_update
 void bmw256_8way_close( bmw256_8way_context *ctx, void *dst );

 #endif

+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+// BMW-256 16 way 32
+
+typedef struct {
+   __m512i buf[16];
+   __m512i H[16];
+   size_t ptr;
+   uint32_t bit_count;  // assume bit_count fits in 32 bits
+} bmw_16way_small_context __attribute__ ((aligned (128)));
+
+typedef bmw_16way_small_context bmw256_16way_context;
+
+void bmw256_16way_init( bmw256_16way_context *ctx );
+void bmw256_16way_update( bmw256_16way_context *ctx, const void *data,
+                          size_t len );
+void bmw256_16way_close( bmw256_16way_context *ctx, void *dst );
+
+#endif
+

 #if defined(__SSE2__)

@@ -107,28 +130,29 @@ typedef struct {
 typedef bmw_2way_big_context bmw512_2way_context;

 void bmw512_2way_init( bmw512_2way_context *ctx );
-void bmw512_2way( bmw512_2way_context *ctx, const void *data, size_t len );
+void bmw512_2way_update( bmw512_2way_context *ctx, const void *data,
+                         size_t len );
 void bmw512_2way_close( bmw512_2way_context *ctx, void *dst );

 #endif // __SSE2__

 #if defined(__AVX2__)

-// BMW-512 4 way 64
+// BMW-512 64 bit 4 way

 typedef struct {
   __m256i buf[16];
   __m256i H[16];
   size_t ptr;
   sph_u64 bit_count;
-} bmw_4way_big_context;
+} bmw_4way_big_context __attribute__((aligned(128)));

 typedef bmw_4way_big_context bmw512_4way_context;

-
 void bmw512_4way_init(void *cc);

-void bmw512_4way(void *cc, const void *data, size_t len);
+void bmw512_4way_update(void *cc, const void *data, size_t len);
+#define bmw512_4way bmw512_4way_update

 void bmw512_4way_close(void *cc, void *dst);

@@ -137,6 +161,25 @@ void bmw512_4way_addbits_and_close(

 #endif  // __AVX2__

+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+// BMW-512 64 bit 8 way
+typedef struct {
+   __m512i buf[16];
+   __m512i H[16];
+   size_t ptr;
+   uint64_t bit_count;
+} bmw512_8way_context __attribute__((aligned(128)));
+
+void bmw512_8way_full( bmw512_8way_context *ctx, void *out, const void *data,
+                         size_t len );
+void bmw512_8way_init( bmw512_8way_context *ctx );
+void bmw512_8way_update( bmw512_8way_context *ctx, const void *data,
+                         size_t len );
+void bmw512_8way_close( bmw512_8way_context *ctx, void *dst );
+
+#endif // AVX512
+
 #ifdef __cplusplus
 }
 #endif
--- a/algo/bmw/bmw256-hash-4way.c
+++ b/algo/bmw/bmw256-hash-4way.c
--- a/algo/bmw/bmw512-4way.c
+++ b/algo/bmw/bmw512-4way.c
@@ -1,34 +1,88 @@
 #include "bmw512-gate.h"
-
-#ifdef BMW512_4WAY
-
 #include <stdlib.h>
 #include <string.h>
 #include <stdint.h>
 //#include "sph_keccak.h"
 #include "bmw-hash-4way.h"

+#if defined(BMW512_8WAY)
+
+void bmw512hash_8way(void *state, const void *input)
+{
+    bmw512_8way_context ctx;
+    bmw512_8way_init( &ctx );
+    bmw512_8way_update( &ctx, input, 80 );
+    bmw512_8way_close( &ctx, state );
+}
+
+int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t vdata[24*8] __attribute__ ((aligned (128)));
+   uint32_t hash[16*8] __attribute__ ((aligned (64)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   uint32_t *hash7 = &(hash[49]);   // 3*16+1
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   uint32_t n = pdata[19];
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 8;
+   __m512i  *noncev = (__m512i*)vdata + 9;   // aligned
+   const uint32_t Htarg = ptarget[7];
+   int thr_id = mythr->id;
+
+   mm512_bswap32_intrlv80_8x64( vdata, pdata );
+   do {
+       *noncev = mm512_intrlv_blend_32( mm512_bswap_32(
+                _mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0 ,
+                                  n+3, 0, n+2, 0, n+1, 0, n  , 0 ) ), *noncev );
+
+      bmw512hash_8way( hash, vdata );
+
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
+      {
+          extr_lane_8x64( lane_hash, hash, lane, 256 );
+          if ( fulltest( lane_hash, ptarget ) )
+          {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+          }
+      }
+      n += 8;
+
+   } while ( likely( ( n < last_nonce ) && !work_restart[thr_id].restart) );
+
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+   
+#elif defined(BMW512_4WAY)
+
+//#ifdef BMW512_4WAY
+
 void bmw512hash_4way(void *state, const void *input)
 {
    bmw512_4way_context ctx;
    bmw512_4way_init( &ctx );
-    bmw512_4way( &ctx, input, 80 );
+    bmw512_4way_update( &ctx, input, 80 );
    bmw512_4way_close( &ctx, state );
 }

 int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
                          uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t vdata[24*4] __attribute__ ((aligned (64)));
-   uint32_t hash[16*4] __attribute__ ((aligned (32)));
-   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t vdata[24*4] __attribute__ ((aligned (128)));
+   uint32_t hash[16*4] __attribute__ ((aligned (64)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
   uint32_t *hash7 = &(hash[25]);   // 3*8+1
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
   uint32_t n = pdata[19];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce -  4;
   __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
-//   const uint32_t Htarg = ptarget[7];
+   const uint32_t Htarg = ptarget[7];
    int thr_id = mythr->id;  // thr_id arg is deprecated

   mm256_bswap32_intrlv80_4x64( vdata, pdata );
@@ -39,20 +93,20 @@ int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
      bmw512hash_4way( hash, vdata );

      for ( int lane = 0; lane < 4; lane++ )
-      if ( ( ( hash7[ lane<<1 ] & 0xFFFFFF00 ) == 0 ) )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
      {
          extr_lane_4x64( lane_hash, hash, lane, 256 );
          if ( fulltest( lane_hash, ptarget ) )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 4;

-   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );

-   *hashes_done = n - first_nonce + 1;
+   *hashes_done = n - first_nonce;
   return 0;
 }

--- a/algo/bmw/bmw512-gate.c
+++ b/algo/bmw/bmw512-gate.c
@@ -1,13 +1,13 @@
 #include "bmw512-gate.h"

-int64_t bmw512_get_max64() { return 0x7ffffLL; }
-
 bool register_bmw512_algo( algo_gate_t* gate )
 {
-  gate->optimizations = AVX2_OPT;
-  gate->set_target      = (void*)&alt_set_target;
-  gate->get_max64       = (void*)&bmw512_get_max64;
-#if defined (BMW512_4WAY)
+  gate->optimizations = AVX2_OPT | AVX512_OPT;
+  opt_target_factor = 256.0;
+#if defined (BMW512_8WAY)
+  gate->scanhash  = (void*)&scanhash_bmw512_8way;
+  gate->hash      = (void*)&bmw512hash_8way;
+#elif defined (BMW512_4WAY)
  gate->scanhash  = (void*)&scanhash_bmw512_4way;
  gate->hash      = (void*)&bmw512hash_4way;
 #else
--- a/algo/bmw/bmw512-gate.h
+++ b/algo/bmw/bmw512-gate.h
@@ -1,23 +1,33 @@
 #ifndef BMW512_GATE_H__
-#define BMW512_GATE_H__
+#define BMW512_GATE_H__ 1

 #include "algo-gate-api.h"
 #include <stdint.h>

-#if defined(__AVX2__)
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+  #define BMW512_8WAY 1
+#elif defined(__AVX2__)
  #define BMW512_4WAY 1
 #endif

-#if defined(BMW512_4WAY)
+#if defined(BMW512_8WAY)
+
+void bmw512hash_8way( void *state, const void *input );
+int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+
+#elif defined(BMW512_4WAY)

 void bmw512hash_4way( void *state, const void *input );
 int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
                         uint64_t *hashes_done, struct thr_info *mythr );

-#endif
+#else

 void bmw512hash( void *state, const void *input );
 int scanhash_bmw512( struct work *work, uint32_t max_nonce,
                    uint64_t *hashes_done, struct thr_info *mythr );

 #endif
+
+#endif
--- a/algo/bmw/bmw512-hash-4way.c
+++ b/algo/bmw/bmw512-hash-4way.c
--- a/algo/bmw/bmw512.c
+++ b/algo/bmw/bmw512.c
@@ -1,5 +1,7 @@
 #include "algo-gate-api.h"

+#if !defined(BMW512_8WAY) && !defined(BMW512_4WAY)
+
 #include <stdlib.h>
 #include <string.h>
 #include <stdint.h>
@@ -50,4 +52,4 @@ int scanhash_bmw512( struct work *work, uint32_t max_nonce,
 	pdata[19] = n;
 	return 0;
 }
-
+#endif
--- a/algo/bmw/sph_bmw.c
+++ b/algo/bmw/sph_bmw.c
@@ -48,6 +48,8 @@ extern "C"{
 #pragma warning (disable: 4146)
 #endif

+#if !defined(__AVX2__)
+
 static const sph_u32 IV224[] = {
 	SPH_C32(0x00010203), SPH_C32(0x04050607),
 	SPH_C32(0x08090A0B), SPH_C32(0x0C0D0E0F),
@@ -70,6 +72,8 @@ static const sph_u32 IV256[] = {
 	SPH_C32(0x78797A7B), SPH_C32(0x7C7D7E7F)
 };

+#endif // !AVX2
+
 #if SPH_64

 static const sph_u64 IV384[] = {
@@ -135,6 +139,8 @@ static const sph_u64 IV512[] = {
 #define M16_30   14, 15,  1,  2,  5,  8,  9
 #define M16_31   15, 16,  2,  3,  6,  9, 10

+#if !defined(__AVX2__)
+
 #define ss0(x)    (((x) >> 1) ^ SPH_T32((x) << 3) \
                  ^ SPH_ROTL32(x,  4) ^ SPH_ROTL32(x, 19))
 #define ss1(x)    (((x) >> 1) ^ SPH_T32((x) << 2) \
@@ -189,6 +195,8 @@ static const sph_u64 IV512[] = {
 #define expand2s_(qf, mf, hf, i16, ix, iy) \
 	expand2s_inner LPAR qf, mf, hf, i16, ix, iy)

+#endif // !AVX2
+
 #if SPH_64

 #define sb0(x)    (((x) >> 1) ^ SPH_T64((x) << 3) \
@@ -291,6 +299,8 @@ static const sph_u64 Kb_tab[] = {
 	tt((M(i0) ^ H(i0)) op01 (M(i1) ^ H(i1)) op12 (M(i2) ^ H(i2)) \
 	op23 (M(i3) ^ H(i3)) op34 (M(i4) ^ H(i4)))

+#if !defined(__AVX2__)
+
 #define Ws0    MAKE_W(SPH_T32,  5, -,  7, +, 10, +, 13, +, 14)
 #define Ws1    MAKE_W(SPH_T32,  6, -,  8, +, 11, +, 14, -, 15)
 #define Ws2    MAKE_W(SPH_T32,  0, +,  7, +,  9, -, 12, +, 15)
@@ -407,6 +417,8 @@ static const sph_u64 Kb_tab[] = {

 #define Qs(j)   (qt[j])

+#endif  // !AVX2
+
 #if SPH_64

 #define Wb0    MAKE_W(SPH_T64,  5, -,  7, +, 10, +, 13, +, 14)
@@ -557,7 +569,6 @@ static const sph_u64 Kb_tab[] = {
 			+ ((xl >> 2) ^ qf(22) ^ qf(15))); \
 	} while (0)

-#define FOLDs   FOLD(sph_u32, MAKE_Qs, SPH_T32, SPH_ROTL32, M, Qs, dH)

 #if SPH_64

@@ -565,6 +576,10 @@ static const sph_u64 Kb_tab[] = {

 #endif

+#if !defined(__AVX2__)
+
+#define FOLDs   FOLD(sph_u32, MAKE_Qs, SPH_T32, SPH_ROTL32, M, Qs, dH)
+
 static void
 compress_small(const unsigned char *data, const sph_u32 h[16], sph_u32 dh[16])
 {
@@ -711,6 +726,8 @@ bmw32_close(sph_bmw_small_context *sc, unsigned ub, unsigned n,
 		sph_enc32le(out + 4 * u, h1[v]);
 }

+#endif // !AVX2
+
 #if SPH_64

 static void
@@ -840,6 +857,8 @@ bmw64_close(sph_bmw_big_context *sc, unsigned ub, unsigned n,

 #endif

+#if !defined(__AVX2__)
+
 /* see sph_bmw.h */
 void
 sph_bmw224_init(void *cc)
@@ -898,6 +917,8 @@ sph_bmw256_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
 //	sph_bmw256_init(cc);
 }

+#endif // !AVX2
+
 #if SPH_64

 /* see sph_bmw.h */
--- a/algo/bmw/sph_bmw.h
+++ b/algo/bmw/sph_bmw.h
@@ -77,6 +77,9 @@ extern "C"{
 * computation can be cloned by copying the context (e.g. with a simple
 * <code>memcpy()</code>).
 */
+
+#if !defined(__AVX2__)
+
 typedef struct {
 #ifndef DOXYGEN_IGNORE
 	unsigned char buf[64];    /* first field, for alignment */
@@ -102,6 +105,8 @@ typedef sph_bmw_small_context sph_bmw224_context;
 */
 typedef sph_bmw_small_context sph_bmw256_context;

+#endif // !AVX2
+
 #if SPH_64

 /**
@@ -137,6 +142,8 @@ typedef sph_bmw_big_context sph_bmw512_context;

 #endif

+#if !defined(__AVX2__)
+
 /**
 * Initialize a BMW-224 context. This process performs no memory allocation.
 *
@@ -227,6 +234,8 @@ void sph_bmw256_close(void *cc, void *dst);
 void sph_bmw256_addbits_and_close(
 	void *cc, unsigned ub, unsigned n, void *dst);

+#endif // !AVX2
+
 #if SPH_64

 /**
--- a/algo/bmw/sse2/bmw.c
+++ b/algo/bmw/sse2/bmw.c
@@ -1,519 +0,0 @@
-/* $Id: bmw.c 227 2010-06-16 17:28:38Z tp $ */
-/*
- * BMW implementation.
- *
- * ==========================(LICENSE BEGIN)============================
- *
- * Copyright (c) 2007-2010  Projet RNRT SAPHIR
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
- * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * ===========================(LICENSE END)=============================
- *
- * @author   Thomas Pornin <thomas.pornin@cryptolog.com>
- */
-
-#include <stddef.h>
-#include <string.h>
-#include <limits.h>
-
-#ifdef __cplusplus
-extern "C"{
-#endif
-
-#include "../sph_bmw.h"
-
-#ifdef _MSC_VER
-#pragma warning (disable: 4146)
-#endif
-
-static const sph_u64 bmwIV512[] = {
-	SPH_C64(0x8081828384858687), SPH_C64(0x88898A8B8C8D8E8F),
-	SPH_C64(0x9091929394959697), SPH_C64(0x98999A9B9C9D9E9F),
-	SPH_C64(0xA0A1A2A3A4A5A6A7), SPH_C64(0xA8A9AAABACADAEAF),
-	SPH_C64(0xB0B1B2B3B4B5B6B7), SPH_C64(0xB8B9BABBBCBDBEBF),
-	SPH_C64(0xC0C1C2C3C4C5C6C7), SPH_C64(0xC8C9CACBCCCDCECF),
-	SPH_C64(0xD0D1D2D3D4D5D6D7), SPH_C64(0xD8D9DADBDCDDDEDF),
-	SPH_C64(0xE0E1E2E3E4E5E6E7), SPH_C64(0xE8E9EAEBECEDEEEF),
-	SPH_C64(0xF0F1F2F3F4F5F6F7), SPH_C64(0xF8F9FAFBFCFDFEFF)
-};
-
-#define XCAT(x, y)    XCAT_(x, y)
-#define XCAT_(x, y)   x ## y
-
-#define LPAR   (
-
-#define I16_16    0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15
-#define I16_17    1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16
-#define I16_18    2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17
-#define I16_19    3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18
-#define I16_20    4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19
-#define I16_21    5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20
-#define I16_22    6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21
-#define I16_23    7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22
-#define I16_24    8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23
-#define I16_25    9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24
-#define I16_26   10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25
-#define I16_27   11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26
-#define I16_28   12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
-#define I16_29   13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28
-#define I16_30   14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29
-#define I16_31   15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30
-
-#define M16_16    0,  1,  3,  4,  7, 10, 11
-#define M16_17    1,  2,  4,  5,  8, 11, 12
-#define M16_18    2,  3,  5,  6,  9, 12, 13
-#define M16_19    3,  4,  6,  7, 10, 13, 14
-#define M16_20    4,  5,  7,  8, 11, 14, 15
-#define M16_21    5,  6,  8,  9, 12, 15, 16
-#define M16_22    6,  7,  9, 10, 13,  0,  1
-#define M16_23    7,  8, 10, 11, 14,  1,  2
-#define M16_24    8,  9, 11, 12, 15,  2,  3
-#define M16_25    9, 10, 12, 13,  0,  3,  4
-#define M16_26   10, 11, 13, 14,  1,  4,  5
-#define M16_27   11, 12, 14, 15,  2,  5,  6
-#define M16_28   12, 13, 15, 16,  3,  6,  7
-#define M16_29   13, 14,  0,  1,  4,  7,  8
-#define M16_30   14, 15,  1,  2,  5,  8,  9
-#define M16_31   15, 16,  2,  3,  6,  9, 10
-
-#define ss0(x)    (((x) >> 1) ^ SPH_T32((x) << 3) \
-                  ^ SPH_ROTL32(x,  4) ^ SPH_ROTL32(x, 19))
-#define ss1(x)    (((x) >> 1) ^ SPH_T32((x) << 2) \
-                  ^ SPH_ROTL32(x,  8) ^ SPH_ROTL32(x, 23))
-#define ss2(x)    (((x) >> 2) ^ SPH_T32((x) << 1) \
-                  ^ SPH_ROTL32(x, 12) ^ SPH_ROTL32(x, 25))
-#define ss3(x)    (((x) >> 2) ^ SPH_T32((x) << 2) \
-                  ^ SPH_ROTL32(x, 15) ^ SPH_ROTL32(x, 29))
-#define ss4(x)    (((x) >> 1) ^ (x))
-#define ss5(x)    (((x) >> 2) ^ (x))
-#define rs1(x)    SPH_ROTL32(x,  3)
-#define rs2(x)    SPH_ROTL32(x,  7)
-#define rs3(x)    SPH_ROTL32(x, 13)
-#define rs4(x)    SPH_ROTL32(x, 16)
-#define rs5(x)    SPH_ROTL32(x, 19)
-#define rs6(x)    SPH_ROTL32(x, 23)
-#define rs7(x)    SPH_ROTL32(x, 27)
-
-#define Ks(j)   SPH_T32((sph_u32)(j) * SPH_C32(0x05555555))
-
-#define add_elt_s(mf, hf, j0m, j1m, j3m, j4m, j7m, j10m, j11m, j16) \
-	(SPH_T32(SPH_ROTL32(mf(j0m), j1m) + SPH_ROTL32(mf(j3m), j4m) \
-		- SPH_ROTL32(mf(j10m), j11m) + Ks(j16)) ^ hf(j7m))
-
-#define expand1s_inner(qf, mf, hf, i16, \
-		i0, i1, i2, i3, i4, i5, i6, i7, i8, \
-		i9, i10, i11, i12, i13, i14, i15, \
-		i0m, i1m, i3m, i4m, i7m, i10m, i11m) \
-	SPH_T32(ss1(qf(i0)) + ss2(qf(i1)) + ss3(qf(i2)) + ss0(qf(i3)) \
-		+ ss1(qf(i4)) + ss2(qf(i5)) + ss3(qf(i6)) + ss0(qf(i7)) \
-		+ ss1(qf(i8)) + ss2(qf(i9)) + ss3(qf(i10)) + ss0(qf(i11)) \
-		+ ss1(qf(i12)) + ss2(qf(i13)) + ss3(qf(i14)) + ss0(qf(i15)) \
-		+ add_elt_s(mf, hf, i0m, i1m, i3m, i4m, i7m, i10m, i11m, i16))
-
-#define expand1s(qf, mf, hf, i16) \
-	expand1s_(qf, mf, hf, i16, I16_ ## i16, M16_ ## i16)
-#define expand1s_(qf, mf, hf, i16, ix, iy) \
-	expand1s_inner LPAR qf, mf, hf, i16, ix, iy)
-
-#define expand2s_inner(qf, mf, hf, i16, \
-		i0, i1, i2, i3, i4, i5, i6, i7, i8, \
-		i9, i10, i11, i12, i13, i14, i15, \
-		i0m, i1m, i3m, i4m, i7m, i10m, i11m) \
-	SPH_T32(qf(i0) + rs1(qf(i1)) + qf(i2) + rs2(qf(i3)) \
-		+ qf(i4) + rs3(qf(i5)) + qf(i6) + rs4(qf(i7)) \
-		+ qf(i8) + rs5(qf(i9)) + qf(i10) + rs6(qf(i11)) \
-		+ qf(i12) + rs7(qf(i13)) + ss4(qf(i14)) + ss5(qf(i15)) \
-		+ add_elt_s(mf, hf, i0m, i1m, i3m, i4m, i7m, i10m, i11m, i16))
-
-#define expand2s(qf, mf, hf, i16) \
-	expand2s_(qf, mf, hf, i16, I16_ ## i16, M16_ ## i16)
-#define expand2s_(qf, mf, hf, i16, ix, iy) \
-	expand2s_inner LPAR qf, mf, hf, i16, ix, iy)
-
-#if SPH_64
-
-#define sb0(x)    (((x) >> 1) ^ SPH_T64((x) << 3) \
-                  ^ SPH_ROTL64(x,  4) ^ SPH_ROTL64(x, 37))
-#define sb1(x)    (((x) >> 1) ^ SPH_T64((x) << 2) \
-                  ^ SPH_ROTL64(x, 13) ^ SPH_ROTL64(x, 43))
-#define sb2(x)    (((x) >> 2) ^ SPH_T64((x) << 1) \
-                  ^ SPH_ROTL64(x, 19) ^ SPH_ROTL64(x, 53))
-#define sb3(x)    (((x) >> 2) ^ SPH_T64((x) << 2) \
-                  ^ SPH_ROTL64(x, 28) ^ SPH_ROTL64(x, 59))
-#define sb4(x)    (((x) >> 1) ^ (x))
-#define sb5(x)    (((x) >> 2) ^ (x))
-#define rb1(x)    SPH_ROTL64(x,  5)
-#define rb2(x)    SPH_ROTL64(x, 11)
-#define rb3(x)    SPH_ROTL64(x, 27)
-#define rb4(x)    SPH_ROTL64(x, 32)
-#define rb5(x)    SPH_ROTL64(x, 37)
-#define rb6(x)    SPH_ROTL64(x, 43)
-#define rb7(x)    SPH_ROTL64(x, 53)
-
-#define Kb(j)   SPH_T64((sph_u64)(j) * SPH_C64(0x0555555555555555))
-
-#if 0
-
-static const sph_u64 Kb_tab[] = {
-	Kb(16), Kb(17), Kb(18), Kb(19), Kb(20), Kb(21), Kb(22), Kb(23),
-	Kb(24), Kb(25), Kb(26), Kb(27), Kb(28), Kb(29), Kb(30), Kb(31)
-};
-
-#define rol_off(mf, j, off) \
-	SPH_ROTL64(mf(((j) + (off)) & 15), (((j) + (off)) & 15) + 1)
-
-#define add_elt_b(mf, hf, j) \
-	(SPH_T64(rol_off(mf, j, 0) + rol_off(mf, j, 3) \
-		- rol_off(mf, j, 10) + Kb_tab[j]) ^ hf(((j) + 7) & 15))
-
-#define expand1b(qf, mf, hf, i) \
-	SPH_T64(sb1(qf((i) - 16)) + sb2(qf((i) - 15)) \
-		+ sb3(qf((i) - 14)) + sb0(qf((i) - 13)) \
-		+ sb1(qf((i) - 12)) + sb2(qf((i) - 11)) \
-		+ sb3(qf((i) - 10)) + sb0(qf((i) - 9)) \
-		+ sb1(qf((i) - 8)) + sb2(qf((i) - 7)) \
-		+ sb3(qf((i) - 6)) + sb0(qf((i) - 5)) \
-		+ sb1(qf((i) - 4)) + sb2(qf((i) - 3)) \
-		+ sb3(qf((i) - 2)) + sb0(qf((i) - 1)) \
-		+ add_elt_b(mf, hf, (i) - 16))
-
-#define expand2b(qf, mf, hf, i) \
-	SPH_T64(qf((i) - 16) + rb1(qf((i) - 15)) \
-		+ qf((i) - 14) + rb2(qf((i) - 13)) \
-		+ qf((i) - 12) + rb3(qf((i) - 11)) \
-		+ qf((i) - 10) + rb4(qf((i) - 9)) \
-		+ qf((i) - 8) + rb5(qf((i) - 7)) \
-		+ qf((i) - 6) + rb6(qf((i) - 5)) \
-		+ qf((i) - 4) + rb7(qf((i) - 3)) \
-		+ sb4(qf((i) - 2)) + sb5(qf((i) - 1)) \
-		+ add_elt_b(mf, hf, (i) - 16))
-
-#else
-
-#define add_elt_b(mf, hf, j0m, j1m, j3m, j4m, j7m, j10m, j11m, j16) \
-	(SPH_T64(SPH_ROTL64(mf(j0m), j1m) + SPH_ROTL64(mf(j3m), j4m) \
-		- SPH_ROTL64(mf(j10m), j11m) + Kb(j16)) ^ hf(j7m))
-
-#define expand1b_inner(qf, mf, hf, i16, \
-		i0, i1, i2, i3, i4, i5, i6, i7, i8, \
-		i9, i10, i11, i12, i13, i14, i15, \
-		i0m, i1m, i3m, i4m, i7m, i10m, i11m) \
-	SPH_T64(sb1(qf(i0)) + sb2(qf(i1)) + sb3(qf(i2)) + sb0(qf(i3)) \
-		+ sb1(qf(i4)) + sb2(qf(i5)) + sb3(qf(i6)) + sb0(qf(i7)) \
-		+ sb1(qf(i8)) + sb2(qf(i9)) + sb3(qf(i10)) + sb0(qf(i11)) \
-		+ sb1(qf(i12)) + sb2(qf(i13)) + sb3(qf(i14)) + sb0(qf(i15)) \
-		+ add_elt_b(mf, hf, i0m, i1m, i3m, i4m, i7m, i10m, i11m, i16))
-
-#define expand1b(qf, mf, hf, i16) \
-	expand1b_(qf, mf, hf, i16, I16_ ## i16, M16_ ## i16)
-#define expand1b_(qf, mf, hf, i16, ix, iy) \
-	expand1b_inner LPAR qf, mf, hf, i16, ix, iy)
-
-#define expand2b_inner(qf, mf, hf, i16, \
-		i0, i1, i2, i3, i4, i5, i6, i7, i8, \
-		i9, i10, i11, i12, i13, i14, i15, \
-		i0m, i1m, i3m, i4m, i7m, i10m, i11m) \
-	SPH_T64(qf(i0) + rb1(qf(i1)) + qf(i2) + rb2(qf(i3)) \
-		+ qf(i4) + rb3(qf(i5)) + qf(i6) + rb4(qf(i7)) \
-		+ qf(i8) + rb5(qf(i9)) + qf(i10) + rb6(qf(i11)) \
-		+ qf(i12) + rb7(qf(i13)) + sb4(qf(i14)) + sb5(qf(i15)) \
-		+ add_elt_b(mf, hf, i0m, i1m, i3m, i4m, i7m, i10m, i11m, i16))
-
-#define expand2b(qf, mf, hf, i16) \
-	expand2b_(qf, mf, hf, i16, I16_ ## i16, M16_ ## i16)
-#define expand2b_(qf, mf, hf, i16, ix, iy) \
-	expand2b_inner LPAR qf, mf, hf, i16, ix, iy)
-
-#endif
-
-#endif
-
-#define MAKE_W(tt, i0, op01, i1, op12, i2, op23, i3, op34, i4) \
-	tt((M(i0) ^ H(i0)) op01 (M(i1) ^ H(i1)) op12 (M(i2) ^ H(i2)) \
-	op23 (M(i3) ^ H(i3)) op34 (M(i4) ^ H(i4)))
-
-#define Ws0    MAKE_W(SPH_T32,  5, -,  7, +, 10, +, 13, +, 14)
-#define Ws1    MAKE_W(SPH_T32,  6, -,  8, +, 11, +, 14, -, 15)
-#define Ws2    MAKE_W(SPH_T32,  0, +,  7, +,  9, -, 12, +, 15)
-#define Ws3    MAKE_W(SPH_T32,  0, -,  1, +,  8, -, 10, +, 13)
-#define Ws4    MAKE_W(SPH_T32,  1, +,  2, +,  9, -, 11, -, 14)
-#define Ws5    MAKE_W(SPH_T32,  3, -,  2, +, 10, -, 12, +, 15)
-#define Ws6    MAKE_W(SPH_T32,  4, -,  0, -,  3, -, 11, +, 13)
-#define Ws7    MAKE_W(SPH_T32,  1, -,  4, -,  5, -, 12, -, 14)
-#define Ws8    MAKE_W(SPH_T32,  2, -,  5, -,  6, +, 13, -, 15)
-#define Ws9    MAKE_W(SPH_T32,  0, -,  3, +,  6, -,  7, +, 14)
-#define Ws10   MAKE_W(SPH_T32,  8, -,  1, -,  4, -,  7, +, 15)
-#define Ws11   MAKE_W(SPH_T32,  8, -,  0, -,  2, -,  5, +,  9)
-#define Ws12   MAKE_W(SPH_T32,  1, +,  3, -,  6, -,  9, +, 10)
-#define Ws13   MAKE_W(SPH_T32,  2, +,  4, +,  7, +, 10, +, 11)
-#define Ws14   MAKE_W(SPH_T32,  3, -,  5, +,  8, -, 11, -, 12)
-#define Ws15   MAKE_W(SPH_T32, 12, -,  4, -,  6, -,  9, +, 13)
-
-#define MAKE_Qas   do { \
-		qt[ 0] = SPH_T32(ss0(Ws0 ) + H( 1)); \
-		qt[ 1] = SPH_T32(ss1(Ws1 ) + H( 2)); \
-		qt[ 2] = SPH_T32(ss2(Ws2 ) + H( 3)); \
-		qt[ 3] = SPH_T32(ss3(Ws3 ) + H( 4)); \
-		qt[ 4] = SPH_T32(ss4(Ws4 ) + H( 5)); \
-		qt[ 5] = SPH_T32(ss0(Ws5 ) + H( 6)); \
-		qt[ 6] = SPH_T32(ss1(Ws6 ) + H( 7)); \
-		qt[ 7] = SPH_T32(ss2(Ws7 ) + H( 8)); \
-		qt[ 8] = SPH_T32(ss3(Ws8 ) + H( 9)); \
-		qt[ 9] = SPH_T32(ss4(Ws9 ) + H(10)); \
-		qt[10] = SPH_T32(ss0(Ws10) + H(11)); \
-		qt[11] = SPH_T32(ss1(Ws11) + H(12)); \
-		qt[12] = SPH_T32(ss2(Ws12) + H(13)); \
-		qt[13] = SPH_T32(ss3(Ws13) + H(14)); \
-		qt[14] = SPH_T32(ss4(Ws14) + H(15)); \
-		qt[15] = SPH_T32(ss0(Ws15) + H( 0)); \
-	} while (0)
-
-#define MAKE_Qbs   do { \
-		qt[16] = expand1s(Qs, M, H, 16); \
-		qt[17] = expand1s(Qs, M, H, 17); \
-		qt[18] = expand2s(Qs, M, H, 18); \
-		qt[19] = expand2s(Qs, M, H, 19); \
-		qt[20] = expand2s(Qs, M, H, 20); \
-		qt[21] = expand2s(Qs, M, H, 21); \
-		qt[22] = expand2s(Qs, M, H, 22); \
-		qt[23] = expand2s(Qs, M, H, 23); \
-		qt[24] = expand2s(Qs, M, H, 24); \
-		qt[25] = expand2s(Qs, M, H, 25); \
-		qt[26] = expand2s(Qs, M, H, 26); \
-		qt[27] = expand2s(Qs, M, H, 27); \
-		qt[28] = expand2s(Qs, M, H, 28); \
-		qt[29] = expand2s(Qs, M, H, 29); \
-		qt[30] = expand2s(Qs, M, H, 30); \
-		qt[31] = expand2s(Qs, M, H, 31); \
-	} while (0)
-
-#define MAKE_Qs   do { \
-		MAKE_Qas; \
-		MAKE_Qbs; \
-	} while (0)
-
-#define Qs(j)   (qt[j])
-
-#define Wb0    MAKE_W(SPH_T64,  5, -,  7, +, 10, +, 13, +, 14)
-#define Wb1    MAKE_W(SPH_T64,  6, -,  8, +, 11, +, 14, -, 15)
-#define Wb2    MAKE_W(SPH_T64,  0, +,  7, +,  9, -, 12, +, 15)
-#define Wb3    MAKE_W(SPH_T64,  0, -,  1, +,  8, -, 10, +, 13)
-#define Wb4    MAKE_W(SPH_T64,  1, +,  2, +,  9, -, 11, -, 14)
-#define Wb5    MAKE_W(SPH_T64,  3, -,  2, +, 10, -, 12, +, 15)
-#define Wb6    MAKE_W(SPH_T64,  4, -,  0, -,  3, -, 11, +, 13)
-#define Wb7    MAKE_W(SPH_T64,  1, -,  4, -,  5, -, 12, -, 14)
-#define Wb8    MAKE_W(SPH_T64,  2, -,  5, -,  6, +, 13, -, 15)
-#define Wb9    MAKE_W(SPH_T64,  0, -,  3, +,  6, -,  7, +, 14)
-#define Wb10   MAKE_W(SPH_T64,  8, -,  1, -,  4, -,  7, +, 15)
-#define Wb11   MAKE_W(SPH_T64,  8, -,  0, -,  2, -,  5, +,  9)
-#define Wb12   MAKE_W(SPH_T64,  1, +,  3, -,  6, -,  9, +, 10)
-#define Wb13   MAKE_W(SPH_T64,  2, +,  4, +,  7, +, 10, +, 11)
-#define Wb14   MAKE_W(SPH_T64,  3, -,  5, +,  8, -, 11, -, 12)
-#define Wb15   MAKE_W(SPH_T64, 12, -,  4, -,  6, -,  9, +, 13)
-
-#define MAKE_Qab   do { \
-		qt[ 0] = SPH_T64(sb0(Wb0 ) + H( 1)); \
-		qt[ 1] = SPH_T64(sb1(Wb1 ) + H( 2)); \
-		qt[ 2] = SPH_T64(sb2(Wb2 ) + H( 3)); \
-		qt[ 3] = SPH_T64(sb3(Wb3 ) + H( 4)); \
-		qt[ 4] = SPH_T64(sb4(Wb4 ) + H( 5)); \
-		qt[ 5] = SPH_T64(sb0(Wb5 ) + H( 6)); \
-		qt[ 6] = SPH_T64(sb1(Wb6 ) + H( 7)); \
-		qt[ 7] = SPH_T64(sb2(Wb7 ) + H( 8)); \
-		qt[ 8] = SPH_T64(sb3(Wb8 ) + H( 9)); \
-		qt[ 9] = SPH_T64(sb4(Wb9 ) + H(10)); \
-		qt[10] = SPH_T64(sb0(Wb10) + H(11)); \
-		qt[11] = SPH_T64(sb1(Wb11) + H(12)); \
-		qt[12] = SPH_T64(sb2(Wb12) + H(13)); \
-		qt[13] = SPH_T64(sb3(Wb13) + H(14)); \
-		qt[14] = SPH_T64(sb4(Wb14) + H(15)); \
-		qt[15] = SPH_T64(sb0(Wb15) + H( 0)); \
-	} while (0)
-
-#define MAKE_Qbb   do { \
-		qt[16] = expand1b(Qb, M, H, 16); \
-		qt[17] = expand1b(Qb, M, H, 17); \
-		qt[18] = expand2b(Qb, M, H, 18); \
-		qt[19] = expand2b(Qb, M, H, 19); \
-		qt[20] = expand2b(Qb, M, H, 20); \
-		qt[21] = expand2b(Qb, M, H, 21); \
-		qt[22] = expand2b(Qb, M, H, 22); \
-		qt[23] = expand2b(Qb, M, H, 23); \
-		qt[24] = expand2b(Qb, M, H, 24); \
-		qt[25] = expand2b(Qb, M, H, 25); \
-		qt[26] = expand2b(Qb, M, H, 26); \
-		qt[27] = expand2b(Qb, M, H, 27); \
-		qt[28] = expand2b(Qb, M, H, 28); \
-		qt[29] = expand2b(Qb, M, H, 29); \
-		qt[30] = expand2b(Qb, M, H, 30); \
-		qt[31] = expand2b(Qb, M, H, 31); \
-	} while (0)
-
-#define MAKE_Qb   do { \
-		MAKE_Qab; \
-		MAKE_Qbb; \
-	} while (0)
-
-#define Qb(j)   (qt[j])
-
-#define FOLD(type, mkQ, tt, rol, mf, qf, dhf)   do { \
-		type qt[32], xl, xh; \
-		mkQ; \
-		xl = qf(16) ^ qf(17) ^ qf(18) ^ qf(19) \
-			^ qf(20) ^ qf(21) ^ qf(22) ^ qf(23); \
-		xh = xl ^ qf(24) ^ qf(25) ^ qf(26) ^ qf(27) \
-			^ qf(28) ^ qf(29) ^ qf(30) ^ qf(31); \
-		dhf( 0) = tt(((xh <<  5) ^ (qf(16) >>  5) ^ mf( 0)) \
-			+ (xl ^ qf(24) ^ qf( 0))); \
-		dhf( 1) = tt(((xh >>  7) ^ (qf(17) <<  8) ^ mf( 1)) \
-			+ (xl ^ qf(25) ^ qf( 1))); \
-		dhf( 2) = tt(((xh >>  5) ^ (qf(18) <<  5) ^ mf( 2)) \
-			+ (xl ^ qf(26) ^ qf( 2))); \
-		dhf( 3) = tt(((xh >>  1) ^ (qf(19) <<  5) ^ mf( 3)) \
-			+ (xl ^ qf(27) ^ qf( 3))); \
-		dhf( 4) = tt(((xh >>  3) ^ (qf(20) <<  0) ^ mf( 4)) \
-			+ (xl ^ qf(28) ^ qf( 4))); \
-		dhf( 5) = tt(((xh <<  6) ^ (qf(21) >>  6) ^ mf( 5)) \
-			+ (xl ^ qf(29) ^ qf( 5))); \
-		dhf( 6) = tt(((xh >>  4) ^ (qf(22) <<  6) ^ mf( 6)) \
-			+ (xl ^ qf(30) ^ qf( 6))); \
-		dhf( 7) = tt(((xh >> 11) ^ (qf(23) <<  2) ^ mf( 7)) \
-			+ (xl ^ qf(31) ^ qf( 7))); \
-		dhf( 8) = tt(rol(dhf(4),  9) + (xh ^ qf(24) ^ mf( 8)) \
-			+ ((xl << 8) ^ qf(23) ^ qf( 8))); \
-		dhf( 9) = tt(rol(dhf(5), 10) + (xh ^ qf(25) ^ mf( 9)) \
-			+ ((xl >> 6) ^ qf(16) ^ qf( 9))); \
-		dhf(10) = tt(rol(dhf(6), 11) + (xh ^ qf(26) ^ mf(10)) \
-			+ ((xl << 6) ^ qf(17) ^ qf(10))); \
-		dhf(11) = tt(rol(dhf(7), 12) + (xh ^ qf(27) ^ mf(11)) \
-			+ ((xl << 4) ^ qf(18) ^ qf(11))); \
-		dhf(12) = tt(rol(dhf(0), 13) + (xh ^ qf(28) ^ mf(12)) \
-			+ ((xl >> 3) ^ qf(19) ^ qf(12))); \
-		dhf(13) = tt(rol(dhf(1), 14) + (xh ^ qf(29) ^ mf(13)) \
-			+ ((xl >> 4) ^ qf(20) ^ qf(13))); \
-		dhf(14) = tt(rol(dhf(2), 15) + (xh ^ qf(30) ^ mf(14)) \
-			+ ((xl >> 7) ^ qf(21) ^ qf(14))); \
-		dhf(15) = tt(rol(dhf(3), 16) + (xh ^ qf(31) ^ mf(15)) \
-			+ ((xl >> 2) ^ qf(22) ^ qf(15))); \
-	} while (0)
-
-#define FOLDs   FOLD(sph_u32, MAKE_Qs, SPH_T32, SPH_ROTL32, M, Qs, dH)
-
-#define FOLDb   FOLD(sph_u64, MAKE_Qb, SPH_T64, SPH_ROTL64, M, Qb, dH)
-
-#define DECL_BMW \
-    sph_u64 bmwH[16]; \
-
-/* load initial constants */
-#define BMW_I \
-do { \
-    memcpy(bmwH, bmwIV512, sizeof bmwH); \
-    hashptr = 0; \
-    hashctA = 0; \
-} while (0) 
-
-/* load hash for loop */
-#define BMW_U \
-do { \
-    const void *data = hash; \
-    size_t len = 64; \
-    unsigned char *buf; \
-    \
-    hashctA += (sph_u64)len << 3; \
-    buf = hashbuf; \
-    memcpy(buf, data, 64); \
-    hashptr = 64; \
-} while (0)  
-
-
-/* bmw512 hash loaded */
-/* hash = blake512(loaded) */
-#define BMW_C \
-do { \
-    void *dst = hash; \
-    size_t out_size_w64 = 8; \
-    unsigned char *data; \
-    sph_u64 *dh; \
-    unsigned char *out; \
-    size_t ptr, u, v; \
-    unsigned z; \
-    sph_u64 h1[16], h2[16], *h; \
-    data = hashbuf; \
-    ptr = hashptr; \
-    z = 0x80 >> 0; \
-    data[ptr ++] = ((0 & -z) | z) & 0xFF; \
-    memset(data + ptr, 0, (sizeof(char)*128) - 8 - ptr); \
-    sph_enc64le_aligned(data + (sizeof(char)*128) - 8, \
-    SPH_T64(hashctA + 0)); \
-    /* for break loop */ \
-    /* one copy of inline FOLD */ \
-    /* FOLD uses, */ \
-    /* uint64 *h, data */ \
-    /* uint64 dh, state */ \
-        h = bmwH; \
-        dh = h2; \
-    for (;;) { \
-        FOLDb; \
-        /* dh gets changed for 2nd run */ \
-        if (dh == h1) break; \
-        for (u = 0; u < 16; u ++) \
-        sph_enc64le_aligned(data + 8 * u, h2[u]); \
-        dh = h1; \
-        h = (sph_u64*)final_b; \
-    } \
-    /* end wrapped for break loop */ \
-    out = dst; \
-    for (u = 0, v = 16 - out_size_w64; u < out_size_w64; u ++, v ++) \
-    sph_enc64le(out + 8 * u, h1[v]); \
-} while (0) 
-
-/*
-static void
-compress_big(const unsigned char *data, const sph_u64 h[16], sph_u64 dh[16])
-{
-
-#define M(x)    sph_dec64le_aligned(data + 8 * (x))
-#define H(x)    (h[x])
-#define dH(x)   (dh[x])
-
-	FOLDb;
-
-#undef M
-#undef H
-#undef dH
-}
-*/
-
-static const sph_u64 final_b[16] = {
-	SPH_C64(0xaaaaaaaaaaaaaaa0), SPH_C64(0xaaaaaaaaaaaaaaa1),
-	SPH_C64(0xaaaaaaaaaaaaaaa2), SPH_C64(0xaaaaaaaaaaaaaaa3),
-	SPH_C64(0xaaaaaaaaaaaaaaa4), SPH_C64(0xaaaaaaaaaaaaaaa5),
-	SPH_C64(0xaaaaaaaaaaaaaaa6), SPH_C64(0xaaaaaaaaaaaaaaa7),
-	SPH_C64(0xaaaaaaaaaaaaaaa8), SPH_C64(0xaaaaaaaaaaaaaaa9),
-	SPH_C64(0xaaaaaaaaaaaaaaaa), SPH_C64(0xaaaaaaaaaaaaaaab),
-	SPH_C64(0xaaaaaaaaaaaaaaac), SPH_C64(0xaaaaaaaaaaaaaaad),
-	SPH_C64(0xaaaaaaaaaaaaaaae), SPH_C64(0xaaaaaaaaaaaaaaaf)
-};
-
-
-#ifdef __cplusplus
-}
-#endif
--- a/algo/bmw/sse2/sph_bmw.h
+++ b/algo/bmw/sse2/sph_bmw.h
@@ -1,61 +0,0 @@
-/* $Id: sph_bmw.h 216 2010-06-08 09:46:57Z tp $ */
-/**
- * BMW interface. BMW (aka "Blue Midnight Wish") is a family of
- * functions which differ by their output size; this implementation
- * defines BMW for output sizes 224, 256, 384 and 512 bits.
- *
- * ==========================(LICENSE BEGIN)============================
- *
- * Copyright (c) 2007-2010  Projet RNRT SAPHIR
- * 
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
- * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * ===========================(LICENSE END)=============================
- *
- * @file     sph_bmw.h
- * @author   Thomas Pornin <thomas.pornin@cryptolog.com>
- */
-
-#ifndef SPH_BMW_H__
-#define SPH_BMW_H__
-
-#ifdef __cplusplus
-extern "C"{
-#endif
-
-#include <stddef.h>
-#include "sph_types.h"
-
-#define SPH_SIZE_bmw512   512
-
-typedef struct {
-#ifndef DOXYGEN_IGNORE
-	sph_u64 bmwH[16];
-#endif
-} sph_bmw_big_context;
-
-typedef sph_bmw_big_context sph_bmw512_context;
-
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif
--- a/algo/cryptonight/cryptolight.c
+++ b/algo/cryptonight/cryptolight.c
@@ -1,369 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-#include "algo-gate-api.h"
-
-#if defined(__arm__) || defined(_MSC_VER)
-#ifndef NOASM
-#define NOASM
-#endif
-#endif
-
-#include "crypto/oaes_lib.h"
-#include "crypto/c_keccak.h"
-#include "crypto/c_groestl.h"
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "crypto/int-util.h"
-#include "crypto/hash-ops.h"
-
-#if USE_INT128
-
-#if __GNUC__ == 4 && __GNUC_MINOR__ >= 4 && __GNUC_MINOR__ < 6
-typedef unsigned int uint128_t __attribute__ ((__mode__ (TI)));
-#elif defined (_MSC_VER)
-/* only for mingw64 on windows */
-#undef  USE_INT128
-#define USE_INT128 (0)
-#else
-typedef __uint128_t uint128_t;
-#endif
-
-#endif
-
-#define LITE 1
-#if LITE /* cryptonight-light */
-#define MEMORY (1 << 20)
-#define ITER   (1 << 19)
-#else
-#define MEMORY (1 << 21) /* 2 MiB */
-#define ITER   (1 << 20)
-#endif
-
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)
-
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-	union hash_state hs;
-	struct {
-		uint8_t k[64];
-		uint8_t init[INIT_SIZE_BYTE];
-	};
-};
-#pragma pack(pop)
-
-static void do_blake_hash(const void* input, size_t len, char* output) {
-	blake256_hash((uint8_t*)output, input, len);
-}
-
-static void do_groestl_hash(const void* input, size_t len, char* output) {
-	groestl(input, len * 8, (uint8_t*)output);
-}
-
-static void do_jh_hash(const void* input, size_t len, char* output) {
-	int r = jh_hash(HASH_SIZE * 8, input, 8 * len, (uint8_t*)output);
-	assert(likely(SUCCESS == r));
-}
-
-static void do_skein_hash(const void* input, size_t len, char* output) {
-	int r = skein_hash(8 * HASH_SIZE, input, 8 * len, (uint8_t*)output);
-	assert(likely(SKEIN_SUCCESS == r));
-}
-
-extern int aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#if !defined(_MSC_VER) && !defined(NOASM)
-extern int fast_aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int fast_aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#else
-#define fast_aesb_single_round     aesb_single_round
-#define fast_aesb_pseudo_round_mut aesb_pseudo_round_mut
-#endif
-
-#if defined(NOASM) || !defined(__x86_64__)
-static uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi) {
-	// multiplier   = ab = a * 2^32 + b
-	// multiplicand = cd = c * 2^32 + d
-	// ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
-	uint64_t a = hi_dword(multiplier);
-	uint64_t b = lo_dword(multiplier);
-	uint64_t c = hi_dword(multiplicand);
-	uint64_t d = lo_dword(multiplicand);
-
-	uint64_t ac = a * c;
-	uint64_t ad = a * d;
-	uint64_t bc = b * c;
-	uint64_t bd = b * d;
-
-	uint64_t adbc = ad + bc;
-	uint64_t adbc_carry = adbc < ad ? 1 : 0;
-
-	// multiplier * multiplicand = product_hi * 2^64 + product_lo
-	uint64_t product_lo = bd + (adbc << 32);
-	uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
-	*product_hi = ac + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
-	assert(ac <= *product_hi);
-
-	return product_lo;
-}
-#else
-extern uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi);
-#endif
-
-static void (* const extra_hashes[4])(const void *, size_t, char *) = {
-		do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash
-};
-
-
-static inline size_t e2i(const uint8_t* a) {
-#if !LITE
-	return ((uint32_t *)a)[0] & 0x1FFFF0;
-#else
-	return ((uint32_t *)a)[0] & 0xFFFF0;
-#endif
-}
-
-static inline void mul_sum_xor_dst(const uint8_t* a, uint8_t* c, uint8_t* dst) {
-	uint64_t hi, lo = mul128(((uint64_t*) a)[0], ((uint64_t*) dst)[0], &hi) + ((uint64_t*) c)[1];
-	hi += ((uint64_t*) c)[0];
-
-	((uint64_t*) c)[0] = ((uint64_t*) dst)[0] ^ hi;
-	((uint64_t*) c)[1] = ((uint64_t*) dst)[1] ^ lo;
-	((uint64_t*) dst)[0] = hi;
-	((uint64_t*) dst)[1] = lo;
-}
-
-static inline void xor_blocks(uint8_t* a, const uint8_t* b) {
-#if USE_INT128
-	*((uint128_t*) a) ^= *((uint128_t*) b);
-#else
-	((uint64_t*) a)[0] ^= ((uint64_t*) b)[0];
-	((uint64_t*) a)[1] ^= ((uint64_t*) b)[1];
-#endif
-}
-
-static inline void xor_blocks_dst(const uint8_t* a, const uint8_t* b, uint8_t* dst) {
-#if USE_INT128
-	*((uint128_t*) dst) = *((uint128_t*) a) ^ *((uint128_t*) b);
-#else
-	((uint64_t*) dst)[0] = ((uint64_t*) a)[0] ^ ((uint64_t*) b)[0];
-	((uint64_t*) dst)[1] = ((uint64_t*) a)[1] ^ ((uint64_t*) b)[1];
-#endif
-}
-
-struct cryptonight_ctx {
-	uint8_t _ALIGN(16) long_state[MEMORY];
-	union cn_slow_hash_state state;
-	uint8_t _ALIGN(16) text[INIT_SIZE_BYTE];
-	uint8_t _ALIGN(16) a[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) b[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) c[AES_BLOCK_SIZE];
-	oaes_ctx* aes_ctx;
-};
-
-static void cryptolight_hash_ctx(void* output, const void* input, int len, struct cryptonight_ctx* ctx)
-{
-        len = 76;
-	hash_process(&ctx->state.hs, (const uint8_t*) input, len);
-	ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
-	size_t i, j;
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx->aes_ctx, ctx->state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 0], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 1], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 2], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 3], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 4], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 5], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 6], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 7], ctx->aes_ctx->key->exp_data);
-		memcpy(&ctx->long_state[i], ctx->text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx->state.k[0], &ctx->state.k[32], ctx->a);
-	xor_blocks_dst(&ctx->state.k[16], &ctx->state.k[48], ctx->b);
-
-	for (i = 0; likely(i < ITER / 4); ++i) {
-		/* Dependency chain: address -> read value ------+
-		 * written value <-+ hard function (AES or MUL) <+
-		 * next address  <-+
-		 */
-		/* Iteration 1 */
-		j = e2i(ctx->a);
-		aesb_single_round(&ctx->long_state[j], ctx->c, ctx->a);
-		xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j]);
-		/* Iteration 2 */
-		mul_sum_xor_dst(ctx->c, ctx->a, &ctx->long_state[e2i(ctx->c)]);
-		/* Iteration 3 */
-		j = e2i(ctx->a);
-		aesb_single_round(&ctx->long_state[j], ctx->b, ctx->a);
-		xor_blocks_dst(ctx->b, ctx->c, &ctx->long_state[j]);
-		/* Iteration 4 */
-		mul_sum_xor_dst(ctx->b, ctx->a, &ctx->long_state[e2i(ctx->b)]);
-	}
-
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		xor_blocks(&ctx->text[0 * AES_BLOCK_SIZE], &ctx->long_state[i + 0 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[0 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[1 * AES_BLOCK_SIZE], &ctx->long_state[i + 1 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[1 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[2 * AES_BLOCK_SIZE], &ctx->long_state[i + 2 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[2 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[3 * AES_BLOCK_SIZE], &ctx->long_state[i + 3 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[3 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[4 * AES_BLOCK_SIZE], &ctx->long_state[i + 4 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[4 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[5 * AES_BLOCK_SIZE], &ctx->long_state[i + 5 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[5 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[6 * AES_BLOCK_SIZE], &ctx->long_state[i + 6 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[6 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[7 * AES_BLOCK_SIZE], &ctx->long_state[i + 7 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[7 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-	}
-	memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
-	hash_permutation(&ctx->state.hs);
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
-	oaes_free((OAES_CTX **) &ctx->aes_ctx);
-}
-
-void cryptolight_hash(void* output, const void* input, int len) {
-	struct cryptonight_ctx *ctx = (struct cryptonight_ctx*)malloc(sizeof(struct cryptonight_ctx));
-	cryptolight_hash_ctx(output, input, len, ctx);
-	free(ctx);
-}
-
-#if defined(__AES__)
-
-static void cryptolight_hash_ctx_aes_ni(void* output, const void* input,
-                       int len, struct cryptonight_ctx* ctx)
-{
-	hash_process(&ctx->state.hs, (const uint8_t*)input, len);
-	ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
-	size_t i, j;
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx->aes_ctx, ctx->state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 0], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 1], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 2], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 3], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 4], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 5], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 6], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 7], ctx->aes_ctx->key->exp_data);
-		memcpy(&ctx->long_state[i], ctx->text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx->state.k[0], &ctx->state.k[32], ctx->a);
-	xor_blocks_dst(&ctx->state.k[16], &ctx->state.k[48], ctx->b);
-
-	for (i = 0; likely(i < ITER / 4); ++i) {
-		/* Dependency chain: address -> read value ------+
-		 * written value <-+ hard function (AES or MUL) <+
-		 * next address  <-+
-		 */
-		/* Iteration 1 */
-		j = e2i(ctx->a);
-		fast_aesb_single_round(&ctx->long_state[j], ctx->c, ctx->a);
-		xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j]);
-		/* Iteration 2 */
-		mul_sum_xor_dst(ctx->c, ctx->a, &ctx->long_state[e2i(ctx->c)]);
-		/* Iteration 3 */
-		j = e2i(ctx->a);
-		fast_aesb_single_round(&ctx->long_state[j], ctx->b, ctx->a);
-		xor_blocks_dst(ctx->b, ctx->c, &ctx->long_state[j]);
-		/* Iteration 4 */
-		mul_sum_xor_dst(ctx->b, ctx->a, &ctx->long_state[e2i(ctx->b)]);
-	}
-
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		xor_blocks(&ctx->text[0 * AES_BLOCK_SIZE], &ctx->long_state[i + 0 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[0 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[1 * AES_BLOCK_SIZE], &ctx->long_state[i + 1 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[1 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[2 * AES_BLOCK_SIZE], &ctx->long_state[i + 2 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[2 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[3 * AES_BLOCK_SIZE], &ctx->long_state[i + 3 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[3 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[4 * AES_BLOCK_SIZE], &ctx->long_state[i + 4 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[4 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[5 * AES_BLOCK_SIZE], &ctx->long_state[i + 5 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[5 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[6 * AES_BLOCK_SIZE], &ctx->long_state[i + 6 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[6 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[7 * AES_BLOCK_SIZE], &ctx->long_state[i + 7 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[7 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-	}
-	memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
-	hash_permutation(&ctx->state.hs);
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
-	oaes_free((OAES_CTX **) &ctx->aes_ctx);
-}
-
-#endif
-
-int scanhash_cryptolight( struct work *work,
-		uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr)
-{
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
-	uint32_t *nonceptr = (uint32_t*) (((char*)pdata) + 39);
-	uint32_t n = *nonceptr - 1;
-	const uint32_t first_nonce = n + 1;
-	//const uint32_t Htarg = ptarget[7];
-	uint32_t _ALIGN(32) hash[HASH_SIZE / 4];
-   int thr_id = mythr->id;
-
-	struct cryptonight_ctx *ctx = (struct cryptonight_ctx*)malloc(sizeof(struct cryptonight_ctx));
-
-#if defined(__AES__)
-		do {
-			*nonceptr = ++n;
-			cryptolight_hash_ctx_aes_ni(hash, pdata, 76, ctx);
-			if (unlikely(hash[7] < ptarget[7])) {
-				*hashes_done = n - first_nonce + 1;
-				free(ctx);
-				return true;
-			}
-		} while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-#else
-		do {
-			*nonceptr = ++n;
-			cryptolight_hash_ctx(hash, pdata, 76, ctx);
-			if (unlikely(hash[7] < ptarget[7])) {
-				*hashes_done = n - first_nonce + 1;
-				free(ctx);
-				return true;
-			}
-		} while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-#endif
-	free(ctx);
-	*hashes_done = n - first_nonce + 1;
-	return 0;
-}
-
-bool register_cryptolight_algo( algo_gate_t* gate )
-{
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash  = (void*)&scanhash_cryptolight;
-  gate->hash      = (void*)&cryptolight_hash;
-  gate->hash_suw  = (void*)&cryptolight_hash; 
-  gate->get_max64 = (void*)&get_max64_0x40LL;
-  return true;
-};
-
--- a/algo/cryptonight/cryptonight-aesni.c
+++ b/algo/cryptonight/cryptonight-aesni.c
@@ -1,357 +0,0 @@
-#if defined(__AES__)
-
-#include <x86intrin.h>
-#include <memory.h>
-#include "cryptonight.h"
-#include "miner.h"
-#include "crypto/c_keccak.h"
-#include <immintrin.h>
-
-static inline void ExpandAESKey256_sub1(__m128i *tmp1, __m128i *tmp2)
-{
-	__m128i tmp4;
-	*tmp2 = _mm_shuffle_epi32(*tmp2, 0xFF);
-	tmp4 = _mm_slli_si128(*tmp1, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	*tmp1 = _mm_xor_si128(*tmp1, *tmp2);
-}
-
-static inline void ExpandAESKey256_sub2(__m128i *tmp1, __m128i *tmp3)
-{
-	__m128i tmp2, tmp4;
-	
-	tmp4 = _mm_aeskeygenassist_si128(*tmp1, 0x00);
-	tmp2 = _mm_shuffle_epi32(tmp4, 0xAA);
-	tmp4 = _mm_slli_si128(*tmp3, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp2);
-}
-
-// Special thanks to Intel for helping me
-// with ExpandAESKey256() and its subroutines
-static inline void ExpandAESKey256(char *keybuf)
-{
-	__m128i tmp1, tmp2, tmp3, *keys;
-	
-	keys = (__m128i *)keybuf;
-	
-	tmp1 = _mm_load_si128((__m128i *)keybuf);
-	tmp3 = _mm_load_si128((__m128i *)(keybuf+0x10));
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x01);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[2] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[3] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x02);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[4] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[5] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x04);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[6] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[7] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x08);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[8] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[9] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x10);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[10] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[11] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x20);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[12] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[13] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x40);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[14] = tmp1;
-}
-
-// align to 64 byte cache line
-typedef struct 
-{
-    uint8_t long_state[MEMORY] __attribute((aligned(64)));
-    union cn_slow_hash_state state;
-    uint8_t text[INIT_SIZE_BYTE] __attribute((aligned(64)));
-    uint64_t a[AES_BLOCK_SIZE >> 3] __attribute__((aligned(64)));
-    uint64_t b[AES_BLOCK_SIZE >> 3] __attribute__((aligned(64)));
-    uint8_t c[AES_BLOCK_SIZE] __attribute__((aligned(64)));
-} cryptonight_ctx;
-
-static __thread cryptonight_ctx ctx;
-
-void cryptonight_hash_aes( void *restrict output, const void *input, int len )
-{
-    uint8_t ExpandedKey[256] __attribute__((aligned(64)));
-    __m128i *longoutput, *expkey, *xmminput;
-    size_t i, j;
-    
-    keccak( (const uint8_t*)input, 76, (char*)&ctx.state.hs.b, 200 );
-
-    if ( cryptonightV7 && len < 43 )
-      return;
-
-    const uint64_t tweak = cryptonightV7 
-                         ? *((const uint64_t*) (((const uint8_t*)input) + 35))
-                           ^ ctx.state.hs.w[24] : 0; 
-
-    memcpy( ExpandedKey, ctx.state.hs.b, AES_KEY_SIZE );
-    ExpandAESKey256( ExpandedKey );
-    memcpy( ctx.text, ctx.state.init, INIT_SIZE_BYTE );
-    
-    longoutput = (__m128i*)ctx.long_state;
-    xmminput   = (__m128i*)ctx.text;
-    expkey     = (__m128i*)ExpandedKey;
-    
-    // prefetch expkey, xmminput and enough longoutput for 4 iterations
-    _mm_prefetch( xmminput,     _MM_HINT_T0 );
-    _mm_prefetch( xmminput + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey,     _MM_HINT_T0 );
-    _mm_prefetch( expkey + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey + 8, _MM_HINT_T0 );
-    for ( i = 0; i < 64; i += 16 )
-    {
-        __builtin_prefetch( longoutput + i,      1, 0 );
-        __builtin_prefetch( longoutput + i +  4, 1, 0 );
-        __builtin_prefetch( longoutput + i +  8, 1, 0 );
-        __builtin_prefetch( longoutput + i + 12, 1, 0 );
-    }
-
-    // n-4 iterations
-    for ( i = 0; likely( i < MEMORY_M128I - 4*INIT_SIZE_M128I );
-                         i += INIT_SIZE_M128I )
-    {
-        // prefetch 4 iterations ahead.
-        __builtin_prefetch( longoutput + i + 64, 1, 0 );
-        __builtin_prefetch( longoutput + i + 68, 1, 0 );
-
-	for ( j = 0; j < 10; j++ )
-	{
-		xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-		xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-		xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-		xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-		xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-		xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-		xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-		xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-	}
-	_mm_store_si128( &( longoutput[i  ] ), xmminput[0] );
-	_mm_store_si128( &( longoutput[i+1] ), xmminput[1] );
-	_mm_store_si128( &( longoutput[i+2] ), xmminput[2] );
-	_mm_store_si128( &( longoutput[i+3] ), xmminput[3] );
-	_mm_store_si128( &( longoutput[i+4] ), xmminput[4] );
-	_mm_store_si128( &( longoutput[i+5] ), xmminput[5] );
-	_mm_store_si128( &( longoutput[i+6] ), xmminput[6] );
-	_mm_store_si128( &( longoutput[i+7] ), xmminput[7] );
-    }
-    // last 4 iterations
-    for ( ; likely( i < MEMORY_M128I ); i += INIT_SIZE_M128I )
-    {
-        for ( j = 0; j < 10; j++ )
-        {
-                xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-                xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-                xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-                xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-                xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-                xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-                xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-                xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-        _mm_store_si128( &( longoutput[i  ] ), xmminput[0] );
-        _mm_store_si128( &( longoutput[i+1] ), xmminput[1] );
-        _mm_store_si128( &( longoutput[i+2] ), xmminput[2] );
-        _mm_store_si128( &( longoutput[i+3] ), xmminput[3] );
-        _mm_store_si128( &( longoutput[i+4] ), xmminput[4] );
-        _mm_store_si128( &( longoutput[i+5] ), xmminput[5] );
-        _mm_store_si128( &( longoutput[i+6] ), xmminput[6] );
-        _mm_store_si128( &( longoutput[i+7] ), xmminput[7] );
-    }
-
-    ctx.a[0] = ((uint64_t *)ctx.state.k)[0] ^ ((uint64_t *)ctx.state.k)[4];
-    ctx.b[0] = ((uint64_t *)ctx.state.k)[2] ^ ((uint64_t *)ctx.state.k)[6];
-    ctx.a[1] = ((uint64_t *)ctx.state.k)[1] ^ ((uint64_t *)ctx.state.k)[5];
-    ctx.b[1] = ((uint64_t *)ctx.state.k)[3] ^ ((uint64_t *)ctx.state.k)[7];
-
-    uint64_t a[2] __attribute((aligned(16))),
-             b[2] __attribute((aligned(16))),
-             c[2] __attribute((aligned(16)));
-    a[0] = ctx.a[0];
-    a[1] = ctx.a[1];
-    __m128i b_x = _mm_load_si128( (__m128i*)ctx.b );
-    __m128i a_x = _mm_load_si128( (__m128i*)a );
-    __m128i* lsa = (__m128i*)&ctx.long_state[ a[0] & 0x1FFFF0 ];
-    __m128i c_x = _mm_load_si128( lsa );
-    uint64_t *nextblock;
-    uint64_t hi, lo;
-
-    // n-1 iterations
-    for( i = 0; __builtin_expect( i < 0x7ffff, 1 ); i++ )
-    {	  
-	c_x = _mm_aesenc_si128( c_x, a_x );
-	_mm_store_si128( (__m128i*)c, c_x );
-        b_x = _mm_xor_si128( b_x, c_x );
-        nextblock = (uint64_t *)&ctx.long_state[c[0] & 0x1FFFF0];
-        _mm_store_si128( lsa, b_x );
-
-        if ( cryptonightV7 )
-        {
-           const uint8_t tmp = ( (const uint8_t*)(lsa) )[11];
-           const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-           ((uint8_t*)(lsa))[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-        } 
-
-	b[0] = nextblock[0];
-	b[1] = nextblock[1];
-
-        // hi,lo = 64bit x 64bit multiply of c[0] and b[0]
-	__asm__( "mulq %3\n\t"
-	         : "=d" ( hi ),
-	           "=a" ( lo )
-	         : "%a" ( c[0] ),
-	           "rm" ( b[0] )
-		 : "cc" );
-
-        b_x = c_x;
-
-        a[0] += hi;
-        a[1] += lo;
-        nextblock[0] = a[0];
-        nextblock[1] = cryptonightV7 ? a[1] ^ tweak : a[1];
-        a[0] ^= b[0];
-        a[1] ^= b[1];
-
-        lsa = (__m128i*)&ctx.long_state[ a[0] & 0x1FFFF0 ];
-        a_x = _mm_load_si128( (__m128i*)a );
-        c_x = _mm_load_si128( lsa );
-    }
-    // abreviated nth iteration
-    c_x = _mm_aesenc_si128( c_x, a_x );
-    _mm_store_si128( (__m128i*)c, c_x );
-    b_x = _mm_xor_si128( b_x, c_x );
-    nextblock = (uint64_t *)&ctx.long_state[c[0] & 0x1FFFF0];
-    _mm_store_si128( lsa, b_x );
-
-    if ( cryptonightV7 )
-    {
-       const uint8_t tmp = ( (const uint8_t*)(lsa) )[11];
-       const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-       ((uint8_t*)(lsa))[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-    }
-
-    b[0] = nextblock[0];
-    b[1] = nextblock[1];
-
-    __asm__( "mulq %3\n\t"
-             : "=d" ( hi ),
-               "=a" ( lo )
-             : "%a" ( c[0] ),
-               "rm" ( b[0] )
-             : "cc" );
-
-    a[0] += hi;
-    a[1] += lo;
-    nextblock[0] = a[0];
-    nextblock[1] = cryptonightV7 ? a[1] ^ tweak : a[1];
-    a[0] ^= b[0];
-    a[1] ^= b[1];
-
-    memcpy( ExpandedKey, &ctx.state.hs.b[32], AES_KEY_SIZE );
-    ExpandAESKey256( ExpandedKey );
-    memcpy( ctx.text, ctx.state.init, INIT_SIZE_BYTE );
-    
-    // prefetch expkey, all of xmminput and enough longoutput for 4 loops
-    _mm_prefetch( xmminput,     _MM_HINT_T0 );
-    _mm_prefetch( xmminput + 4, _MM_HINT_T0 );
-    for ( i = 0; i < 64; i += 16 )
-    {
-       _mm_prefetch( longoutput + i,      _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i +  4, _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i +  8, _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i + 12, _MM_HINT_T0 );
-    }
-    _mm_prefetch( expkey,     _MM_HINT_T0 );
-    _mm_prefetch( expkey + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey + 8, _MM_HINT_T0 );
-
-    // n-4 iterations
-    for ( i = 0; likely( i < MEMORY_M128I - 4*INIT_SIZE_M128I );
-                         i += INIT_SIZE_M128I )
-    {
-        // stay 4 iterations ahead.
-        _mm_prefetch( longoutput + i + 64, _MM_HINT_T0 );
-        _mm_prefetch( longoutput + i + 68, _MM_HINT_T0 );
-
-        xmminput[0] = _mm_xor_si128( longoutput[i  ], xmminput[0] );
-        xmminput[1] = _mm_xor_si128( longoutput[i+1], xmminput[1] );
-        xmminput[2] = _mm_xor_si128( longoutput[i+2], xmminput[2] );
-        xmminput[3] = _mm_xor_si128( longoutput[i+3], xmminput[3] );
-        xmminput[4] = _mm_xor_si128( longoutput[i+4], xmminput[4] );
-        xmminput[5] = _mm_xor_si128( longoutput[i+5], xmminput[5] );
-        xmminput[6] = _mm_xor_si128( longoutput[i+6], xmminput[6] );
-        xmminput[7] = _mm_xor_si128( longoutput[i+7], xmminput[7] );
-		
-        for( j = 0; j < 10; j++ )
-        {
-            xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-	    xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-	    xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-	    xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-	    xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-	    xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-	    xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-	    xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-    }
-    // last 4 iterations 
-    for ( ; likely( i < MEMORY_M128I ); i += INIT_SIZE_M128I )
-    {
-        xmminput[0] = _mm_xor_si128( longoutput[i  ], xmminput[0] );
-        xmminput[1] = _mm_xor_si128( longoutput[i+1], xmminput[1] );
-        xmminput[2] = _mm_xor_si128( longoutput[i+2], xmminput[2] );
-        xmminput[3] = _mm_xor_si128( longoutput[i+3], xmminput[3] );
-        xmminput[4] = _mm_xor_si128( longoutput[i+4], xmminput[4] );
-        xmminput[5] = _mm_xor_si128( longoutput[i+5], xmminput[5] );
-        xmminput[6] = _mm_xor_si128( longoutput[i+6], xmminput[6] );
-        xmminput[7] = _mm_xor_si128( longoutput[i+7], xmminput[7] );
-
-        for( j = 0; j < 10; j++ )
-        {
-            xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-            xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-            xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-            xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-            xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-            xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-            xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-            xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-    }
-
-    memcpy( ctx.state.init, ctx.text, INIT_SIZE_BYTE);
-    keccakf( (uint64_t*)&ctx.state.hs.w, 24 );
-    extra_hashes[ctx.state.hs.b[0] & 3](&ctx.state, 200, output);
-
-}
-#endif
--- a/algo/cryptonight/cryptonight-common.c
+++ b/algo/cryptonight/cryptonight-common.c
@@ -1,129 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-// Modified for CPUminer by Lucas Jones
-
-#include "cpuminer-config.h"
-#include "algo-gate-api.h"
-
-#if defined(__AES__)
-  #include "algo/groestl/aes_ni/hash-groestl256.h"
-#else
-#include "crypto/c_groestl.h"
-#endif
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "cryptonight.h"
-
-/*
-#if defined __unix__ && (!defined __APPLE__)
-#include <sys/mman.h>
-#elif defined _WIN32
-#include <windows.h>
-#endif
-*/
-
-void do_blake_hash(const void* input, size_t len, char* output) {
-    blake256_hash((uint8_t*)output, input, len);
-}
-
-void do_groestl_hash(const void* input, size_t len, char* output) {
-#if defined(__AES__)
-    hashState_groestl256 ctx;
-    init_groestl256( &ctx, 32 );
-    update_and_final_groestl256( &ctx, output, input, len * 8 );
-#else
-    groestl(input, len * 8, (uint8_t*)output);
-#endif
-}
-
-void do_jh_hash(const void* input, size_t len, char* output) {
-    jh_hash(32 * 8, input, 8 * len, (uint8_t*)output);
-}
-
-void do_skein_hash(const void* input, size_t len, char* output) {
-    skein_hash(8 * 32, input, 8 * len, (uint8_t*)output);
-}
-
-void (* const extra_hashes[4])( const void *, size_t, char *) =
-    { do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash };
-
-void cryptonight_hash( void *restrict output, const void *input, int len )
-{
-#if defined(__AES__)
-  cryptonight_hash_aes( output, input, len );
-#else
-  cryptonight_hash_ctx ( output, input, len );
-#endif
-}
-
-void cryptonight_hash_suw( void *restrict output, const void *input )
-{
-#if defined(__AES__)
-  cryptonight_hash_aes( output, input, 76 );
-#else
-  cryptonight_hash_ctx ( output, input, 76 );
-#endif
-}
-
-bool cryptonightV7 = false;
-
-int scanhash_cryptonight( struct work *work, uint32_t max_nonce,
-                   uint64_t *hashes_done, struct thr_info *mythr )
- {
-    uint32_t *pdata = work->data;
-    uint32_t *ptarget = work->target;
-    int thr_id = mythr->id;
-
-    uint32_t *nonceptr = (uint32_t*) (((char*)pdata) + 39);
-    uint32_t n = *nonceptr - 1;
-    const uint32_t first_nonce = n + 1;
-    const uint32_t Htarg = ptarget[7];
-    uint32_t hash[32 / 4] __attribute__((aligned(32)));
-
-//    if (  (  cryptonightV7 && ( *(uint8_t*)pdata <  7 ) )
-//       || ( !cryptonightV7 && ( *(uint8_t*)pdata == 7 ) ) )
-//          applog(LOG_WARNING,"Cryptonight variant mismatch, shares may be rejected.");
-
-    do
-    {
-       *nonceptr = ++n;
-       cryptonight_hash( hash, pdata, 76 );
-       if (unlikely( hash[7] < Htarg ))
-       {
-           *hashes_done = n - first_nonce + 1;
-//           work_set_target_ratio( work, hash );
-	   return true;
-       }
-    } while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-    
-    *hashes_done = n - first_nonce + 1;
-    return 0;
-}
-
-bool register_cryptonight_algo( algo_gate_t* gate )
-{
-  cryptonightV7 = false;
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash         = (void*)&scanhash_cryptonight;
-  gate->hash             = (void*)&cryptonight_hash;
-  gate->hash_suw         = (void*)&cryptonight_hash_suw;  
-  gate->get_max64        = (void*)&get_max64_0x40LL;
-  return true;
-};
-
-bool register_cryptonightv7_algo( algo_gate_t* gate )
-{
-  cryptonightV7 = true;
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash      = (void*)&scanhash_cryptonight;
-  gate->hash          = (void*)&cryptonight_hash;
-  gate->hash_suw      = (void*)&cryptonight_hash_suw;
-  gate->get_max64     = (void*)&get_max64_0x40LL;
-  return true;
-};
-
--- a/algo/cryptonight/cryptonight.c
+++ b/algo/cryptonight/cryptonight.c
@@ -1,310 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-// Modified for CPUminer by Lucas Jones
-
-#include "miner.h"
-#include <memory.h>
-
-#if defined(__arm__) || defined(_MSC_VER)
-#ifndef NOASM
-#define NOASM
-#endif
-#endif
-
-#include "crypto/oaes_lib.h"
-#include "crypto/c_keccak.h"
-#include "crypto/c_groestl.h"
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "crypto/int-util.h"
-//#include "crypto/hash-ops.h"
-#include "cryptonight.h"
-
-#if USE_INT128
-
-#if __GNUC__ == 4 && __GNUC_MINOR__ >= 4 && __GNUC_MINOR__ < 6
-typedef unsigned int uint128_t __attribute__ ((__mode__ (TI)));
-#elif defined (_MSC_VER)
-/* only for mingw64 on windows */
-#undef  USE_INT128
-#define USE_INT128 (0)
-#else
-typedef __uint128_t uint128_t;
-#endif
-
-#endif
-
-#define LITE 0
-#if LITE /* cryptonight-light */
-#define MEMORY (1 << 20)
-#define ITER   (1 << 19)
-#else
-#define MEMORY (1 << 21) /* 2 MiB */
-#define ITER   (1 << 20)
-#endif
-
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)
-
-/*
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-	union hash_state hs;
-	struct {
-		uint8_t k[64];
-		uint8_t init[INIT_SIZE_BYTE];
-	};
-};
-#pragma pack(pop)
-
-static void do_blake_hash(const void* input, size_t len, char* output) {
-	blake256_hash((uint8_t*)output, input, len);
-}
-
-static void do_groestl_hash(const void* input, size_t len, char* output) {
-	groestl(input, len * 8, (uint8_t*)output);
-}
-
-static void do_jh_hash(const void* input, size_t len, char* output) {
-	int r = jh_hash(HASH_SIZE * 8, input, 8 * len, (uint8_t*)output);
-	assert(likely(SUCCESS == r));
-}
-
-static void do_skein_hash(const void* input, size_t len, char* output) {
-	int r = skein_hash(8 * HASH_SIZE, input, 8 * len, (uint8_t*)output);
-	assert(likely(SKEIN_SUCCESS == r));
-}
-*/
-
-extern int aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#if !defined(_MSC_VER) && !defined(NOASM)
-extern int fast_aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int fast_aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#else
-#define fast_aesb_single_round     aesb_single_round
-#define fast_aesb_pseudo_round_mut aesb_pseudo_round_mut
-#endif
-
-
-#if defined(NOASM) || !defined(__x86_64__)
-static uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi) {
-	// multiplier   = ab = a * 2^32 + b
-	// multiplicand = cd = c * 2^32 + d
-	// ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
-	uint64_t a = hi_dword(multiplier);
-	uint64_t b = lo_dword(multiplier);
-	uint64_t c = hi_dword(multiplicand);
-	uint64_t d = lo_dword(multiplicand);
-
-	uint64_t ac = a * c;
-	uint64_t ad = a * d;
-	uint64_t bc = b * c;
-	uint64_t bd = b * d;
-
-	uint64_t adbc = ad + bc;
-	uint64_t adbc_carry = adbc < ad ? 1 : 0;
-
-	// multiplier * multiplicand = product_hi * 2^64 + product_lo
-	uint64_t product_lo = bd + (adbc << 32);
-	uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
-	*product_hi = ac + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
-	assert(ac <= *product_hi);
-
-	return product_lo;
-}
-#else
-extern uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi);
-#endif
-
-/*
-static void (* const extra_hashes[4])(const void *, size_t, char *) = {
-		do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash
-};
-*/
-
-static inline size_t e2i(const uint8_t* a) {
-#if !LITE
-	return ((uint32_t *)a)[0] & 0x1FFFF0;
-#else
-	return ((uint32_t *)a)[0] & 0xFFFF0;
-#endif
-}
-
-static inline void mul_sum_xor_dst( const uint8_t* a, uint8_t* c, uint8_t* dst, 
-         const uint64_t tweak )
-{
-	uint64_t hi, lo = mul128(((uint64_t*) a)[0], ((uint64_t*) dst)[0], &hi) + ((uint64_t*) c)[1];
-	hi += ((uint64_t*) c)[0];
-
-	((uint64_t*) c)[0] = ((uint64_t*) dst)[0] ^ hi;
-	((uint64_t*) c)[1] = ((uint64_t*) dst)[1] ^ lo;
-	((uint64_t*) dst)[0] = hi;
-	((uint64_t*) dst)[1] = cryptonightV7 ? lo ^ tweak : lo;
-}
-
-static inline void xor_blocks(uint8_t* a, const uint8_t* b) {
-#if USE_INT128
-	*((uint128_t*) a) ^= *((uint128_t*) b);
-#else
-	((uint64_t*) a)[0] ^= ((uint64_t*) b)[0];
-	((uint64_t*) a)[1] ^= ((uint64_t*) b)[1];
-#endif
-}
-
-static inline void xor_blocks_dst(const uint8_t* a, const uint8_t* b, uint8_t* dst) {
-#if USE_INT128
-	*((uint128_t*) dst) = *((uint128_t*) a) ^ *((uint128_t*) b);
-#else
-	((uint64_t*) dst)[0] = ((uint64_t*) a)[0] ^ ((uint64_t*) b)[0];
-	((uint64_t*) dst)[1] = ((uint64_t*) a)[1] ^ ((uint64_t*) b)[1];
-#endif
-}
-
-typedef struct {
-	uint8_t _ALIGN(16) long_state[MEMORY];
-	union cn_slow_hash_state state;
-	uint8_t _ALIGN(16) text[INIT_SIZE_BYTE];
-	uint8_t _ALIGN(16) a[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) b[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) c[AES_BLOCK_SIZE];
-	oaes_ctx* aes_ctx;
-} cryptonight_ctx;
-
-static __thread cryptonight_ctx ctx;
-
-void cryptonight_hash_ctx(void* output, const void* input, int len)
-{
-//    hash_process(&ctx.state.hs, (const uint8_t*) input, len);
-    keccak( (const uint8_t*)input, 76, (char*)&ctx.state.hs.b, 200 );
-
-    if ( cryptonightV7 && len < 43 )
-      return;
-    const uint64_t tweak = cryptonightV7
-                         ? *((const uint64_t*) (((const uint8_t*)input) + 35))
-                           ^ ctx.state.hs.w[24] : 0;
-
-    ctx.aes_ctx = (oaes_ctx*) oaes_alloc();
-
-    __builtin_prefetch( ctx.text,             0, 3 );
-    __builtin_prefetch( ctx.text       +  64, 0, 3 );
-    __builtin_prefetch( ctx.long_state,       1, 0 );
-    __builtin_prefetch( ctx.long_state +  64, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 128, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 192, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 256, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 320, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 384, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 448, 1, 0 );
-
-	size_t i, j;
-	memcpy(ctx.text, ctx.state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx.aes_ctx, ctx.state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-
-    __builtin_prefetch( ctx.long_state + i + 512, 1, 0 );
-    __builtin_prefetch( ctx.long_state + i + 576, 1, 0 );
-
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 0], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 1], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 2], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 3], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 4], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 5], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 6], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 7], ctx.aes_ctx->key->exp_data);
-		memcpy(&ctx.long_state[i], ctx.text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx.state.k[0], &ctx.state.k[32], ctx.a);
-	xor_blocks_dst(&ctx.state.k[16], &ctx.state.k[48], ctx.b);
-
-	for (i = 0; likely(i < ITER / 4); ++i)
-        {
-           /* Dependency chain: address -> read value ------+
-            * written value <-+ hard function (AES or MUL) <+
-            * next address  <-+
-            */
-           /* Iteration 1 */
-           j = e2i(ctx.a);
-           aesb_single_round(&ctx.long_state[j], ctx.c, ctx.a);
-           xor_blocks_dst(ctx.c, ctx.b, &ctx.long_state[j]);
-
-           if ( cryptonightV7 )
-           {
-              uint8_t *lsa = (uint8_t*)&ctx.long_state[((uint64_t *)(ctx.a))[0] & 0x1FFFF0];
-              const uint8_t tmp = lsa[11];
-              const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-              lsa[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-           }
-
-           /* Iteration 2 */
-           mul_sum_xor_dst(ctx.c, ctx.a, &ctx.long_state[e2i(ctx.c)], tweak );
-
-           /* Iteration 3 */
-           j = e2i(ctx.a);
-           aesb_single_round(&ctx.long_state[j], ctx.b, ctx.a);
-           xor_blocks_dst(ctx.b, ctx.c, &ctx.long_state[j]);
-
-           if ( cryptonightV7 )
-           {
-              uint8_t *lsa = (uint8_t*)&ctx.long_state[((uint64_t *)(ctx.a))[0] & 0x1FFFF0];
-              const uint8_t tmp = lsa[11];
-              const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-              lsa[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-           }
-
-           /* Iteration 4 */
-           mul_sum_xor_dst(ctx.b, ctx.a, &ctx.long_state[e2i(ctx.b)], tweak );
-
-	}
-
-    __builtin_prefetch( ctx.text,             0, 3 );
-    __builtin_prefetch( ctx.text       +  64, 0, 3 );
-    __builtin_prefetch( ctx.long_state,       1, 0 );
-    __builtin_prefetch( ctx.long_state +  64, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 128, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 192, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 256, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 320, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 384, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 448, 1, 0 );
-
-	memcpy(ctx.text, ctx.state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx.aes_ctx, &ctx.state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-
-    __builtin_prefetch( ctx.long_state + i + 512, 1, 0 );
-    __builtin_prefetch( ctx.long_state + i + 576, 1, 0 );
-
-		xor_blocks(&ctx.text[0 * AES_BLOCK_SIZE], &ctx.long_state[i + 0 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[0 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[1 * AES_BLOCK_SIZE], &ctx.long_state[i + 1 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[1 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[2 * AES_BLOCK_SIZE], &ctx.long_state[i + 2 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[2 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[3 * AES_BLOCK_SIZE], &ctx.long_state[i + 3 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[3 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[4 * AES_BLOCK_SIZE], &ctx.long_state[i + 4 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[4 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[5 * AES_BLOCK_SIZE], &ctx.long_state[i + 5 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[5 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[6 * AES_BLOCK_SIZE], &ctx.long_state[i + 6 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[6 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[7 * AES_BLOCK_SIZE], &ctx.long_state[i + 7 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[7 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-	}
-	memcpy(ctx.state.init, ctx.text, INIT_SIZE_BYTE);
-//	hash_permutation(&ctx.state.hs);
-        keccakf( (uint64_t*)&ctx.state.hs.w, 24 );
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx.state.hs.b[0] & 3](&ctx.state, 200, output);
-	oaes_free((OAES_CTX **) &ctx.aes_ctx);
-}
-
--- a/algo/cryptonight/cryptonight.h
+++ b/algo/cryptonight/cryptonight.h
@@ -1,51 +0,0 @@
-#ifndef __CRYPTONIGHT_H_INCLUDED
-#define __CRYPTONIGHT_H_INCLUDED
-
-#include <stddef.h>
-#include "crypto/oaes_lib.h"
-#include "miner.h"
-
-#define MEMORY         (1 << 21) /* 2 MiB */
-#define MEMORY_M128I   (MEMORY >> 4) // 2 MiB / 16 = 128 ki * __m128i
-#define ITER           (1 << 20)
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)	// 128
-#define INIT_SIZE_M128I (INIT_SIZE_BYTE >> 4) // 8
-
-
-#pragma pack(push, 1)
-union hash_state {
-  uint8_t b[200];
-  uint64_t w[25];
-};
-#pragma pack(pop)
-
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-    union hash_state hs;
-    struct {
-        uint8_t k[64];
-        uint8_t init[INIT_SIZE_BYTE];
-    };
-};
-#pragma pack(pop)
-
-void do_blake_hash(const void* input, size_t len, char* output);
-void do_groestl_hash(const void* input, size_t len, char* output);
-void do_jh_hash(const void* input, size_t len, char* output);
-void do_skein_hash(const void* input, size_t len, char* output);
-void cryptonight_hash_ctx(void* output, const void* input, int len);
-void keccakf(uint64_t st[25], int rounds);
-extern void (* const extra_hashes[4])(const void *, size_t, char *);
-
-int scanhash_cryptonight( struct work *work, uint32_t max_nonce,
-                           uint64_t *hashes_done, struct thr_info *mythr );
-
-void cryptonight_hash_aes( void *restrict output, const void *input, int len );
-
-extern bool cryptonightV7;
-
-#endif
-
--- a/algo/cubehash/cube-hash-2way.c
+++ b/algo/cubehash/cube-hash-2way.c
@@ -7,7 +7,7 @@

 // 2x128

-/*
+
 // The result of hashing 10 rounds of initial data which consists of params
 // zero padded.
 static const uint64_t IV256[] =
@@ -25,7 +25,239 @@ static const uint64_t IV512[] =
 0x148FE485FCD398D9, 0xB64445321B017BEF, 0x2FF5781C6A536159, 0x0DBADEA991FA7934,
 0xA5A70E75D65C8A2B, 0xBC796576B1C62456, 0xE7989AF11921C8F7, 0xD43E3B447795D246
 };
-*/
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+// 4 way 128 is handy to avoid reinterleaving in many algos.
+// If reinterleaving is necessary it may be more efficient to use
+// 2 way 256. The same transform code should work for both.
+
+static void transform_4way( cube_4way_context *sp )
+{
+    int r;
+    const int rounds = sp->rounds;
+
+    __m512i x0, x1, x2, x3, x4, x5, x6, x7, y0, y1;
+
+    x0 = _mm512_load_si512( (__m512i*)sp->h     );
+    x1 = _mm512_load_si512( (__m512i*)sp->h + 1 );
+    x2 = _mm512_load_si512( (__m512i*)sp->h + 2 );
+    x3 = _mm512_load_si512( (__m512i*)sp->h + 3 );
+    x4 = _mm512_load_si512( (__m512i*)sp->h + 4 );
+    x5 = _mm512_load_si512( (__m512i*)sp->h + 5 );
+    x6 = _mm512_load_si512( (__m512i*)sp->h + 6 );
+    x7 = _mm512_load_si512( (__m512i*)sp->h + 7 );
+
+    for ( r = 0; r < rounds; ++r )
+    {
+        x4 = _mm512_add_epi32( x0, x4 );
+        x5 = _mm512_add_epi32( x1, x5 );
+        x6 = _mm512_add_epi32( x2, x6 );
+        x7 = _mm512_add_epi32( x3, x7 );
+        y0 = x0;
+        y1 = x1;
+        x0 = mm512_rol_32( x2, 7 );
+        x1 = mm512_rol_32( x3, 7 );
+        x2 = mm512_rol_32( y0, 7 );
+        x3 = mm512_rol_32( y1, 7 );
+        x0 = _mm512_xor_si512( x0, x4 );
+        x1 = _mm512_xor_si512( x1, x5 );
+        x2 = _mm512_xor_si512( x2, x6 );
+        x3 = _mm512_xor_si512( x3, x7 );
+        x4 = mm512_swap128_64( x4 );
+        x5 = mm512_swap128_64( x5 );
+        x6 = mm512_swap128_64( x6 );
+        x7 = mm512_swap128_64( x7 );
+        x4 = _mm512_add_epi32( x0, x4 );
+        x5 = _mm512_add_epi32( x1, x5 );
+        x6 = _mm512_add_epi32( x2, x6 );
+        x7 = _mm512_add_epi32( x3, x7 );
+        y0 = x0;
+        y1 = x2;
+        x0 = mm512_rol_32( x1, 11 );
+        x1 = mm512_rol_32( y0, 11 );
+        x2 = mm512_rol_32( x3, 11 );
+        x3 = mm512_rol_32( y1, 11 );
+        x0 = _mm512_xor_si512( x0, x4 );
+        x1 = _mm512_xor_si512( x1, x5 );
+        x2 = _mm512_xor_si512( x2, x6 );
+        x3 = _mm512_xor_si512( x3, x7 );
+        x4 = mm512_swap64_32( x4 );
+        x5 = mm512_swap64_32( x5 );
+        x6 = mm512_swap64_32( x6 );
+        x7 = mm512_swap64_32( x7 );
+    }
+
+    _mm512_store_si512( (__m512i*)sp->h,     x0 );
+    _mm512_store_si512( (__m512i*)sp->h + 1, x1 );
+    _mm512_store_si512( (__m512i*)sp->h + 2, x2 );
+    _mm512_store_si512( (__m512i*)sp->h + 3, x3 );
+    _mm512_store_si512( (__m512i*)sp->h + 4, x4 );
+    _mm512_store_si512( (__m512i*)sp->h + 5, x5 );
+    _mm512_store_si512( (__m512i*)sp->h + 6, x6 );
+    _mm512_store_si512( (__m512i*)sp->h + 7, x7 );
+}
+
+int cube_4way_init( cube_4way_context *sp, int hashbitlen, int rounds,
+                    int blockbytes )
+{
+    __m512i *h = (__m512i*)sp->h;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = blockbytes/16;
+    sp->rounds    = rounds;
+    sp->pos       = 0;
+
+    h[ 0] = m512_const1_128( iv[0] );
+    h[ 1] = m512_const1_128( iv[1] );
+    h[ 2] = m512_const1_128( iv[2] );
+    h[ 3] = m512_const1_128( iv[3] );
+    h[ 4] = m512_const1_128( iv[4] );
+    h[ 5] = m512_const1_128( iv[5] );
+    h[ 6] = m512_const1_128( iv[6] );
+    h[ 7] = m512_const1_128( iv[7] );
+    h[ 0] = m512_const1_128( iv[0] );
+    h[ 1] = m512_const1_128( iv[1] );
+    h[ 2] = m512_const1_128( iv[2] );
+    h[ 3] = m512_const1_128( iv[3] );
+    h[ 4] = m512_const1_128( iv[4] );
+    h[ 5] = m512_const1_128( iv[5] );
+    h[ 6] = m512_const1_128( iv[6] );
+    h[ 7] = m512_const1_128( iv[7] );
+
+    return 0;
+}
+
+int cube_4way_update( cube_4way_context *sp, const void *data, size_t size )
+{
+    const int len = size >> 4;
+    const __m512i *in = (__m512i*)data;
+    int i;
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform_4way( sp );
+           sp->pos = 0;
+        }
+    }
+    return 0;
+}
+
+int cube_4way_close( cube_4way_context *sp, void *output )
+{
+    __m512i *hash = (__m512i*)output;
+    int i;
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ],
+                                 m512_const2_64( 0, 0x0000000000000080 ) );
+    transform_4way( sp );
+
+    sp->h[7] = _mm512_xor_si512( sp->h[7],
+                                 m512_const2_64( 0x0000000100000000, 0 ) );
+
+    for ( i = 0; i < 10; ++i ) 
+       transform_4way( sp );
+
+    memcpy( hash, sp->h, sp->hashlen<<6 );
+    return 0;
+}
+
+int cube_4way_full( cube_4way_context *sp, void *output,  int hashbitlen, 
+                    const void *data, size_t size )
+{
+    __m512i *h = (__m512i*)sp->h;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = 32/16;
+    sp->rounds    = 16;
+    sp->pos       = 0;
+
+    h[ 0] = m512_const1_128( iv[0] );
+    h[ 1] = m512_const1_128( iv[1] );
+    h[ 2] = m512_const1_128( iv[2] );
+    h[ 3] = m512_const1_128( iv[3] );
+    h[ 4] = m512_const1_128( iv[4] );
+    h[ 5] = m512_const1_128( iv[5] );
+    h[ 6] = m512_const1_128( iv[6] );
+    h[ 7] = m512_const1_128( iv[7] );
+
+    const int len = size >> 4;
+    const __m512i *in = (__m512i*)data;
+    __m512i *hash = (__m512i*)output;
+    int i;
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform_4way( sp );
+           sp->pos = 0;
+        }
+    }
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ],
+                                    m512_const2_64( 0, 0x0000000000000080 ) );
+    transform_4way( sp );
+
+    sp->h[7] = _mm512_xor_si512( sp->h[7],
+                                    m512_const2_64( 0x0000000100000000, 0 ) );
+
+    for ( i = 0; i < 10; ++i )
+       transform_4way( sp );
+
+    memcpy( hash, sp->h, sp->hashlen<<6);
+    return 0;
+}
+
+
+int cube_4way_update_close( cube_4way_context *sp, void *output,
+                               const void *data, size_t size )
+{
+    const int len = size >> 4;
+    const __m512i *in = (__m512i*)data;
+    __m512i *hash = (__m512i*)output;
+    int i;
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform_4way( sp );
+           sp->pos = 0;
+        }
+    }
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    sp->h[ sp->pos ] = _mm512_xor_si512( sp->h[ sp->pos ],
+                                    m512_const2_64( 0, 0x0000000000000080 ) );
+    transform_4way( sp );
+
+    sp->h[7] = _mm512_xor_si512( sp->h[7],
+                                    m512_const2_64( 0x0000000100000000, 0 ) );
+
+    for ( i = 0; i < 10; ++i )
+       transform_4way( sp );
+
+    memcpy( hash, sp->h, sp->hashlen<<6);
+    return 0;
+}
+
+
+#endif // AVX512
+
+// 2 way 128 

 static void transform_2way( cube_2way_context *sp )
 {
@@ -59,10 +291,10 @@ static void transform_2way( cube_2way_context *sp )
        x1 = _mm256_xor_si256( x1, x5 );
        x2 = _mm256_xor_si256( x2, x6 );
        x3 = _mm256_xor_si256( x3, x7 );
-        x4 = mm256_swap64_128( x4 );
-        x5 = mm256_swap64_128( x5 );
-        x6 = mm256_swap64_128( x6 );
-        x7 = mm256_swap64_128( x7 );
+        x4 = mm256_swap128_64( x4 );
+        x5 = mm256_swap128_64( x5 );
+        x6 = mm256_swap128_64( x6 );
+        x7 = mm256_swap128_64( x7 );
        x4 = _mm256_add_epi32( x0, x4 );
        x5 = _mm256_add_epi32( x1, x5 );
        x6 = _mm256_add_epi32( x2, x6 );
@@ -77,10 +309,10 @@ static void transform_2way( cube_2way_context *sp )
        x1 = _mm256_xor_si256( x1, x5 );
        x2 = _mm256_xor_si256( x2, x6 );
        x3 = _mm256_xor_si256( x3, x7 );
-        x4 = mm256_swap32_64( x4 );
-        x5 = mm256_swap32_64( x5 );
-        x6 = mm256_swap32_64( x6 );
-        x7 = mm256_swap32_64( x7 );
+        x4 = mm256_swap64_32( x4 );
+        x5 = mm256_swap64_32( x5 );
+        x6 = mm256_swap64_32( x6 );
+        x7 = mm256_swap64_32( x7 );
    }

    _mm256_store_si256( (__m256i*)sp->h,     x0 );
@@ -91,45 +323,35 @@ static void transform_2way( cube_2way_context *sp )
    _mm256_store_si256( (__m256i*)sp->h + 5, x5 );
    _mm256_store_si256( (__m256i*)sp->h + 6, x6 );
    _mm256_store_si256( (__m256i*)sp->h + 7, x7 );
-
 }

 int cube_2way_init( cube_2way_context *sp, int hashbitlen, int rounds,
                    int blockbytes )
 {
-    __m128i* h = (__m128i*)sp->h;
+    __m256i *h = (__m256i*)sp->h;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
    sp->hashlen   = hashbitlen/128;
    sp->blocksize = blockbytes/16;
    sp->rounds    = rounds;
    sp->pos       = 0;

-    if ( hashbitlen == 512 )
-    {
-
-       h[ 0] = m128_const_64( 0x4167D83E2D538B8B, 0x50F494D42AEA2A61 );
-       h[ 2] = m128_const_64( 0x50AC5695CC39968E, 0xC701CF8C3FEE2313 );
-       h[ 4] = m128_const_64( 0x825B453797CF0BEF, 0xA647A8B34D42C787 );
-       h[ 6] = m128_const_64( 0xA23911AED0E5CD33, 0xF22090C4EEF864D2 );
-       h[ 8] = m128_const_64( 0xB64445321B017BEF, 0x148FE485FCD398D9 );
-       h[10] = m128_const_64( 0x0DBADEA991FA7934, 0x2FF5781C6A536159 );
-       h[12] = m128_const_64( 0xBC796576B1C62456, 0xA5A70E75D65C8A2B );
-       h[14] = m128_const_64( 0xD43E3B447795D246, 0xE7989AF11921C8F7 );
-       h[1] = h[ 0];  h[ 3] = h[ 2]; h[ 5] = h[ 4]; h[ 7] = h[ 6];
-       h[9] = h[ 8];  h[11] = h[10]; h[13] = h[12]; h[15] = h[14];
-    }
-    else
-    {
-       h[ 0] = m128_const_64( 0x35481EAE63117E71, 0xCCD6F29FEA2BD4B4 );
-       h[ 2] = m128_const_64( 0xF4CC12BE7E624131, 0xE5D94E6322512D5B );
-       h[ 4] = m128_const_64( 0x3361DA8CD0720C35, 0x42AF2070C2D0B696 );
-       h[ 6] = m128_const_64( 0x40E5FBAB4680AC00, 0x8EF8AD8328CCECA4 );
-       h[ 8] = m128_const_64( 0xF0B266796C859D41, 0x6107FBD5D89041C3 );
-       h[10] = m128_const_64( 0x93CB628565C892FD, 0x5FA2560309392549 );
-       h[12] = m128_const_64( 0x85254725774ABFDD, 0x9E4B4E602AF2B5AE );
-       h[14] = m128_const_64( 0xD6032C0A9CDAF8AF, 0x4AB6AAD615815AEB );
-       h[1] = h[ 0];  h[ 3] = h[ 2]; h[ 5] = h[ 4]; h[ 7] = h[ 6];
-       h[9] = h[ 8];  h[11] = h[10]; h[13] = h[12]; h[15] = h[14];
-    }
+    h[ 0] = m256_const1_128( iv[0] );
+    h[ 1] = m256_const1_128( iv[1] );
+    h[ 2] = m256_const1_128( iv[2] );
+    h[ 3] = m256_const1_128( iv[3] );
+    h[ 4] = m256_const1_128( iv[4] );
+    h[ 5] = m256_const1_128( iv[5] );
+    h[ 6] = m256_const1_128( iv[6] );
+    h[ 7] = m256_const1_128( iv[7] );
+    h[ 0] = m256_const1_128( iv[0] );
+    h[ 1] = m256_const1_128( iv[1] );
+    h[ 2] = m256_const1_128( iv[2] );
+    h[ 3] = m256_const1_128( iv[3] );
+    h[ 4] = m256_const1_128( iv[4] );
+    h[ 5] = m256_const1_128( iv[5] );
+    h[ 6] = m256_const1_128( iv[6] );
+    h[ 7] = m256_const1_128( iv[7] );
    
    return 0;
 }
@@ -141,9 +363,6 @@ int cube_2way_update( cube_2way_context *sp, const void *data, size_t size )
    const __m256i *in = (__m256i*)data;
    int i;

-    // It is assumed data is aligned to 256 bits and is a multiple of 128 bits.
-    // Current usage sata is either 64 or 80 bytes.
-
    for ( i = 0; i < len; i++ )
    {
        sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ], in[i] );
@@ -164,11 +383,11 @@ int cube_2way_close( cube_2way_context *sp, void *output )

    // pos is zero for 64 byte data, 1 for 80 byte data.
    sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ],
-                                _mm256_set_epi32( 0,0,0,0x80,  0,0,0,0x80 ) );
+                                   m256_const2_64( 0, 0x0000000000000080 ) );
    transform_2way( sp );

    sp->h[7] = _mm256_xor_si256( sp->h[7],
-		                 _mm256_set_epi32( 1,0,0,0,  1,0,0,0 ) );
+                                   m256_const2_64( 0x0000000100000000, 0 ) );

    for ( i = 0; i < 10; ++i )           transform_2way( sp );

@@ -197,13 +416,63 @@ int cube_2way_update_close( cube_2way_context *sp, void *output,

    // pos is zero for 64 byte data, 1 for 80 byte data.
    sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ],
-                    _mm256_set_epi32( 0,0,0,0x80,  0,0,0,0x80 ) );
+                                    m256_const2_64( 0, 0x0000000000000080 ) );
    transform_2way( sp );

-    sp->h[7] = _mm256_xor_si256( sp->h[7], _mm256_set_epi32( 1,0,0,0,
-                                                             1,0,0,0 ) );
+    sp->h[7] = _mm256_xor_si256( sp->h[7],
+                                    m256_const2_64( 0x0000000100000000, 0 ) );

-    for ( i = 0; i < 10; ++i )            transform_2way( sp );
+    for ( i = 0; i < 10; ++i )    transform_2way( sp );
+
+    memcpy( hash, sp->h, sp->hashlen<<5 );
+    return 0;
+}
+
+int cube_2way_full( cube_2way_context *sp, void *output, int hashbitlen,
+                               const void *data, size_t size )
+{
+    __m256i *h = (__m256i*)sp->h;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = 32/16;
+    sp->rounds    = 16;
+    sp->pos       = 0;
+
+    h[ 0] = m256_const1_128( iv[0] );
+    h[ 1] = m256_const1_128( iv[1] );
+    h[ 2] = m256_const1_128( iv[2] );
+    h[ 3] = m256_const1_128( iv[3] );
+    h[ 4] = m256_const1_128( iv[4] );
+    h[ 5] = m256_const1_128( iv[5] );
+    h[ 6] = m256_const1_128( iv[6] );
+    h[ 7] = m256_const1_128( iv[7] );
+
+    const int len = size >> 4;
+    const __m256i *in = (__m256i*)data;
+    __m256i *hash = (__m256i*)output;
+    int i;
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform_2way( sp );
+           sp->pos = 0;
+        }
+    }
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    sp->h[ sp->pos ] = _mm256_xor_si256( sp->h[ sp->pos ],
+                                    m256_const2_64( 0, 0x0000000000000080 ) );
+    transform_2way( sp );
+
+    sp->h[7] = _mm256_xor_si256( sp->h[7],
+                                    m256_const2_64( 0x0000000100000000, 0 ) );
+
+    for ( i = 0; i < 10; ++i )    transform_2way( sp );

    memcpy( hash, sp->h, sp->hashlen<<5 );
    return 0;
--- a/algo/cubehash/cube-hash-2way.h
+++ b/algo/cubehash/cube-hash-2way.h
@@ -1,11 +1,56 @@
 #ifndef CUBE_HASH_2WAY_H__
-#define CUBE_HASH_2WAY_H__
-
-#if defined(__AVX2__)
+#define CUBE_HASH_2WAY_H__ 1

 #include <stdint.h>
 #include "simd-utils.h"

+#if defined(__AVX2__)
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+struct _cube_4way_context
+{
+    __m512i h[8];
+    int hashlen;
+    int rounds;
+    int blocksize;
+    int pos; 
+} __attribute__ ((aligned (128)));
+
+typedef struct _cube_4way_context cube_4way_context;
+
+int cube_4way_init( cube_4way_context* sp, int hashbitlen, int rounds,
+                       int blockbytes );
+int cube_4way_update( cube_4way_context *sp, const void *data, size_t size );
+int cube_4way_close( cube_4way_context *sp, void *output );
+int cube_4way_update_close( cube_4way_context *sp, void *output,
+                            const void *data, size_t size );
+int cube_4way_full( cube_4way_context *sp, void *output, int hashbitlen,
+                    const void *data, size_t size );
+
+int cube_4x256_full( cube_4way_context *sp, void *output, int hashbitlen,
+                     const void *data, size_t size );
+
+#define cube512_4way_init( sp ) cube_4way_update( sp, 512 )
+#define cube512_4way_update cube_4way_update
+#define cube512_4way_update_close cube_4way_update
+#define cube512_4way_close cube_4way_update
+#define cube512_4way_full( sp, output, data, size ) \
+           cube_4way_full( sp, output, 512, data, size )
+#define cube512_4x256_full( sp, output, data, size ) \
+           cube_4x256_full( sp, output, 512, data, size )
+
+#define cube256_4way_init( sp ) cube_4way_update( sp, 256 )
+#define cube256_4way_update cube_4way_update
+#define cube256_4way_update_close cube_4way_update
+#define cube256_4way_close cube_4way_update
+#define cube256_4way_full( sp, output, data, size ) \
+           cube_4way_full( sp, output, 256, data, size )
+#define cube256_4x256_full( sp, output, data, size ) \
+           cube_4x256_full( sp, output, 256, data, size )
+
+#endif
+
 // 2x128, 2 way parallel SSE2

 struct _cube_2way_context
@@ -15,21 +60,18 @@ struct _cube_2way_context
    int rounds;
    int blocksize;         // __m128i
    int pos;               // number of __m128i read into x from current block
-} __attribute__ ((aligned (64)));
+} __attribute__ ((aligned (128)));

 typedef struct _cube_2way_context cube_2way_context;

 int cube_2way_init( cube_2way_context* sp, int hashbitlen, int rounds,
                       int blockbytes );
-// reinitialize context with same parameters, much faster.
-int cube_2way_reinit( cube_2way_context *sp );
-
 int cube_2way_update( cube_2way_context *sp, const void *data, size_t size );
-
 int cube_2way_close( cube_2way_context *sp, void *output );
-
 int cube_2way_update_close( cube_2way_context *sp, void *output,
                            const void *data, size_t size );
+int cube_2way_full( cube_2way_context *sp, void *output, int hashbitlen,
+                    const void *data, size_t size );


 #endif
--- a/algo/cubehash/cubehash_sse2.c
+++ b/algo/cubehash/cubehash_sse2.c
@@ -21,7 +21,27 @@ static void transform( cubehashParam *sp )
    int r;
    const int rounds = sp->rounds;

-#ifdef __AVX2__
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+    register __m512i x0, x1;
+
+    x0 = _mm512_load_si512( (__m512i*)sp->x     );
+    x1 = _mm512_load_si512( (__m512i*)sp->x + 1 );
+
+    for ( r = 0; r < rounds; ++r )
+    { 
+        x1 = _mm512_add_epi32( x0, x1 );
+        x0 = _mm512_xor_si512( mm512_rol_32( mm512_swap_256( x0 ), 7 ), x1 );
+        x1 = _mm512_add_epi32( x0, mm512_swap128_64( x1 ) );
+        x0 = _mm512_xor_si512( mm512_rol_32(
+                                         mm512_swap256_128( x0 ), 11 ), x1 );
+        x1 = mm512_swap64_32( x1 );
+    }
+
+    _mm512_store_si512( (__m512i*)sp->x,     x0 );
+    _mm512_store_si512( (__m512i*)sp->x + 1, x1 );
+
+#elif defined(__AVX2__)

    register __m256i x0, x1, x2, x3, y0, y1;

@@ -39,8 +59,8 @@ static void transform( cubehashParam *sp )
        x1 = mm256_rol_32( y0, 7 );
        x0 = _mm256_xor_si256( x0, x2 );
        x1 = _mm256_xor_si256( x1, x3 );
-        x2 = mm256_swap64_128( x2 );
-        x3 = mm256_swap64_128( x3 );
+        x2 = mm256_swap128_64( x2 );
+        x3 = mm256_swap128_64( x3 );
        x2 = _mm256_add_epi32( x0, x2 );
        x3 = _mm256_add_epi32( x1, x3 );
        y0 = mm256_swap_128( x0 );
@@ -49,8 +69,8 @@ static void transform( cubehashParam *sp )
        x1 = mm256_rol_32( y1, 11 );
        x0 = _mm256_xor_si256( x0, x2 );
        x1 = _mm256_xor_si256( x1, x3 );
-        x2 = mm256_swap32_64( x2 );
-        x3 = mm256_swap32_64( x3 );
+        x2 = mm256_swap64_32( x2 );
+        x3 = mm256_swap64_32( x3 );
    }

    _mm256_store_si256( (__m256i*)sp->x,     x0 );
@@ -210,11 +230,10 @@ int cubehashDigest( cubehashParam *sp, byte *digest )

    // pos is zero for 64 byte data, 1 for 80 byte data.
    sp->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ],
-                                      _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                                    0,0,0,0, 0,0,0,0x80 ) );
+                                      m128_const_64( 0, 0x80 ) );
    transform( sp );

-    sp->x[7] = _mm_xor_si128( sp->x[7], _mm_set_epi32( 1,0,0,0 ) );
+    sp->x[7] = _mm_xor_si128( sp->x[7], m128_const_64( 0x100000000, 0 ) );
    transform( sp );
    transform( sp );
    transform( sp );
@@ -256,11 +275,89 @@ int cubehashUpdateDigest( cubehashParam *sp, byte *digest,

    // pos is zero for 64 byte data, 1 for 80 byte data.
    sp->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ],
-                                      _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                                    0,0,0,0, 0,0,0,0x80 ) );
+                                      m128_const_64( 0, 0x80 ) );
    transform( sp );

-    sp->x[7] = _mm_xor_si128( sp->x[7], _mm_set_epi32( 1,0,0,0 ) );
+    sp->x[7] = _mm_xor_si128( sp->x[7], m128_const_64( 0x100000000, 0 ) );
+
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+    transform( sp );
+
+    for ( i = 0; i < sp->hashlen; i++ )
+       hash[i] = sp->x[i];
+
+    return SUCCESS;
+}
+
+int cubehash_full( cubehashParam *sp, byte *digest, int hashbitlen,
+                          const byte *data, size_t size )
+{
+    __m128i *x = (__m128i*)sp->x;
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = 32/16;
+    sp->rounds    = 16;
+    sp->pos       = 0;
+
+    if ( hashbitlen == 512 )
+    {
+
+       x[0] = m128_const_64( 0x4167D83E2D538B8B, 0x50F494D42AEA2A61 );
+       x[1] = m128_const_64( 0x50AC5695CC39968E, 0xC701CF8C3FEE2313 );
+       x[2] = m128_const_64( 0x825B453797CF0BEF, 0xA647A8B34D42C787 );
+       x[3] = m128_const_64( 0xA23911AED0E5CD33, 0xF22090C4EEF864D2 );
+       x[4] = m128_const_64( 0xB64445321B017BEF, 0x148FE485FCD398D9 );
+       x[5] = m128_const_64( 0x0DBADEA991FA7934, 0x2FF5781C6A536159 );
+       x[6] = m128_const_64( 0xBC796576B1C62456, 0xA5A70E75D65C8A2B );
+       x[7] = m128_const_64( 0xD43E3B447795D246, 0xE7989AF11921C8F7 );
+    }
+    else
+    {
+       x[0] = m128_const_64( 0x35481EAE63117E71, 0xCCD6F29FEA2BD4B4 );
+       x[1] = m128_const_64( 0xF4CC12BE7E624131, 0xE5D94E6322512D5B );
+       x[2] = m128_const_64( 0x3361DA8CD0720C35, 0x42AF2070C2D0B696 );
+       x[3] = m128_const_64( 0x40E5FBAB4680AC00, 0x8EF8AD8328CCECA4 );
+       x[4] = m128_const_64( 0xF0B266796C859D41, 0x6107FBD5D89041C3 );
+       x[5] = m128_const_64( 0x93CB628565C892FD, 0x5FA2560309392549 );
+       x[6] = m128_const_64( 0x85254725774ABFDD, 0x9E4B4E602AF2B5AE );
+       x[7] = m128_const_64( 0xD6032C0A9CDAF8AF, 0x4AB6AAD615815AEB );
+    }
+
+
+
+
+    const int len = size / 16;
+    const __m128i* in = (__m128i*)data;
+    __m128i* hash = (__m128i*)digest;
+    int i;
+
+    // It is assumed data is aligned to 256 bits and is a multiple of 128 bits.
+    // Current usage sata is either 64 or 80 bytes.
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform( sp );
+           sp->pos = 0;
+        }
+    }
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    sp->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ],
+                                      m128_const_64( 0, 0x80 ) );
+    transform( sp );
+
+    sp->x[7] = _mm_xor_si128( sp->x[7], m128_const_64( 0x100000000, 0 ) );

    transform( sp );
    transform( sp );
--- a/algo/cubehash/cubehash_sse2.h
+++ b/algo/cubehash/cubehash_sse2.h
@@ -19,7 +19,7 @@ struct _cubehashParam
    int rounds;
    int blocksize;         // __m128i
    int pos;	           // number of __m128i read into x from current block
-    __m128i _ALIGN(256) x[8];  // aligned for __m256i
+    __m128i _ALIGN(64) x[8];  // aligned for __m256i
 };

 typedef struct _cubehashParam cubehashParam;
@@ -39,6 +39,9 @@ int cubehashDigest(cubehashParam* sp, byte *digest);
 int cubehashUpdateDigest( cubehashParam *sp, byte *digest, const byte *data,
                          size_t size );

+int cubehash_full( cubehashParam* sp, byte *digest, int hashbitlen,
+                   const byte *data, size_t size );
+
 #ifdef __cplusplus
 }
 #endif
--- a/algo/echo/aes_ni/hash.c
+++ b/algo/echo/aes_ni/hash.c
@@ -7,7 +7,6 @@
 * - implements NIST hash api
 * - assumes that message lenght is multiple of 8-bits
 * - _ECHO_VPERM_ must be defined if compiling with ../main.c
- * -  define NO_AES_NI for aes_ni version
 *
 * Cagdas Calik
 * ccalik@metu.edu.tr
@@ -21,13 +20,7 @@
 #include "hash_api.h"
 //#include "vperm.h"
 #include <immintrin.h>
-/*
-#ifndef NO_AES_NI
-#include <wmmintrin.h>
-#else
-#include <tmmintrin.h>
-#endif
-*/
+#include "simd-utils.h"

 MYALIGN const unsigned int _k_s0F[] = {0x0F0F0F0F, 0x0F0F0F0F, 0x0F0F0F0F, 0x0F0F0F0F};
 MYALIGN const unsigned int _k_ipt[] = {0x5A2A7000, 0xC2B2E898, 0x52227808, 0xCABAE090, 0x317C4D00, 0x4C01307D, 0xB0FDCC81, 0xCD80B1FC};
@@ -62,8 +55,8 @@ MYALIGN const unsigned int	mul2ipt[]		= {0x728efc00, 0x6894e61a, 0x3fc3b14d, 0x2

 #define ECHO_SUBBYTES(state, i, j) \
 	state[i][j] = _mm_aesenc_si128(state[i][j], k1);\
-	state[i][j] = _mm_aesenc_si128(state[i][j], M128(zero));\
-	k1 = _mm_add_epi32(k1, M128(const1))
+   k1 = _mm_add_epi32(k1, M128(const1));\
+	state[i][j] = _mm_aesenc_si128(state[i][j], M128(zero))

 #define ECHO_MIXBYTES(state1, state2, j, t1, t2, s2) \
 	s2 = _mm_add_epi8(state1[0][j], state1[0][j]);\
@@ -179,53 +172,53 @@ void Compress(hashState_echo *ctx, const unsigned char *pmsg, unsigned int uBloc

   for(b = 0; b < uBlockCount; b++)
   {
-	ctx->k = _mm_add_epi64(ctx->k, ctx->const1536);
+   	ctx->k = _mm_add_epi64(ctx->k, ctx->const1536);

-	// load message
-	for(j = ctx->uHashSize / 256; j < 4; j++)
-	{
-	   for(i = 0; i < 4; i++)
+   	// load message
+	   for(j = ctx->uHashSize / 256; j < 4; j++)
 	   {
-		_state[i][j] = _mm_loadu_si128((__m128i*)pmsg + 4 * (j - (ctx->uHashSize / 256)) + i);
+	      for(i = 0; i < 4; i++)
+	      {
+		     _state[i][j] = _mm_load_si128((__m128i*)pmsg + 4 * (j - (ctx->uHashSize / 256)) + i);
+	      }
 	   }
-	}

-	// save state
-	SAVESTATE(_statebackup, _state);
+	   // save state
+	   SAVESTATE(_statebackup, _state);

-	k1 = ctx->k;
+	   k1 = ctx->k;

-	for(r = 0; r < ctx->uRounds / 2; r++)
-	{
-		ECHO_ROUND_UNROLL2;
-	}
+	   for(r = 0; r < ctx->uRounds / 2; r++)
+   	{
+	   	ECHO_ROUND_UNROLL2;
+	   }
 		
-	if(ctx->uHashSize == 256)
-	{
-	   for(i = 0; i < 4; i++)
+	   if(ctx->uHashSize == 256)
 	   {
-		_state[i][0] = _mm_xor_si128(_state[i][0], _state[i][1]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _state[i][2]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _state[i][3]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][0]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][1]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][2]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][3]);
+	      for(i = 0; i < 4; i++)
+	      {
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _state[i][1]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _state[i][2]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _state[i][3]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][0]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][1]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][2]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][3]);
+	      }
 	   }
-	}
-	else
-	{
-	   for(i = 0; i < 4; i++)
-	   {
-		_state[i][0] = _mm_xor_si128(_state[i][0], _state[i][2]);
-		_state[i][1] = _mm_xor_si128(_state[i][1], _state[i][3]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][0]);
-		_state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][2]);
-		_state[i][1] = _mm_xor_si128(_state[i][1], _statebackup[i][1]);
-		_state[i][1] = _mm_xor_si128(_state[i][1], _statebackup[i][3]);
-           }
-	}
-	pmsg += ctx->uBlockLength;
+	   else
+    	{
+	      for(i = 0; i < 4; i++)
+	      {
+      		_state[i][0] = _mm_xor_si128(_state[i][0], _state[i][2]);
+		      _state[i][1] = _mm_xor_si128(_state[i][1], _state[i][3]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][0]);
+		      _state[i][0] = _mm_xor_si128(_state[i][0], _statebackup[i][2]);
+		      _state[i][1] = _mm_xor_si128(_state[i][1], _statebackup[i][1]);
+		      _state[i][1] = _mm_xor_si128(_state[i][1], _statebackup[i][3]);
+         }
+   	}
+	   pmsg += ctx->uBlockLength;
   }
 	SAVESTATE(ctx->state, _state);

@@ -390,13 +383,13 @@ HashReturn final_echo(hashState_echo *state, BitSequence *hashval)
 	}

 	// Store the hash value
-	_mm_storeu_si128((__m128i*)hashval + 0, state->state[0][0]);
-	_mm_storeu_si128((__m128i*)hashval + 1, state->state[1][0]);
+	_mm_store_si128((__m128i*)hashval + 0, state->state[0][0]);
+	_mm_store_si128((__m128i*)hashval + 1, state->state[1][0]);

 	if(state->uHashSize == 512)
 	{
-		_mm_storeu_si128((__m128i*)hashval + 2, state->state[2][0]);
-		_mm_storeu_si128((__m128i*)hashval + 3, state->state[3][0]);
+		_mm_store_si128((__m128i*)hashval + 2, state->state[2][0]);
+		_mm_store_si128((__m128i*)hashval + 3, state->state[3][0]);
 	}

 	return SUCCESS;
@@ -513,18 +506,177 @@ HashReturn update_final_echo( hashState_echo *state, BitSequence *hashval,
   }

   // Store the hash value
-   _mm_storeu_si128( (__m128i*)hashval + 0, state->state[0][0] );
-   _mm_storeu_si128( (__m128i*)hashval + 1, state->state[1][0] );
+   _mm_store_si128( (__m128i*)hashval + 0, state->state[0][0] );
+   _mm_store_si128( (__m128i*)hashval + 1, state->state[1][0] );

   if( state->uHashSize == 512 )
   {
-        _mm_storeu_si128( (__m128i*)hashval + 2, state->state[2][0] );
-        _mm_storeu_si128( (__m128i*)hashval + 3, state->state[3][0] );
+        _mm_store_si128( (__m128i*)hashval + 2, state->state[2][0] );
+        _mm_store_si128( (__m128i*)hashval + 3, state->state[3][0] );

   }
   return SUCCESS;
 }

+HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
+            int nHashSize, const BitSequence *data, DataLength datalen )
+{
+   int i, j;
+
+   state->k = m128_zero;
+   state->processed_bits = 0;
+   state->uBufferBytes = 0;
+
+   switch( nHashSize )
+   {
+      case 256:
+         state->uHashSize = 256;
+         state->uBlockLength = 192;
+         state->uRounds = 8;
+         state->hashsize = m128_const_64( 0, 0x100 );
+         state->const1536 = m128_const_64( 0, 0x600 );
+         break;
+
+      case 512:
+         state->uHashSize = 512;
+         state->uBlockLength = 128;
+         state->uRounds = 10;
+         state->hashsize = m128_const_64( 0, 0x200 );
+         state->const1536 = m128_const_64( 0, 0x400 );
+         break;
+
+      default:
+         return BAD_HASHBITLEN;
+   }
+
+   for(i = 0; i < 4; i++)
+      for(j = 0; j < nHashSize / 256; j++)
+         state->state[i][j] = state->hashsize;
+
+   for(i = 0; i < 4; i++)
+      for(j = nHashSize / 256; j < 4; j++)
+         state->state[i][j] = m128_zero;
+
+
+   unsigned int uBlockCount, uRemainingBytes;
+
+   if( (state->uBufferBytes + datalen) >= state->uBlockLength )
+   {
+        if( state->uBufferBytes != 0 )
+        {
+           // Fill the buffer
+           memcpy( state->buffer + state->uBufferBytes,
+                   (void*)data, state->uBlockLength - state->uBufferBytes );
+
+           // Process buffer
+           Compress( state, state->buffer, 1 );
+           state->processed_bits += state->uBlockLength * 8;
+
+           data += state->uBlockLength - state->uBufferBytes;
+           datalen -= state->uBlockLength - state->uBufferBytes;
+        }
+
+        // buffer now does not contain any unprocessed bytes
+
+        uBlockCount = datalen / state->uBlockLength;
+        uRemainingBytes = datalen % state->uBlockLength;
+
+        if( uBlockCount > 0 )
+        {
+           Compress( state, data, uBlockCount );
+           state->processed_bits += uBlockCount * state->uBlockLength * 8;
+           data += uBlockCount * state->uBlockLength;
+        }
+
+        if( uRemainingBytes > 0 )
+        memcpy(state->buffer, (void*)data, uRemainingBytes);
+
+        state->uBufferBytes = uRemainingBytes;
+   }
+   else
+   {
+        memcpy( state->buffer + state->uBufferBytes, (void*)data, datalen );
+        state->uBufferBytes += datalen;
+   }
+
+   __m128i remainingbits;
+
+   // Add remaining bytes in the buffer
+   state->processed_bits += state->uBufferBytes * 8;
+
+   remainingbits = _mm_set_epi32( 0, 0, 0, state->uBufferBytes * 8 );
+
+   // Pad with 0x80
+   state->buffer[state->uBufferBytes++] = 0x80;
+   // Enough buffer space for padding in this block?
+   if( (state->uBlockLength - state->uBufferBytes) >= 18 )
+   {
+        // Pad with zeros
+        memset( state->buffer + state->uBufferBytes, 0, state->uBlockLength - (state->uBufferBytes + 18) );
+
+        // Hash size
+        *( (unsigned short*)(state->buffer + state->uBlockLength - 18) ) = state->uHashSize;
+
+        // Processed bits
+        *( (DataLength*)(state->buffer + state->uBlockLength - 16) ) =
+                   state->processed_bits;
+        *( (DataLength*)(state->buffer + state->uBlockLength - 8) ) = 0;
+
+        // Last block contains message bits?
+        if( state->uBufferBytes == 1 )
+        {
+           state->k = _mm_xor_si128( state->k, state->k );
+           state->k = _mm_sub_epi64( state->k, state->const1536 );
+        }
+        else
+        {
+           state->k = _mm_add_epi64( state->k, remainingbits );
+           state->k = _mm_sub_epi64( state->k, state->const1536 );
+        }
+
+        // Compress
+        Compress( state, state->buffer, 1 );
+   }
+   else
+   {
+        // Fill with zero and compress
+        memset( state->buffer + state->uBufferBytes, 0,
+                state->uBlockLength - state->uBufferBytes );
+        state->k = _mm_add_epi64( state->k, remainingbits );
+        state->k = _mm_sub_epi64( state->k, state->const1536 );
+        Compress( state, state->buffer, 1 );
+
+        // Last block
+        memset( state->buffer, 0, state->uBlockLength - 18 );
+
+        // Hash size
+        *( (unsigned short*)(state->buffer + state->uBlockLength - 18) ) =
+                 state->uHashSize;
+
+        // Processed bits
+        *( (DataLength*)(state->buffer + state->uBlockLength - 16) ) =
+                   state->processed_bits;
+        *( (DataLength*)(state->buffer + state->uBlockLength - 8) ) = 0;
+        // Compress the last block
+        state->k = _mm_xor_si128( state->k, state->k );
+        state->k = _mm_sub_epi64( state->k, state->const1536 );
+        Compress( state, state->buffer, 1) ;
+   }
+
+   // Store the hash value
+   _mm_store_si128( (__m128i*)hashval + 0, state->state[0][0] );
+   _mm_store_si128( (__m128i*)hashval + 1, state->state[1][0] );
+
+   if( state->uHashSize == 512 )
+   {
+        _mm_store_si128( (__m128i*)hashval + 2, state->state[2][0] );
+        _mm_store_si128( (__m128i*)hashval + 3, state->state[3][0] );
+
+   }
+   return SUCCESS;
+}
+
+

 HashReturn hash_echo(int hashbitlen, const BitSequence *data, DataLength databitlen, BitSequence *hashval)
 {
--- a/algo/echo/aes_ni/hash_api.h
+++ b/algo/echo/aes_ni/hash_api.h
@@ -15,7 +15,7 @@
 #ifndef HASH_API_H
 #define HASH_API_H

-#ifndef NO_AES_NI
+#ifdef __AES__
 #define HASH_IMPL_STR	"ECHO-aesni"
 #else
 #define HASH_IMPL_STR	"ECHO-vperm"
@@ -55,6 +55,8 @@ HashReturn hash_echo(int hashbitlen, const BitSequence *data, DataLength databit

 HashReturn update_final_echo( hashState_echo *state, BitSequence *hashval,
                              const BitSequence *data, DataLength databitlen );
+HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
+            int nHashSize, const BitSequence *data, DataLength databitlen );

 #endif // HASH_API_H

--- a/algo/echo/echo-hash-4way.c
+++ b/algo/echo/echo-hash-4way.c
@@ -0,0 +1,776 @@
+#if defined(__VAES__)
+
+#include "simd-utils.h"
+#include "echo-hash-4way.h"
+
+/*
+static const unsigned int mul2ipt[] __attribute__ ((aligned (64))) =
+{  
+   0x728efc00, 0x6894e61a, 0x3fc3b14d, 0x25d9ab57,
+   0xfd5ba600, 0x2a8c71d7, 0x1eb845e3, 0xc96f9234
+};
+*/
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+
+//#define mul2mask    m512_const2_64( 0, 0x00001b00 )
+//_mm512_set4_epi32( 0, 0, 0, 0x00001b00 ) 
+//_mm512_set4_epi32( 0x00001b00, 0, 0, 0 )  
+
+//#define lsbmask    m512_const1_32( 0x01010101 ) 
+
+#define ECHO_SUBBYTES( state, i, j ) \
+	state[i][j] = _mm512_aesenc_epi128( state[i][j], k1 ); \
+   k1 = _mm512_add_epi32( k1, one ); \
+	state[i][j] = _mm512_aesenc_epi128( state[i][j], m512_zero );
+
+#define ECHO_MIXBYTES( state1, state2, j, t1, t2, s2 ) do \
+{ \
+   const int j1 = ( (j)+1 ) & 3; \
+   const int j2 = ( (j)+2 ) & 3; \
+   const int j3 = ( (j)+3 ) & 3; \
+   s2 = _mm512_add_epi8( state1[ 0 ] [j ], state1[ 0 ][ j ] ); \
+   t1 = _mm512_srli_epi16( state1[ 0 ][ j ], 7 ); \
+   t1 = _mm512_and_si512( t1, lsbmask );\
+   t2 = _mm512_shuffle_epi8( mul2mask, t1 ); \
+   s2 = _mm512_xor_si512( s2, t2 ); \
+   state2[ 0 ] [j ] = s2; \
+   state2[ 1 ] [j ] = state1[ 0 ][ j ]; \
+   state2[ 2 ] [j ] = state1[ 0 ][ j ]; \
+   state2[ 3 ] [j ] = _mm512_xor_si512( s2, state1[ 0 ][ j ] );\
+   s2 = _mm512_add_epi8( state1[ 1 ][ j1 ], state1[ 1 ][ j1 ] ); \
+   t1 = _mm512_srli_epi16( state1[ 1 ][ j1 ], 7 ); \
+   t1 = _mm512_and_si512( t1, lsbmask ); \
+   t2 = _mm512_shuffle_epi8( mul2mask, t1 ); \
+   s2 = _mm512_xor_si512( s2, t2 );\
+   state2[ 0 ][ j ] = _mm512_xor_si512( state2[ 0 ][ j ], \
+                              _mm512_xor_si512( s2, state1[ 1 ][ j1 ] ) ); \
+   state2[ 1 ][ j ] = _mm512_xor_si512( state2[ 1 ][ j ], s2 ); \
+   state2[ 2 ][ j ] = _mm512_xor_si512( state2[ 2 ][ j ], state1[ 1 ][ j1 ] ); \
+   state2[ 3 ][ j ] = _mm512_xor_si512( state2[ 3 ][ j ], state1[ 1 ][ j1 ] ); \
+   s2 = _mm512_add_epi8( state1[ 2 ][ j2 ], state1[ 2 ][ j2 ] ); \
+   t1 = _mm512_srli_epi16( state1[ 2 ][ j2 ], 7 ); \
+   t1 = _mm512_and_si512( t1, lsbmask ); \
+   t2 = _mm512_shuffle_epi8( mul2mask, t1 ); \
+   s2 = _mm512_xor_si512( s2, t2 ); \
+   state2[ 0 ][ j ] = _mm512_xor_si512( state2[ 0 ][ j ], state1[ 2 ][ j2 ] ); \
+   state2[ 1 ][ j ] = _mm512_xor_si512( state2[ 1 ][ j ], \
+                            _mm512_xor_si512( s2, state1[ 2 ][ j2 ] ) ); \
+   state2[ 2 ][ j ] = _mm512_xor_si512( state2[ 2 ][ j ], s2 ); \
+   state2[ 3 ][ j ] = _mm512_xor_si512( state2[ 3][ j ], state1[ 2 ][ j2 ] ); \
+   s2 = _mm512_add_epi8( state1[ 3 ][ j3 ], state1[ 3 ][ j3 ] ); \
+   t1 = _mm512_srli_epi16( state1[ 3 ][ j3 ], 7 ); \
+   t1 = _mm512_and_si512( t1, lsbmask ); \
+   t2 = _mm512_shuffle_epi8( mul2mask, t1 ); \
+   s2 = _mm512_xor_si512( s2, t2 ); \
+   state2[ 0 ][ j ] = _mm512_xor_si512( state2[ 0 ][ j ], state1[ 3 ][ j3 ] ); \
+   state2[ 1 ][ j ] = _mm512_xor_si512( state2[ 1 ][ j ], state1[ 3 ][ j3 ] ); \
+   state2[ 2 ][ j ] = _mm512_xor_si512( state2[ 2 ][ j ], \
+                            _mm512_xor_si512( s2, state1[ 3 ][ j3] ) ); \
+   state2[ 3 ][ j ] = _mm512_xor_si512( state2[ 3 ][ j ], s2 ); \
+} while(0)
+
+#define ECHO_ROUND_UNROLL2 \
+   ECHO_SUBBYTES(_state, 0, 0);\
+   ECHO_SUBBYTES(_state, 1, 0);\
+   ECHO_SUBBYTES(_state, 2, 0);\
+   ECHO_SUBBYTES(_state, 3, 0);\
+   ECHO_SUBBYTES(_state, 0, 1);\
+   ECHO_SUBBYTES(_state, 1, 1);\
+   ECHO_SUBBYTES(_state, 2, 1);\
+   ECHO_SUBBYTES(_state, 3, 1);\
+   ECHO_SUBBYTES(_state, 0, 2);\
+   ECHO_SUBBYTES(_state, 1, 2);\
+   ECHO_SUBBYTES(_state, 2, 2);\
+   ECHO_SUBBYTES(_state, 3, 2);\
+   ECHO_SUBBYTES(_state, 0, 3);\
+   ECHO_SUBBYTES(_state, 1, 3);\
+   ECHO_SUBBYTES(_state, 2, 3);\
+   ECHO_SUBBYTES(_state, 3, 3);\
+   ECHO_MIXBYTES(_state, _state2, 0, t1, t2, s2);\
+   ECHO_MIXBYTES(_state, _state2, 1, t1, t2, s2);\
+   ECHO_MIXBYTES(_state, _state2, 2, t1, t2, s2);\
+   ECHO_MIXBYTES(_state, _state2, 3, t1, t2, s2);\
+   ECHO_SUBBYTES(_state2, 0, 0);\
+   ECHO_SUBBYTES(_state2, 1, 0);\
+   ECHO_SUBBYTES(_state2, 2, 0);\
+   ECHO_SUBBYTES(_state2, 3, 0);\
+   ECHO_SUBBYTES(_state2, 0, 1);\
+   ECHO_SUBBYTES(_state2, 1, 1);\
+   ECHO_SUBBYTES(_state2, 2, 1);\
+   ECHO_SUBBYTES(_state2, 3, 1);\
+   ECHO_SUBBYTES(_state2, 0, 2);\
+   ECHO_SUBBYTES(_state2, 1, 2);\
+   ECHO_SUBBYTES(_state2, 2, 2);\
+   ECHO_SUBBYTES(_state2, 3, 2);\
+   ECHO_SUBBYTES(_state2, 0, 3);\
+   ECHO_SUBBYTES(_state2, 1, 3);\
+   ECHO_SUBBYTES(_state2, 2, 3);\
+   ECHO_SUBBYTES(_state2, 3, 3);\
+   ECHO_MIXBYTES(_state2, _state, 0, t1, t2, s2);\
+   ECHO_MIXBYTES(_state2, _state, 1, t1, t2, s2);\
+   ECHO_MIXBYTES(_state2, _state, 2, t1, t2, s2);\
+   ECHO_MIXBYTES(_state2, _state, 3, t1, t2, s2)
+
+#define SAVESTATE(dst, src)\
+	dst[0][0] = src[0][0];\
+	dst[0][1] = src[0][1];\
+	dst[0][2] = src[0][2];\
+	dst[0][3] = src[0][3];\
+	dst[1][0] = src[1][0];\
+	dst[1][1] = src[1][1];\
+	dst[1][2] = src[1][2];\
+	dst[1][3] = src[1][3];\
+	dst[2][0] = src[2][0];\
+	dst[2][1] = src[2][1];\
+	dst[2][2] = src[2][2];\
+	dst[2][3] = src[2][3];\
+	dst[3][0] = src[3][0];\
+	dst[3][1] = src[3][1];\
+	dst[3][2] = src[3][2];\
+	dst[3][3] = src[3][3]
+
+// blockcount always 1
+void echo_4way_compress( echo_4way_context *ctx, const __m512i *pmsg,
+               unsigned int uBlockCount )
+{
+  unsigned int r, b, i, j;
+  __m512i t1, t2, s2, k1;
+  __m512i _state[4][4], _state2[4][4], _statebackup[4][4]; 
+  __m512i one = m512_one_128;
+  __m512i mul2mask = m512_const2_64( 0, 0x00001b00 );
+  __m512i lsbmask  = m512_const1_32( 0x01010101 ); 
+
+  _state[ 0 ][ 0 ] = ctx->state[ 0 ][ 0 ];
+  _state[ 0 ][ 1 ] = ctx->state[ 0 ][ 1 ];
+  _state[ 0 ][ 2 ] = ctx->state[ 0 ][ 2 ];
+  _state[ 0 ][ 3 ] = ctx->state[ 0 ][ 3 ];
+  _state[ 1 ][ 0 ] = ctx->state[ 1 ][ 0 ];
+  _state[ 1 ][ 1 ] = ctx->state[ 1 ][ 1 ];
+  _state[ 1 ][ 2 ] = ctx->state[ 1 ][ 2 ];
+  _state[ 1 ][ 3 ] = ctx->state[ 1 ][ 3 ];
+  _state[ 2 ][ 0 ] = ctx->state[ 2 ][ 0 ];
+  _state[ 2 ][ 1 ] = ctx->state[ 2 ][ 1 ];
+  _state[ 2 ][ 2 ] = ctx->state[ 2 ][ 2 ];
+  _state[ 2 ][ 3 ] = ctx->state[ 2 ][ 3 ];
+  _state[ 3 ][ 0 ] = ctx->state[ 3 ][ 0 ];
+  _state[ 3 ][ 1 ] = ctx->state[ 3 ][ 1 ];
+  _state[ 3 ][ 2 ] = ctx->state[ 3 ][ 2 ];
+  _state[ 3 ][ 3 ] = ctx->state[ 3 ][ 3 ];
+
+  for ( b = 0; b < uBlockCount; b++ )
+  {
+    ctx->k = _mm512_add_epi64( ctx->k, ctx->const1536 );
+
+    for( j = ctx->uHashSize / 256; j < 4; j++ )
+    {
+      for ( i = 0; i < 4; i++ )
+	   {
+        _state[ i ][ j ] = _mm512_load_si512( 
+                     pmsg + 4 * (j - (ctx->uHashSize / 256)) + i );
+	   }
+	 }
+    
+    // save state
+	 SAVESTATE( _statebackup, _state );
+
+	 k1 = ctx->k;
+
+	 for ( r = 0; r < ctx->uRounds / 2; r++ )
+	 {
+		ECHO_ROUND_UNROLL2;
+	 }
+		
+	 if ( ctx->uHashSize == 256 )
+	 {
+	   for ( i = 0; i < 4; i++ )
+	   {
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _state[ i ][ 1 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _state[ i ][ 2 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _state[ i ][ 3 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 0 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 1 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 2 ] ) ;
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 3 ] );
+	   }
+	 }
+	 else
+	 {
+	   for ( i = 0; i < 4; i++ )
+	   {
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _state[ i ][ 2 ] );
+		   _state[ i ][ 1 ] = _mm512_xor_si512( _state[ i ][ 1 ],
+                                              _state[ i ][ 3 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 0 ] );
+		   _state[ i ][ 0 ] = _mm512_xor_si512( _state[ i ] [0 ],
+                                              _statebackup[ i ][ 2 ] );
+		   _state[ i ][ 1 ] = _mm512_xor_si512( _state[ i ][ 1 ],
+                                              _statebackup[ i ][ 1 ] );
+		   _state[ i ][ 1 ] = _mm512_xor_si512( _state[ i ][ 1 ],
+                                              _statebackup[ i ][ 3 ] );
+      }
+	 }
+    pmsg += ctx->uBlockLength;
+  }
+  SAVESTATE(ctx->state, _state);
+
+}
+
+int echo_4way_init( echo_4way_context *ctx, int nHashSize )
+{
+   int i, j;
+
+   ctx->k = m512_zero; 
+   ctx->processed_bits = 0;
+   ctx->uBufferBytes = 0;
+
+   switch( nHashSize )
+   {
+	case 256:
+		ctx->uHashSize = 256;
+		ctx->uBlockLength = 192;
+		ctx->uRounds = 8;
+		ctx->hashsize = m512_const2_64( 0, 0x100 );
+		ctx->const1536 = m512_const2_64( 0, 0x600 );
+		break;
+
+	case 512:
+		ctx->uHashSize = 512;
+		ctx->uBlockLength = 128;
+		ctx->uRounds = 10;
+		ctx->hashsize = m512_const2_64( 0, 0x200 );
+		ctx->const1536 = m512_const2_64( 0, 0x400);
+		break;
+
+	default:
+        	return 1;
+   }
+
+   for( i = 0; i < 4; i++ )
+	for( j = 0; j < nHashSize / 256; j++ )
+		ctx->state[ i ][ j ] = ctx->hashsize;
+
+   for( i = 0; i < 4; i++ )
+	for( j = nHashSize / 256; j < 4; j++ )
+		ctx->state[ i ][ j ] = m512_zero;
+
+   return 0;
+}
+
+int echo_4way_update_close( echo_4way_context *state, void *hashval,
+                              const void *data, int databitlen )
+{
+// bytelen is either 32 (maybe), 64 or 80 or 128!
+// all are less than full block.
+
+   int vlen = databitlen / 128;  // * 4 lanes / 128 bits per lane
+   const int vblen = state->uBlockLength / 16; //  16 bytes per lane
+   __m512i remainingbits;
+
+   if ( databitlen == 1024 )
+   {
+      echo_4way_compress( state, data, 1 );
+      state->processed_bits = 1024;
+      remainingbits = m512_const2_64( 0, -1024 );
+      vlen = 0;
+   }
+   else
+   {
+      vlen = databitlen / 128;  // * 4 lanes / 128 bits per lane
+      memcpy_512( state->buffer, data, vlen );
+      state->processed_bits += (unsigned int)( databitlen );
+      remainingbits = m512_const2_64( 0, (uint64_t)databitlen );
+   }
+
+   state->buffer[ vlen ] = m512_const2_64( 0, 0x80 );
+   memset_zero_512( state->buffer + vlen + 1, vblen - vlen - 2 );
+   state->buffer[ vblen-2 ] = m512_const2_64( (uint64_t)state->uHashSize << 48, 0 );
+   state->buffer[ vblen-1 ] = m512_const2_64( 0, state->processed_bits);
+
+   state->k = _mm512_add_epi64( state->k, remainingbits );
+   state->k = _mm512_sub_epi64( state->k, state->const1536 );
+
+   echo_4way_compress( state, state->buffer, 1 );
+
+   _mm512_store_si512( (__m512i*)hashval + 0, state->state[ 0 ][ 0] );
+   _mm512_store_si512( (__m512i*)hashval + 1, state->state[ 1 ][ 0] );
+
+   if ( state->uHashSize == 512 )
+   {
+      _mm512_store_si512( (__m512i*)hashval + 2, state->state[ 2 ][ 0 ] );
+      _mm512_store_si512( (__m512i*)hashval + 3, state->state[ 3 ][ 0 ] );
+   }
+   return 0;
+}
+
+int echo_4way_full( echo_4way_context *ctx, void *hashval, int nHashSize, 
+                    const void *data, int datalen )
+{
+   int i, j;
+   int databitlen = datalen * 8;
+   ctx->k = m512_zero;
+   ctx->processed_bits = 0;
+   ctx->uBufferBytes = 0;
+
+   switch( nHashSize )
+   {
+      case 256:
+         ctx->uHashSize = 256;
+         ctx->uBlockLength = 192;
+         ctx->uRounds = 8;
+         ctx->hashsize = m512_const2_64( 0, 0x100 );
+         ctx->const1536 = m512_const2_64( 0, 0x600 );
+         break;
+
+      case 512:
+         ctx->uHashSize = 512;
+         ctx->uBlockLength = 128;
+         ctx->uRounds = 10;
+         ctx->hashsize = m512_const2_64( 0, 0x200 );
+         ctx->const1536 = m512_const2_64( 0, 0x400 );
+         break;
+
+      default:
+         return 1;
+   }
+
+   for( i = 0; i < 4; i++ )
+      for( j = 0; j < nHashSize / 256; j++ )
+         ctx->state[ i ][ j ] = ctx->hashsize;
+
+   for( i = 0; i < 4; i++ )
+      for( j = nHashSize / 256; j < 4; j++ )
+         ctx->state[ i ][ j ] = m512_zero;
+
+   
+// bytelen is either 32 (maybe), 64 or 80 or 128!
+// all are less than full block.
+
+   int vlen = datalen / 32;  
+   const int vblen = ctx->uBlockLength / 16; //  16 bytes per lane
+   __m512i remainingbits;
+
+   if ( databitlen == 1024 )
+   {
+      echo_4way_compress( ctx, data, 1 );
+      ctx->processed_bits = 1024;
+      remainingbits = m512_const2_64( 0, -1024 );
+      vlen = 0;
+   }
+   else
+   {
+      vlen = databitlen / 128;  // * 4 lanes / 128 bits per lane
+      memcpy_512( ctx->buffer, data, vlen );
+      ctx->processed_bits += (unsigned int)( databitlen );
+      remainingbits = m512_const2_64( 0, databitlen );
+   }
+
+   ctx->buffer[ vlen ] = m512_const2_64( 0, 0x80 );
+   memset_zero_512( ctx->buffer + vlen + 1, vblen - vlen - 2 );
+   ctx->buffer[ vblen-2 ] =
+                     m512_const2_64( (uint64_t)ctx->uHashSize << 48, 0 );
+   ctx->buffer[ vblen-1 ] = m512_const2_64( 0, ctx->processed_bits);
+
+   ctx->k = _mm512_add_epi64( ctx->k, remainingbits );
+   ctx->k = _mm512_sub_epi64( ctx->k, ctx->const1536 );
+
+   echo_4way_compress( ctx, ctx->buffer, 1 );
+
+   _mm512_store_si512( (__m512i*)hashval + 0, ctx->state[ 0 ][ 0] );
+   _mm512_store_si512( (__m512i*)hashval + 1, ctx->state[ 1 ][ 0] );
+
+   if ( ctx->uHashSize == 512 )
+   {
+      _mm512_store_si512( (__m512i*)hashval + 2, ctx->state[ 2 ][ 0 ] );
+      _mm512_store_si512( (__m512i*)hashval + 3, ctx->state[ 3 ][ 0 ] );
+   }
+   return 0;
+}
+
+#endif  // AVX512
+
+// AVX2 + VAES
+
+#define mul2mask_2way   m256_const2_64( 0, 0x0000000000001b00 ) 
+
+#define lsbmask_2way    m256_const1_32( 0x01010101 ) 
+
+#define ECHO_SUBBYTES_2WAY( state, i, j ) \
+        state[i][j] = _mm256_aesenc_epi128( state[i][j], k1 ); \
+        k1 = _mm256_add_epi32( k1, m256_one_128 ); \
+        state[i][j] = _mm256_aesenc_epi128( state[i][j], m256_zero ); \
+
+#define ECHO_MIXBYTES_2WAY( state1, state2, j, t1, t2, s2 ) do \
+{ \
+   const int j1 = ( (j)+1 ) & 3; \
+   const int j2 = ( (j)+2 ) & 3; \
+   const int j3 = ( (j)+3 ) & 3; \
+   s2 = _mm256_add_epi8( state1[ 0 ] [j ], state1[ 0 ][ j ] ); \
+   t1 = _mm256_srli_epi16( state1[ 0 ][ j ], 7 ); \
+   t1 = _mm256_and_si256( t1, lsbmask_2way );\
+   t2 = _mm256_shuffle_epi8( mul2mask_2way, t1 ); \
+   s2 = _mm256_xor_si256( s2, t2 ); \
+   state2[ 0 ] [j ] = s2; \
+   state2[ 1 ] [j ] = state1[ 0 ][ j ]; \
+   state2[ 2 ] [j ] = state1[ 0 ][ j ]; \
+   state2[ 3 ] [j ] = _mm256_xor_si256( s2, state1[ 0 ][ j ] );\
+   s2 = _mm256_add_epi8( state1[ 1 ][ j1 ], state1[ 1 ][ j1 ] ); \
+   t1 = _mm256_srli_epi16( state1[ 1 ][ j1 ], 7 ); \
+   t1 = _mm256_and_si256( t1, lsbmask_2way ); \
+   t2 = _mm256_shuffle_epi8( mul2mask_2way, t1 ); \
+   s2 = _mm256_xor_si256( s2, t2 );\
+   state2[ 0 ][ j ] = _mm256_xor_si256( state2[ 0 ][ j ], \
+                              _mm256_xor_si256( s2, state1[ 1 ][ j1 ] ) ); \
+   state2[ 1 ][ j ] = _mm256_xor_si256( state2[ 1 ][ j ], s2 ); \
+   state2[ 2 ][ j ] = _mm256_xor_si256( state2[ 2 ][ j ], state1[ 1 ][ j1 ] ); \
+   state2[ 3 ][ j ] = _mm256_xor_si256( state2[ 3 ][ j ], state1[ 1 ][ j1 ] ); \
+   s2 = _mm256_add_epi8( state1[ 2 ][ j2 ], state1[ 2 ][ j2 ] ); \
+   t1 = _mm256_srli_epi16( state1[ 2 ][ j2 ], 7 ); \
+   t1 = _mm256_and_si256( t1, lsbmask_2way ); \
+   t2 = _mm256_shuffle_epi8( mul2mask_2way, t1 ); \
+   s2 = _mm256_xor_si256( s2, t2 ); \
+   state2[ 0 ][ j ] = _mm256_xor_si256( state2[ 0 ][ j ], state1[ 2 ][ j2 ] ); \
+   state2[ 1 ][ j ] = _mm256_xor_si256( state2[ 1 ][ j ], \
+                            _mm256_xor_si256( s2, state1[ 2 ][ j2 ] ) ); \
+   state2[ 2 ][ j ] = _mm256_xor_si256( state2[ 2 ][ j ], s2 ); \
+   state2[ 3 ][ j ] = _mm256_xor_si256( state2[ 3][ j ], state1[ 2 ][ j2 ] ); \
+   s2 = _mm256_add_epi8( state1[ 3 ][ j3 ], state1[ 3 ][ j3 ] ); \
+   t1 = _mm256_srli_epi16( state1[ 3 ][ j3 ], 7 ); \
+   t1 = _mm256_and_si256( t1, lsbmask_2way ); \
+   t2 = _mm256_shuffle_epi8( mul2mask_2way, t1 ); \
+   s2 = _mm256_xor_si256( s2, t2 ); \
+   state2[ 0 ][ j ] = _mm256_xor_si256( state2[ 0 ][ j ], state1[ 3 ][ j3 ] ); \
+   state2[ 1 ][ j ] = _mm256_xor_si256( state2[ 1 ][ j ], state1[ 3 ][ j3 ] ); \
+   state2[ 2 ][ j ] = _mm256_xor_si256( state2[ 2 ][ j ], \
+                            _mm256_xor_si256( s2, state1[ 3 ][ j3] ) ); \
+   state2[ 3 ][ j ] = _mm256_xor_si256( state2[ 3 ][ j ], s2 ); \
+} while(0)
+
+#define ECHO_ROUND_UNROLL2_2WAY \
+   ECHO_SUBBYTES_2WAY(_state, 0, 0);\
+   ECHO_SUBBYTES_2WAY(_state, 1, 0);\
+   ECHO_SUBBYTES_2WAY(_state, 2, 0);\
+   ECHO_SUBBYTES_2WAY(_state, 3, 0);\
+   ECHO_SUBBYTES_2WAY(_state, 0, 1);\
+   ECHO_SUBBYTES_2WAY(_state, 1, 1);\
+   ECHO_SUBBYTES_2WAY(_state, 2, 1);\
+   ECHO_SUBBYTES_2WAY(_state, 3, 1);\
+   ECHO_SUBBYTES_2WAY(_state, 0, 2);\
+   ECHO_SUBBYTES_2WAY(_state, 1, 2);\
+   ECHO_SUBBYTES_2WAY(_state, 2, 2);\
+   ECHO_SUBBYTES_2WAY(_state, 3, 2);\
+   ECHO_SUBBYTES_2WAY(_state, 0, 3);\
+   ECHO_SUBBYTES_2WAY(_state, 1, 3);\
+   ECHO_SUBBYTES_2WAY(_state, 2, 3);\
+   ECHO_SUBBYTES_2WAY(_state, 3, 3);\
+   ECHO_MIXBYTES_2WAY(_state, _state2, 0, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state, _state2, 1, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state, _state2, 2, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state, _state2, 3, t1, t2, s2);\
+   ECHO_SUBBYTES_2WAY(_state2, 0, 0);\
+   ECHO_SUBBYTES_2WAY(_state2, 1, 0);\
+   ECHO_SUBBYTES_2WAY(_state2, 2, 0);\
+   ECHO_SUBBYTES_2WAY(_state2, 3, 0);\
+   ECHO_SUBBYTES_2WAY(_state2, 0, 1);\
+   ECHO_SUBBYTES_2WAY(_state2, 1, 1);\
+   ECHO_SUBBYTES_2WAY(_state2, 2, 1);\
+   ECHO_SUBBYTES_2WAY(_state2, 3, 1);\
+   ECHO_SUBBYTES_2WAY(_state2, 0, 2);\
+   ECHO_SUBBYTES_2WAY(_state2, 1, 2);\
+   ECHO_SUBBYTES_2WAY(_state2, 2, 2);\
+   ECHO_SUBBYTES_2WAY(_state2, 3, 2);\
+   ECHO_SUBBYTES_2WAY(_state2, 0, 3);\
+   ECHO_SUBBYTES_2WAY(_state2, 1, 3);\
+   ECHO_SUBBYTES_2WAY(_state2, 2, 3);\
+   ECHO_SUBBYTES_2WAY(_state2, 3, 3);\
+   ECHO_MIXBYTES_2WAY(_state2, _state, 0, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state2, _state, 1, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state2, _state, 2, t1, t2, s2);\
+   ECHO_MIXBYTES_2WAY(_state2, _state, 3, t1, t2, s2)
+
+#define SAVESTATE_2WAY(dst, src)\
+        dst[0][0] = src[0][0];\
+        dst[0][1] = src[0][1];\
+        dst[0][2] = src[0][2];\
+        dst[0][3] = src[0][3];\
+        dst[1][0] = src[1][0];\
+        dst[1][1] = src[1][1];\
+        dst[1][2] = src[1][2];\
+        dst[1][3] = src[1][3];\
+        dst[2][0] = src[2][0];\
+        dst[2][1] = src[2][1];\
+        dst[2][2] = src[2][2];\
+        dst[2][3] = src[2][3];\
+        dst[3][0] = src[3][0];\
+        dst[3][1] = src[3][1];\
+        dst[3][2] = src[3][2];\
+        dst[3][3] = src[3][3]
+
+// blockcount always 1
+void echo_2way_compress( echo_2way_context *ctx, const __m256i *pmsg,
+               unsigned int uBlockCount )
+{
+  unsigned int r, b, i, j;
+  __m256i t1, t2, s2, k1;
+  __m256i _state[4][4], _state2[4][4], _statebackup[4][4];
+
+  _state[ 0 ][ 0 ] = ctx->state[ 0 ][ 0 ];
+  _state[ 0 ][ 1 ] = ctx->state[ 0 ][ 1 ];
+  _state[ 0 ][ 2 ] = ctx->state[ 0 ][ 2 ];
+  _state[ 0 ][ 3 ] = ctx->state[ 0 ][ 3 ];
+  _state[ 1 ][ 0 ] = ctx->state[ 1 ][ 0 ];
+  _state[ 1 ][ 1 ] = ctx->state[ 1 ][ 1 ];
+  _state[ 1 ][ 2 ] = ctx->state[ 1 ][ 2 ];
+  _state[ 1 ][ 3 ] = ctx->state[ 1 ][ 3 ];
+  _state[ 2 ][ 0 ] = ctx->state[ 2 ][ 0 ];
+  _state[ 2 ][ 1 ] = ctx->state[ 2 ][ 1 ];
+  _state[ 2 ][ 2 ] = ctx->state[ 2 ][ 2 ];
+  _state[ 2 ][ 3 ] = ctx->state[ 2 ][ 3 ];
+  _state[ 3 ][ 0 ] = ctx->state[ 3 ][ 0 ];
+  _state[ 3 ][ 1 ] = ctx->state[ 3 ][ 1 ];
+  _state[ 3 ][ 2 ] = ctx->state[ 3 ][ 2 ];
+  _state[ 3 ][ 3 ] = ctx->state[ 3 ][ 3 ];
+
+  for ( b = 0; b < uBlockCount; b++ )
+  {
+    ctx->k = _mm256_add_epi64( ctx->k, ctx->const1536 );
+
+    for( j = ctx->uHashSize / 256; j < 4; j++ )
+    {
+      for ( i = 0; i < 4; i++ )
+      {
+        _state[ i ][ j ] = _mm256_load_si256(
+                     pmsg + 4 * (j - (ctx->uHashSize / 256)) + i );
+      }
+   }
+
+   // save state
+   SAVESTATE_2WAY( _statebackup, _state );
+
+   k1 = ctx->k;
+
+   for ( r = 0; r < ctx->uRounds / 2; r++ )
+   {
+       ECHO_ROUND_UNROLL2_2WAY;
+   }
+
+   if ( ctx->uHashSize == 256 )
+   {
+      for ( i = 0; i < 4; i++ )
+      {
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _state[ i ][ 1 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _state[ i ][ 2 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _state[ i ][ 3 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 0 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 1 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 2 ] ) ;
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 3 ] );
+       }
+    }
+    else
+    {
+       for ( i = 0; i < 4; i++ )
+       {
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _state[ i ][ 2 ] );
+         _state[ i ][ 1 ] = _mm256_xor_si256( _state[ i ][ 1 ],
+                                              _state[ i ][ 3 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ][ 0 ],
+                                              _statebackup[ i ][ 0 ] );
+         _state[ i ][ 0 ] = _mm256_xor_si256( _state[ i ] [0 ],
+                                              _statebackup[ i ][ 2 ] );
+         _state[ i ][ 1 ] = _mm256_xor_si256( _state[ i ][ 1 ],
+                                              _statebackup[ i ][ 1 ] );
+         _state[ i ][ 1 ] = _mm256_xor_si256( _state[ i ][ 1 ],
+                                              _statebackup[ i ][ 3 ] );
+      }
+    }
+    pmsg += ctx->uBlockLength;
+  }
+  SAVESTATE_2WAY(ctx->state, _state);
+
+}
+int echo_2way_init( echo_2way_context *ctx, int nHashSize )
+{
+        int i, j;
+
+   ctx->k = m256_zero;
+   ctx->processed_bits = 0;
+   ctx->uBufferBytes = 0;
+
+   switch( nHashSize )
+   {
+                case 256:
+                        ctx->uHashSize = 256;
+                        ctx->uBlockLength = 192;
+                        ctx->uRounds = 8;
+                        ctx->hashsize = m256_const2_64( 0, 0x100 );
+                        ctx->const1536 = m256_const2_64( 0, 0x600 );
+                        break;
+
+                case 512:
+                        ctx->uHashSize = 512;
+                        ctx->uBlockLength = 128;
+                        ctx->uRounds = 10;
+                        ctx->hashsize = m256_const2_64( 0, 0x200 );
+                        ctx->const1536 = m256_const2_64( 0, 0x400 );
+                        break;
+
+                default:
+                        return 1;
+        }
+
+        for( i = 0; i < 4; i++ )
+                for( j = 0; j < nHashSize / 256; j++ )
+                        ctx->state[ i ][ j ] = ctx->hashsize;
+
+        for( i = 0; i < 4; i++ )
+                for( j = nHashSize / 256; j < 4; j++ )
+                        ctx->state[ i ][ j ] = m256_zero;
+
+        return 0;
+}
+
+int echo_2way_update_close( echo_2way_context *state, void *hashval,
+                              const void *data, int databitlen )
+{
+// bytelen is either 32 (maybe), 64 or 80 or 128!
+// all are less than full block.
+
+   int vlen = databitlen / 128;  // * 4 lanes / 128 bits per lane
+   const int vblen = state->uBlockLength / 16; //  16 bytes per lane
+   __m256i remainingbits;
+
+   if ( databitlen == 1024 )
+   {
+      echo_2way_compress( state, data, 1 );
+      state->processed_bits = 1024;
+      remainingbits = m256_const2_64( 0, -1024 );
+      vlen = 0;
+   }
+   else
+   {
+      memcpy_256( state->buffer, data, vlen );
+      state->processed_bits += (unsigned int)( databitlen );
+      remainingbits = m256_const2_64( 0, databitlen );
+   }
+
+   state->buffer[ vlen ] = m256_const2_64( 0, 0x80 );
+   memset_zero_256( state->buffer + vlen + 1, vblen - vlen - 2 );
+   state->buffer[ vblen-2 ] = m256_const2_64( (uint64_t)state->uHashSize << 48, 0 );
+   state->buffer[ vblen-1 ] = m256_const2_64( 0, state->processed_bits );
+
+   state->k = _mm256_add_epi64( state->k, remainingbits );
+   state->k = _mm256_sub_epi64( state->k, state->const1536 );
+
+   echo_2way_compress( state, state->buffer, 1 );
+
+   _mm256_store_si256( (__m256i*)hashval + 0, state->state[ 0 ][ 0] );
+   _mm256_store_si256( (__m256i*)hashval + 1, state->state[ 1 ][ 0] );
+
+   if ( state->uHashSize == 512 )
+   {
+      _mm256_store_si256( (__m256i*)hashval + 2, state->state[ 2 ][ 0 ] );
+      _mm256_store_si256( (__m256i*)hashval + 3, state->state[ 3 ][ 0 ] );
+   }
+   return 0;
+}
+
+int echo_2way_full( echo_2way_context *ctx, void *hashval, int nHashSize,
+                    const void *data, int datalen )
+{
+   int i, j;
+   int databitlen = datalen * 8;
+   ctx->k = m256_zero;
+   ctx->processed_bits = 0;
+   ctx->uBufferBytes = 0;
+
+   switch( nHashSize )
+   {
+      case 256:
+         ctx->uHashSize = 256;
+         ctx->uBlockLength = 192;
+         ctx->uRounds = 8;
+         ctx->hashsize = m256_const2_64( 0, 0x100 );
+         ctx->const1536 = m256_const2_64( 0, 0x600 );
+         break;
+
+      case 512:
+         ctx->uHashSize = 512;
+         ctx->uBlockLength = 128;
+         ctx->uRounds = 10;
+         ctx->hashsize = m256_const2_64( 0, 0x200 );
+         ctx->const1536 = m256_const2_64( 0, 0x400 );
+         break;
+
+      default:
+         return 1;
+   }
+
+   for( i = 0; i < 4; i++ )
+      for( j = 0; j < nHashSize / 256; j++ )
+         ctx->state[ i ][ j ] = ctx->hashsize;
+
+   for( i = 0; i < 4; i++ )
+      for( j = nHashSize / 256; j < 4; j++ )
+         ctx->state[ i ][ j ] = m256_zero;
+
+   int vlen = datalen / 32;
+   const int vblen = ctx->uBlockLength / 16; //  16 bytes per lane
+   __m256i remainingbits;
+
+   if ( databitlen == 1024 )
+   {
+      echo_2way_compress( ctx, data, 1 );
+      ctx->processed_bits = 1024;
+      remainingbits = m256_const2_64( 0, -1024 );
+      vlen = 0;
+   }
+   else
+   {
+      vlen = databitlen / 128;  // * 4 lanes / 128 bits per lane
+      memcpy_256( ctx->buffer, data, vlen );
+      ctx->processed_bits += (unsigned int)( databitlen );
+      remainingbits = m256_const2_64( 0, databitlen );
+   }
+
+   ctx->buffer[ vlen ] = m256_const2_64( 0, 0x80 );
+   memset_zero_256( ctx->buffer + vlen + 1, vblen - vlen - 2 );
+   ctx->buffer[ vblen-2 ] = m256_const2_64( (uint64_t)ctx->uHashSize << 48, 0 );
+   ctx->buffer[ vblen-1 ] = m256_const2_64( 0, ctx->processed_bits );
+
+   ctx->k = _mm256_add_epi64( ctx->k, remainingbits );
+   ctx->k = _mm256_sub_epi64( ctx->k, ctx->const1536 );
+
+   echo_2way_compress( ctx, ctx->buffer, 1 );
+
+   _mm256_store_si256( (__m256i*)hashval + 0, ctx->state[ 0 ][ 0] );
+   _mm256_store_si256( (__m256i*)hashval + 1, ctx->state[ 1 ][ 0] );
+
+   if ( ctx->uHashSize == 512 )
+   {
+      _mm256_store_si256( (__m256i*)hashval + 2, ctx->state[ 2 ][ 0 ] );
+      _mm256_store_si256( (__m256i*)hashval + 3, ctx->state[ 3 ][ 0 ] );
+   }
+   return 0;
+}
+
+
+#endif   // VAES
--- a/algo/echo/echo-hash-4way.h
+++ b/algo/echo/echo-hash-4way.h
@@ -0,0 +1,90 @@
+#if !defined(ECHO_HASH_4WAY_H__)
+#define ECHO_HASH_4WAY_H__ 1
+
+#if defined(__VAES__)
+
+#include "simd-utils.h"
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+
+typedef struct
+{
+   __m512i    state[4][4];
+   __m512i    buffer[ 4 * 192 / 16 ];  // 4x128 interleaved 192 bytes
+   __m512i    k;
+   __m512i    hashsize;
+   __m512i    const1536;
+
+   unsigned int   uRounds;
+   unsigned int   uHashSize;
+   unsigned int   uBlockLength;
+   unsigned int   uBufferBytes;
+   unsigned int   processed_bits;
+
+} echo_4way_context __attribute__ ((aligned (64)));
+#define echo512_4way_context echo_4way_context
+
+int echo_4way_init( echo_4way_context *state, int hashbitlen );
+#define echo512_4way_init( state ) echo_4way_init( state, 512 )
+#define echo256_4way_init( state ) echo_4way_init( state, 256 )
+
+int echo_4way_update( echo_4way_context *state, const void *data,
+    unsigned int databitlen);
+#define echo512_4way_update echo_4way_update
+
+// int echo_4way_close( echo_4way_context *state, void *hashval );
+// #define echo512_4way_close echo_4way_close
+
+int echo_4way_update_close( echo_4way_context *state, void *hashval,
+                              const void *data, int databitlen );
+#define echo512_4way_update_close echo_4way_update_close
+
+int echo_4way_full( echo_4way_context *ctx, void *hashval, int nHashSize,
+                    const void *data, int datalen );
+#define echo512_4way_full( state, hashval, data, datalen ) \
+           echo_4way_full( state, hashval, 512, data, datalen )
+#define echo256_4way_full( state, hashval, data, datalen ) \
+           echo_4way_full( state, hashval, 256, data, datalen )
+
+#endif   // AVX512
+
+typedef struct
+{
+   __m256i    state[4][4];
+   __m256i    buffer[ 4 * 192 / 16 ];  // 4x128 interleaved 192 bytes
+   __m256i    k;
+   __m256i    hashsize;
+   __m256i    const1536;
+
+   unsigned int   uRounds;
+   unsigned int   uHashSize;
+   unsigned int   uBlockLength;
+   unsigned int   uBufferBytes;
+   unsigned int   processed_bits;
+
+} echo_2way_context __attribute__ ((aligned (64)));
+#define echo512_2way_context echo_2way_context
+
+int echo_2way_init( echo_2way_context *state, int hashbitlen );
+#define echo512_2way_init( state ) echo_2way_init( state, 512 )
+#define echo256_2way_init( state ) echo_2way_init( state, 256 )
+
+int echo_2way_update( echo_2way_context *state, const void *data,
+    unsigned int databitlen);
+#define echo512_2way_update echo_2way_update
+
+int echo_2way_update_close( echo_2way_context *state, void *hashval,
+                              const void *data, int databitlen );
+#define echo512_2way_update_close echo_2way_update_close
+
+int echo_2way_full( echo_2way_context *ctx, void *hashval, int nHashSize,
+                    const void *data, int datalen );
+#define echo512_2way_full( state, hashval, data, datalen ) \
+           echo_2way_full( state, hashval, 512, data, datalen )
+#define echo256_2way_full( state, hashval, data, datalen ) \
+           echo_2way_full( state, hashval, 256, data, datalen )
+
+
+#endif  // VAES
+
+#endif   // ECHO_HASH_4WAY_H__
--- a/algo/echo/sph_echo.c
+++ b/algo/echo/sph_echo.c
@@ -36,6 +36,8 @@

 #include "sph_echo.h"

+#if !defined(__AES__)
+
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -1028,4 +1030,5 @@ sph_echo512_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
 }
 #ifdef __cplusplus
 }
-#endif
+#endif 
+#endif  // !AES
--- a/algo/echo/sph_echo.h
+++ b/algo/echo/sph_echo.h
@@ -36,6 +36,8 @@
 #ifndef SPH_ECHO_H__
 #define SPH_ECHO_H__

+#if !defined(__AES__)
+
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -316,5 +318,5 @@ void sph_echo512_addbits_and_close(
 #ifdef __cplusplus
 }
 #endif
-
+#endif // !AES
 #endif
--- a/algo/fugue/fugue-aesni.c
+++ b/algo/fugue/fugue-aesni.c
@@ -0,0 +1,565 @@
+/*
+ * file        : fugue_vperm.c
+ * version     : 1.0.208
+ * date        : 14.12.2010
+ * 
+ * - vperm and aes_ni implementations of hash function Fugue
+ * - implements NIST hash api
+ * - assumes that message lenght is multiple of 8-bits
+ * - _FUGUE_VPERM_ must be defined if compiling with ../main.c
+ * - default version is vperm, define AES_NI for aes_ni version
+ * 
+ * Cagdas Calik
+ * ccalik@metu.edu.tr
+ * Institute of Applied Mathematics, Middle East Technical University, Turkey.
+ *
+ */
+
+#if defined(__AES__)
+
+#include <x86intrin.h>
+
+#include <memory.h>
+#include "fugue-aesni.h"
+
+
+MYALIGN const unsigned long long _supermix1a[]	= {0x0202010807020100, 0x0a05000f06010c0b};
+MYALIGN const unsigned long long _supermix1b[]	= {0x0b0d080703060504, 0x0e0a090c050e0f0a};
+MYALIGN const unsigned long long _supermix1c[]	= {0x0402060c070d0003, 0x090a060580808080};
+MYALIGN const unsigned long long _supermix1d[]	= {0x808080800f0e0d0c, 0x0f0e0d0c80808080};
+MYALIGN const unsigned long long _supermix2a[]	= {0x07020d0880808080, 0x0b06010c050e0f0a};
+MYALIGN const unsigned long long _supermix4a[]	= {0x000f0a050c0b0601, 0x0302020404030e09};
+MYALIGN const unsigned long long _supermix4b[]	= {0x07020d08080e0d0d, 0x07070908050e0f0a};
+MYALIGN const unsigned long long _supermix4c[]	= {0x0706050403020000, 0x0302000007060504};
+MYALIGN const unsigned long long _supermix7a[]	= {0x010c0b060d080702, 0x0904030e03000104};
+MYALIGN const unsigned long long _supermix7b[]	= {0x8080808080808080, 0x0504070605040f06};
+MYALIGN const unsigned long long _k_n[] = {0x4E4E4E4E4E4E4E4E, 0x1B1B1B1B0E0E0E0E};
+MYALIGN const unsigned char _shift_one_mask[]   = {7, 4, 5, 6, 11, 8, 9, 10, 15, 12, 13, 14, 3, 0, 1, 2};
+MYALIGN const unsigned char _shift_four_mask[]  = {13, 14, 15, 12, 1, 2, 3, 0, 5, 6, 7, 4, 9, 10, 11, 8};
+MYALIGN const unsigned char _shift_seven_mask[] = {10, 11, 8, 9, 14, 15, 12, 13, 2, 3, 0, 1, 6, 7, 4, 5};
+MYALIGN const unsigned char _aes_shift_rows[]   = {0, 5, 10, 15, 4, 9, 14, 3, 8, 13, 2, 7, 12, 1, 6, 11};
+MYALIGN const unsigned int _inv_shift_rows[] = {0x070a0d00, 0x0b0e0104, 0x0f020508, 0x0306090c};
+MYALIGN const unsigned int _mul2mask[] = {0x1b1b0000, 0x00000000, 0x00000000, 0x00000000};
+MYALIGN const unsigned int _mul4mask[] = {0x2d361b00, 0x00000000, 0x00000000, 0x00000000};
+MYALIGN const unsigned int _lsbmask2[] = {0x03030303, 0x03030303, 0x03030303, 0x03030303};
+
+
+MYALIGN const unsigned int _IV512[] = {		
+	0x00000000, 0x00000000,	0x7ea50788, 0x00000000,
+	0x75af16e6, 0xdbe4d3c5, 0x27b09aac, 0x00000000,
+	0x17f115d9, 0x54cceeb6, 0x0b02e806, 0x00000000,
+	0xd1ef924a, 0xc9e2c6aa, 0x9813b2dd, 0x00000000,
+	0x3858e6ca, 0x3f207f43, 0xe778ea25, 0x00000000,
+	0xd6dd1f95, 0x1dd16eda, 0x67353ee1, 0x00000000};
+
+#if defined(__SSE4_1__)
+
+#define PACK_S0(s0, s1, t1)\
+   s0 = _mm_castps_si128(_mm_insert_ps(_mm_castsi128_ps(s0), _mm_castsi128_ps(s1), 0x30))
+
+#define UNPACK_S0(s0, s1, t1)\
+   s1 = _mm_castps_si128(_mm_insert_ps(_mm_castsi128_ps(s1), _mm_castsi128_ps(s0), 0xc0));\
+   s0 = mm128_mask_32( s0, 8 )
+
+#define CMIX(s1, s2, r1, r2, t1, t2)\
+   t1 = s1;\
+   t1 = _mm_castps_si128(_mm_shuffle_ps(_mm_castsi128_ps(t1), _mm_castsi128_ps(s2), _MM_SHUFFLE(3, 0, 2, 1)));\
+   r1 = _mm_xor_si128(r1, t1);\
+   r2 = _mm_xor_si128(r2, t1);
+
+#else   // SSE2
+
+#define PACK_S0(s0, s1, t1)\
+   t1 = _mm_shuffle_epi32(s1, _MM_SHUFFLE(0, 3, 3, 3));\
+   s0 = _mm_xor_si128(s0, t1);
+
+#define UNPACK_S0(s0, s1, t1)\
+   t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 3, 3));\
+   s1 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s1), _mm_castsi128_ps(t1)));\
+   s0 = mm128_mask_32( s0, 8 )
+
+#define CMIX(s1, s2, r1, r2, t1, t2)\
+   t1 = _mm_shuffle_epi32(s1, 0xf9);\
+   t2 = _mm_shuffle_epi32(s2, 0xcf);\
+   t1 = _mm_xor_si128(t1, t2);\
+   r1 = _mm_xor_si128(r1, t1);\
+   r2 = _mm_xor_si128(r2, t1)
+
+#endif
+
+#define TIX256(msg, s10, s8, s24, s0, t1, t2, t3)\
+	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
+	s10 = _mm_xor_si128(s10, t1);\
+	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
+	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
+	t1 = _mm_slli_si128(t1, 8);\
+	s8 = _mm_xor_si128(s8, t1);\
+	t1 = _mm_shuffle_epi32(s24, _MM_SHUFFLE(3, 3, 0, 3));\
+	s0 = _mm_xor_si128(s0, t1)
+
+
+#define TIX384(msg, s16, s8, s27, s30, s0, s4, t1, t2, t3)\
+	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
+	s16 = _mm_xor_si128(s16, t1);\
+	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
+	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
+	t1 = _mm_slli_si128(t1, 8);\
+	s8 = _mm_xor_si128(s8, t1);\
+	t1 = _mm_shuffle_epi32(s27, _MM_SHUFFLE(3, 3, 0, 3));\
+	s0 = _mm_xor_si128(s0, t1);\
+	t1 = _mm_shuffle_epi32(s30, _MM_SHUFFLE(3, 3, 0, 3));\
+	s4 = _mm_xor_si128(s4, t1)
+
+#define TIX512(msg, s22, s8, s24, s27, s30, s0, s4, s7, t1, t2, t3)\
+	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
+	s22 = _mm_xor_si128(s22, t1);\
+	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
+	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
+	t1 = _mm_slli_si128(t1, 8);\
+	s8 = _mm_xor_si128(s8, t1);\
+	t1 = _mm_shuffle_epi32(s24, _MM_SHUFFLE(3, 3, 0, 3));\
+	s0 = _mm_xor_si128(s0, t1);\
+	t1 = _mm_shuffle_epi32(s27, _MM_SHUFFLE(3, 3, 0, 3));\
+	s4 = _mm_xor_si128(s4, t1);\
+	t1 = _mm_shuffle_epi32(s30, _MM_SHUFFLE(3, 3, 0, 3));\
+	s7 = _mm_xor_si128(s7, t1)
+
+
+#define PRESUPERMIX(x, t1, s1, s2, t2)\
+	s1 = x;\
+	s2 = _mm_add_epi8(x, x);\
+	t2 = _mm_add_epi8(s2, s2);\
+	t1 = _mm_srli_epi16(x, 6);\
+	t1 = _mm_and_si128(t1, M128(_lsbmask2));\
+	s2 = _mm_xor_si128(s2, _mm_shuffle_epi8(M128(_mul2mask), t1));\
+	x  = _mm_xor_si128(t2, _mm_shuffle_epi8(M128(_mul4mask), t1))
+
+#define SUBSTITUTE(r0, _t1, _t2, _t3, _t0)\
+	_t2 = _mm_shuffle_epi8(r0, M128(_inv_shift_rows));\
+	_t2 = _mm_aesenclast_si128( _t2, m128_zero )
+	
+#define SUPERMIX(t0, t1, t2, t3, t4)\
+	PRESUPERMIX(t0, t1, t2, t3, t4);\
+	POSTSUPERMIX(t0, t1, t2, t3, t4)
+
+
+#define POSTSUPERMIX(t0, t1, t2, t3, t4)\
+	t1 = t2;\
+	t1 = _mm_shuffle_epi8(t1, M128(_supermix1b));\
+	t4 = t1;\
+	t1 = _mm_shuffle_epi8(t1, M128(_supermix1c));\
+	t4 = _mm_xor_si128(t4, t1);\
+	t1 = t4;\
+	t1 = _mm_shuffle_epi8(t1, M128(_supermix1d));\
+	t4 = _mm_xor_si128(t4, t1);\
+	t1 = t2;\
+	t1 = _mm_shuffle_epi8(t1, M128(_supermix1a));\
+	t4 = _mm_xor_si128(t4, t1);\
+	t2 = _mm_xor_si128(t2, t3);\
+	t2 = _mm_xor_si128(t2, t0);\
+	t2 = _mm_shuffle_epi8(t2, M128(_supermix7a));\
+	t4 = _mm_xor_si128(t4, t2);\
+	t2 = _mm_shuffle_epi8(t2, M128(_supermix7b));\
+	t4 = _mm_xor_si128(t4, t2);\
+	t3 = _mm_shuffle_epi8(t3, M128(_supermix2a));\
+	t1 = t0;\
+	t1 = _mm_shuffle_epi8(t1, M128(_supermix4a));\
+	t4 = _mm_xor_si128(t4, t1);\
+	t0 = _mm_shuffle_epi8(t0, M128(_supermix4b));\
+	t0 = _mm_xor_si128(t0, t3);\
+	t4 = _mm_xor_si128(t4, t0);\
+	t0 = _mm_shuffle_epi8(t0, M128(_supermix4c));\
+	t4 = _mm_xor_si128(t4, t0)
+
+
+#define SUBROUND512_3(r1a, r1b, r1c, r1d, r2a, r2b, r2c, r2d, r3a, r3b, r3c, r3d)\
+	CMIX(r1a, r1b, r1c, r1d, _t0, _t1);\
+	PACK_S0(r1c, r1a, _t0);\
+	SUBSTITUTE(r1c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r1c);\
+	_t0 = _mm_shuffle_epi32(r1c, 0x39);\
+	r2c = _mm_xor_si128(r2c, _t0);\
+   _t0 = mm128_mask_32( _t0, 8 ); \
+	r2d = _mm_xor_si128(r2d, _t0);\
+	UNPACK_S0(r1c, r1a, _t3);\
+	SUBSTITUTE(r2c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r2c);\
+	_t0 = _mm_shuffle_epi32(r2c, 0x39);\
+	r3c = _mm_xor_si128(r3c, _t0);\
+   _t0 = mm128_mask_32( _t0, 8 ); \
+	r3d = _mm_xor_si128(r3d, _t0);\
+	UNPACK_S0(r2c, r2a, _t3);\
+	SUBSTITUTE(r3c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r3c);\
+	UNPACK_S0(r3c, r3a, _t3)
+
+
+#define SUBROUND512_4(r1a, r1b, r1c, r1d, r2a, r2b, r2c, r2d, r3a, r3b, r3c, r3d, r4a, r4b, r4c, r4d)\
+	CMIX(r1a, r1b, r1c, r1d, _t0, _t1);\
+	PACK_S0(r1c, r1a, _t0);\
+	SUBSTITUTE(r1c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r1c);\
+	_t0 = _mm_shuffle_epi32(r1c, 0x39);\
+	r2c = _mm_xor_si128(r2c, _t0);\
+   _t0 = mm128_mask_32( _t0, 8 ); \
+	r2d = _mm_xor_si128(r2d, _t0);\
+	UNPACK_S0(r1c, r1a, _t3);\
+	SUBSTITUTE(r2c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r2c);\
+	_t0 = _mm_shuffle_epi32(r2c, 0x39);\
+	r3c = _mm_xor_si128(r3c, _t0);\
+   _t0 = mm128_mask_32( _t0, 8 ); \
+	r3d = _mm_xor_si128(r3d, _t0);\
+	UNPACK_S0(r2c, r2a, _t3);\
+	SUBSTITUTE(r3c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r3c);\
+	_t0 = _mm_shuffle_epi32(r3c, 0x39);\
+	r4c = _mm_xor_si128(r4c, _t0);\
+   _t0 = mm128_mask_32( _t0, 8 ); \
+	r4d = _mm_xor_si128(r4d, _t0);\
+	UNPACK_S0(r3c, r3a, _t3);\
+	SUBSTITUTE(r4c, _t1, _t2, _t3, _t0);\
+	SUPERMIX(_t2, _t3, _t0, _t1, r4c);\
+	UNPACK_S0(r4c, r4a, _t3)
+
+
+
+#define LOADCOLUMN(x, s, a)\
+	block[0] = col[(base + a + 0) % s];\
+	block[1] = col[(base + a + 1) % s];\
+	block[2] = col[(base + a + 2) % s];\
+	block[3] = col[(base + a + 3) % s];\
+	x = _mm_load_si128((__m128i*)block)
+
+#define STORECOLUMN(x, s)\
+	_mm_store_si128((__m128i*)block, x);\
+	col[(base + 0) % s] = block[0];\
+	col[(base + 1) % s] = block[1];\
+	col[(base + 2) % s] = block[2];\
+	col[(base + 3) % s] = block[3]
+
+void Compress512(hashState_fugue *ctx, const unsigned char *pmsg, unsigned int uBlockCount)
+{
+   __m128i _t0, _t1, _t2, _t3;
+
+   switch(ctx->base)
+   {
+      case 1:
+         TIX512( pmsg, ctx->state[3], ctx->state[10], ctx->state[4],
+                       ctx->state[5], ctx->state[ 6], ctx->state[8],
+		       ctx->state[9], ctx->state[10], _t0, _t1, _t2 );
+
+	 SUBROUND512_4( ctx->state[8], ctx->state[9], ctx->state[7],
+                        ctx->state[1], ctx->state[7], ctx->state[8],
+		       	ctx->state[6], ctx->state[0], ctx->state[6],
+		       	ctx->state[7], ctx->state[5], ctx->state[11],
+		       	ctx->state[5], ctx->state[6], ctx->state[4],
+		       	ctx->state[10] );
+         ctx->base++;
+         pmsg += 4;
+         uBlockCount--;
+      if( uBlockCount == 0 ) break;
+
+      case 2:
+         TIX512( pmsg, ctx->state[11], ctx->state[6], ctx->state[0],
+                       ctx->state[ 1], ctx->state[2], ctx->state[4],
+		       ctx->state[ 5], ctx->state[6], _t0, _t1, _t2);
+
+         SUBROUND512_4( ctx->state[4], ctx->state[5], ctx->state[3],
+                        ctx->state[9], ctx->state[3], ctx->state[4],
+		       	ctx->state[2], ctx->state[8], ctx->state[2],
+		       	ctx->state[3], ctx->state[1], ctx->state[7],
+		       	ctx->state[1], ctx->state[2], ctx->state[0],
+		       	ctx->state[6]);
+
+         ctx->base = 0;
+         pmsg += 4;
+         uBlockCount--;
+      break;
+   }
+
+
+   while( uBlockCount > 0 )
+   {
+      TIX512( pmsg, ctx->state[ 7], ctx->state[2], ctx->state[8], ctx->state[9],
+                    ctx->state[10], ctx->state[0], ctx->state[1], ctx->state[2],
+              _t0, _t1, _t2 );
+      SUBROUND512_4( ctx->state[0], ctx->state[1], ctx->state[11],
+                     ctx->state[5], ctx->state[11], ctx->state[0],
+		     ctx->state[10], ctx->state[4], ctx->state[10],
+		     ctx->state[11], ctx->state[9], ctx->state[3],
+		     ctx->state[9], ctx->state[10], ctx->state[8],
+		     ctx->state[2] );
+
+      ctx->base++;
+      pmsg += 4;
+      uBlockCount--;
+      if( uBlockCount == 0 ) break;
+
+      TIX512( pmsg, ctx->state[3], ctx->state[10], ctx->state[4], ctx->state[5],
+                    ctx->state[6], ctx->state[8], ctx->state[9], ctx->state[10],
+              _t0, _t1, _t2 );
+
+      SUBROUND512_4( ctx->state[8], ctx->state[9], ctx->state[7], ctx->state[1],                     ctx->state[7], ctx->state[8], ctx->state[6], ctx->state[0],
+		     ctx->state[6], ctx->state[7], ctx->state[5], ctx->state[11],
+		     ctx->state[5], ctx->state[6, ctx->state[4], ctx->state[10]);
+
+      ctx->base++;
+      pmsg += 4;
+      uBlockCount--;
+      if( uBlockCount == 0 ) break;
+
+      TIX512( pmsg, ctx->state[11], ctx->state[6], ctx->state[0], ctx->state[1],
+		    ctx->state[2], ctx->state[4], ctx->state[5], ctx->state[6],
+               _t0, _t1, _t2);
+      SUBROUND512_4( ctx->state[4], ctx->state[5], ctx->state[3], ctx->state[9],
+		     ctx->state[3], ctx->state[4], ctx->state[2], ctx->state[8],
+		     ctx->state[2], ctx->state[3], ctx->state[1], ctx->state[7],
+		     ctx->state[1], ctx->state[2], ctx->state[0], ctx->state[6]);
+
+      ctx->base = 0;
+      pmsg += 4;
+      uBlockCount--;
+   }
+
+}
+
+void Final512(hashState_fugue *ctx, BitSequence *hashval)
+{
+        unsigned int block[4] __attribute__ ((aligned (32)));
+        unsigned int col[36] __attribute__ ((aligned (16)));
+	unsigned int i, base;
+	__m128i r0, _t0, _t1, _t2, _t3;
+
+	for(i = 0; i < 12; i++)
+	{
+		_mm_store_si128((__m128i*)block, ctx->state[i]);
+
+		col[3 * i + 0] = block[0];
+		col[3 * i + 1] = block[1];
+		col[3 * i + 2] = block[2];
+	}
+
+	base = (36 - (12 * ctx->base)) % 36;
+
+	for(i = 0; i < 32; i++)
+	{
+		// ROR3
+		base = (base + 33) % 36;
+
+		// CMIX
+		col[(base +  0) % 36] ^= col[(base + 4) % 36];
+		col[(base +  1) % 36] ^= col[(base + 5) % 36];
+		col[(base +  2) % 36] ^= col[(base + 6) % 36];
+		col[(base +  18) % 36] ^= col[(base + 4) % 36];
+		col[(base +  19) % 36] ^= col[(base + 5) % 36];
+		col[(base +  20) % 36] ^= col[(base + 6) % 36];
+
+		// SMIX
+		LOADCOLUMN(r0, 36, 0);
+		SUBSTITUTE(r0, _t1, _t2, _t3, _t0);
+		SUPERMIX(_t2, _t3, _t0, _t1, r0);
+		STORECOLUMN(r0, 36);
+	}
+
+	for(i = 0; i < 13; i++)
+	{
+		// S4 += S0; S9 += S0; S18 += S0; S27 += S0;
+		col[(base +  4) % 36] ^= col[(base + 0) % 36];
+		col[(base +  9) % 36] ^= col[(base + 0) % 36];
+		col[(base + 18) % 36] ^= col[(base + 0) % 36];
+		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+
+		// ROR9
+		base = (base + 27) % 36;
+
+		// SMIX
+		LOADCOLUMN(r0, 36, 0);
+		SUBSTITUTE(r0, _t1, _t2, _t3, _t0);
+		SUPERMIX(_t2, _t3, _t0, _t1, r0);
+		STORECOLUMN(r0, 36);
+
+		// S4 += S0; S10 += S0; S18 += S0; S27 += S0;
+		col[(base +  4) % 36] ^= col[(base + 0) % 36];
+		col[(base + 10) % 36] ^= col[(base + 0) % 36];
+		col[(base + 18) % 36] ^= col[(base + 0) % 36];
+		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+
+		// ROR9
+		base = (base + 27) % 36;
+
+		// SMIX
+		LOADCOLUMN(r0, 36, 0);
+		SUBSTITUTE(r0, _t1, _t2, _t3, _t0);
+		SUPERMIX(_t2, _t3, _t0, _t1, r0);
+		STORECOLUMN(r0, 36);
+
+		// S4 += S0; S10 += S0; S19 += S0; S27 += S0;
+		col[(base +  4) % 36] ^= col[(base + 0) % 36];
+		col[(base + 10) % 36] ^= col[(base + 0) % 36];
+		col[(base + 19) % 36] ^= col[(base + 0) % 36];
+		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+
+		// ROR9
+		base = (base + 27) % 36;
+
+		// SMIX
+		LOADCOLUMN(r0, 36, 0);
+		SUBSTITUTE(r0, _t1, _t2, _t3, _t0);
+		SUPERMIX(_t2, _t3, _t0, _t1, r0);
+		STORECOLUMN(r0, 36);
+
+		// S4 += S0; S10 += S0; S19 += S0; S28 += S0;
+		col[(base +  4) % 36] ^= col[(base + 0) % 36];
+		col[(base + 10) % 36] ^= col[(base + 0) % 36];
+		col[(base + 19) % 36] ^= col[(base + 0) % 36];
+		col[(base + 28) % 36] ^= col[(base + 0) % 36];
+
+		// ROR8
+		base = (base + 28) % 36;
+
+		// SMIX
+		LOADCOLUMN(r0, 36, 0);
+		SUBSTITUTE(r0, _t1, _t2, _t3, _t0);
+		SUPERMIX(_t2, _t3, _t0, _t1, r0);
+		STORECOLUMN(r0, 36);
+	}
+
+	// S4 += S0; S9 += S0; S18 += S0; S27 += S0;
+	col[(base +  4) % 36] ^= col[(base + 0) % 36];
+	col[(base +  9) % 36] ^= col[(base + 0) % 36];
+	col[(base + 18) % 36] ^= col[(base + 0) % 36];
+	col[(base + 27) % 36] ^= col[(base + 0) % 36];
+
+	// Transform to the standard basis and store output; S1 || S2 || S3 || S4
+	LOADCOLUMN(r0, 36, 1);
+	_mm_store_si128((__m128i*)hashval, r0);
+
+	// Transform to the standard basis and store output; S9 || S10 || S11 || S12
+	LOADCOLUMN(r0, 36, 9);
+	_mm_store_si128((__m128i*)hashval + 1, r0);
+
+	// Transform to the standard basis and store output; S18 || S19 || S20 || S21
+	LOADCOLUMN(r0, 36, 18);
+	_mm_store_si128((__m128i*)hashval + 2, r0);
+
+	// Transform to the standard basis and store output; S27 || S28 || S29 || S30
+	LOADCOLUMN(r0, 36, 27);
+	_mm_store_si128((__m128i*)hashval + 3, r0);
+}
+
+HashReturn fugue512_Init(hashState_fugue *ctx, int nHashSize)
+{
+	int i;
+	ctx->processed_bits = 0;
+	ctx->uBufferBytes = 0;
+	ctx->base = 0;
+
+
+	ctx->uHashSize = 512;
+	ctx->uBlockLength = 4;
+
+	for(i = 0; i < 6; i++)
+		ctx->state[i] = m128_zero;
+
+	ctx->state[6]  = _mm_load_si128((__m128i*)_IV512 + 0);
+	ctx->state[7]  = _mm_load_si128((__m128i*)_IV512 + 1);
+	ctx->state[8]  = _mm_load_si128((__m128i*)_IV512 + 2);
+	ctx->state[9]  = _mm_load_si128((__m128i*)_IV512 + 3);
+	ctx->state[10] = _mm_load_si128((__m128i*)_IV512 + 4);
+	ctx->state[11] = _mm_load_si128((__m128i*)_IV512 + 5);
+
+	return SUCCESS;
+}
+
+
+HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength databitlen)
+{
+	unsigned int uByteLength, uBlockCount, uRemainingBytes;
+
+	uByteLength = (unsigned int)(databitlen / 8);
+
+	if(state->uBufferBytes + uByteLength >= state->uBlockLength)
+	{
+		if(state->uBufferBytes != 0)
+		{
+			// Fill the buffer
+			memcpy(state->buffer + state->uBufferBytes, (void*)data, state->uBlockLength - state->uBufferBytes);
+
+			// Process the buffer
+			Compress512(state, state->buffer, 1);
+
+			state->processed_bits += state->uBlockLength * 8;
+			data += state->uBlockLength - state->uBufferBytes;
+			uByteLength -= state->uBlockLength - state->uBufferBytes;
+		}
+
+		// buffer now does not contain any unprocessed bytes
+
+		uBlockCount = uByteLength / state->uBlockLength;
+		uRemainingBytes = uByteLength % state->uBlockLength;
+
+		if(uBlockCount > 0)
+		{
+			Compress512(state, data, uBlockCount);
+
+			state->processed_bits += uBlockCount * state->uBlockLength * 8;
+			data += uBlockCount * state->uBlockLength;
+		}
+
+		if(uRemainingBytes > 0)
+		{
+			memcpy(state->buffer, (void*)data, uRemainingBytes);
+		}
+
+		state->uBufferBytes = uRemainingBytes;
+	}
+	else
+	{
+		memcpy(state->buffer + state->uBufferBytes, (void*)data, uByteLength);
+		state->uBufferBytes += uByteLength;
+	}
+
+	return SUCCESS;
+}
+
+HashReturn fugue512_Final(hashState_fugue *state, void *hashval)
+{
+	unsigned int i;
+	BitSequence lengthbuf[8] __attribute__((aligned(64)));
+
+	// Update message bit count
+	state->processed_bits += state->uBufferBytes * 8;
+
+	// Pad the remaining buffer bytes with zero
+	if(state->uBufferBytes != 0)
+	{
+	   if ( state->uBufferBytes != state->uBlockLength)
+		memset(state->buffer + state->uBufferBytes, 0, state->uBlockLength - state->uBufferBytes);
+
+	   Compress512(state, state->buffer, 1);
+	}
+
+	// Last two blocks are message length in bits
+	for(i = 0; i < 8; i++)
+           lengthbuf[i] = ((state->processed_bits) >> (8 * (7 - i))) & 0xff;
+
+	// Process the last two blocks
+	Compress512(state, lengthbuf, 2);
+
+	// Finalization
+	Final512(state, hashval);
+
+	return SUCCESS;
+}
+
+
+HashReturn fugue512_full(hashState_fugue *hs, void *hashval, const void *data, DataLength databitlen)
+{
+	fugue512_Init(hs, 512);
+	fugue512_Update(hs, data, databitlen*8);
+	fugue512_Final(hs, hashval);
+	return SUCCESS;
+}
+
+#endif  // AES
--- a/algo/fugue/fugue-aesni.h
+++ b/algo/fugue/fugue-aesni.h
@@ -0,0 +1,50 @@
+/*
+ * file        : hash_api.h
+ * version     : 1.0.208
+ * date        : 14.12.2010
+ * 
+ * Fugue vperm implementation Hash API
+ *
+ * Cagdas Calik
+ * ccalik@metu.edu.tr
+ * Institute of Applied Mathematics, Middle East Technical University, Turkey.
+ *
+ */
+
+#ifndef FUGUE_HASH_API_H
+#define FUGUE_HASH_API_H
+
+#if defined(__AES__) 
+
+#if !defined(__SSE4_1__)
+#error "Unsupported configuration, AES needs SSE4.1. Compile without AES."
+#endif
+
+#include "algo/sha/sha3_common.h"
+#include "simd-utils.h"
+
+
+typedef struct
+{
+	__m128i			state[12];
+	unsigned int	base;
+
+	unsigned int	uHashSize;
+	unsigned int	uBlockLength;
+	unsigned int	uBufferBytes;
+	DataLength		processed_bits;
+	BitSequence		buffer[4];
+
+} hashState_fugue __attribute__ ((aligned (64)));
+
+HashReturn fugue512_Init(hashState_fugue *state, int hashbitlen);
+
+HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength databitlen);
+
+HashReturn fugue512_Final(hashState_fugue *state, void *hashval);
+
+HashReturn fugue512_full(hashState_fugue *hs, void *hashval, const void *data, DataLength databitlen);
+
+#endif // AES
+#endif // HASH_API_H
+
--- a/algo/fugue/sph_fugue.h
+++ b/algo/fugue/sph_fugue.h
@@ -74,6 +74,14 @@ void sph_fugue512_close(void *cc, void *dst);
 void sph_fugue512_addbits_and_close(
 	void *cc, unsigned ub, unsigned n, void *dst);

+#define sph_fugue512_full( cc, dst, data, len ) \
+do{ \
+   sph_fugue512_init( cc ); \
+   sph_fugue512( cc, data, len ); \
+   sph_fugue512_close( cc, dst ); \
+}while(0)
+
+
 #ifdef __cplusplus
 }
 #endif
--- a/algo/gost/sph_gost.c
+++ b/algo/gost/sph_gost.c
@@ -4,7 +4,7 @@
 #include <stdlib.h>
 #include <memory.h>
 #include <math.h>
-
+#include "simd-utils.h"
 #include "sph_gost.h"

 #ifdef __cplusplus
@@ -696,9 +696,26 @@ static void AddModulo512(const void *a,const void *b,void *c)

 static void AddXor512(const void *a,const void *b,void *c)
 {
-	const unsigned long long *A=a, *B=b;
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+   casti_m512i( c, 0 ) = _mm512_xor_si512( casti_m512i( a, 0 ),
+                                           casti_m512i( b, 0 ) );
+#elif defined(__AVX2__)
+   casti_m256i( c, 0 ) = _mm256_xor_si256( casti_m256i( a, 0 ),
+                                           casti_m256i( b, 0 ) );
+   casti_m256i( c, 1 ) = _mm256_xor_si256( casti_m256i( a, 1 ),
+                                           casti_m256i( b, 1 ) );
+#elif defined(__SSE2__)
+   casti_m128i( c, 0 ) = _mm_xor_si128( casti_m128i( a, 0 ),
+                                        casti_m128i( b, 0 ) );
+   casti_m128i( c, 1 ) = _mm_xor_si128( casti_m128i( a, 1 ),
+                                        casti_m128i( b, 1 ) );
+   casti_m128i( c, 2 ) = _mm_xor_si128( casti_m128i( a, 2 ),
+                                        casti_m128i( b, 2 ) );
+   casti_m128i( c, 3 ) = _mm_xor_si128( casti_m128i( a, 3 ),
+                                        casti_m128i( b, 3 ) );
+#else
+   const unsigned long long *A=a, *B=b;
 	unsigned long long *C=c;
-#ifdef FULL_UNROLL
 	C[0] = A[0] ^ B[0];
 	C[1] = A[1] ^ B[1];
 	C[2] = A[2] ^ B[2];
@@ -707,12 +724,6 @@ static void AddXor512(const void *a,const void *b,void *c)
 	C[5] = A[5] ^ B[5];
 	C[6] = A[6] ^ B[6];
 	C[7] = A[7] ^ B[7];
-#else
-	int i = 0;
-
-	for(i=0; i<8; i++) {
-		C[i] = A[i] ^ B[i];
-	}
 #endif
 }

@@ -893,31 +904,32 @@ static void g_N(const unsigned char *N,unsigned char *h,const unsigned char *m)

 static void hash_X(unsigned char *IV,const unsigned char *message,unsigned long long length,unsigned char *out)
 {
-	unsigned char v512[64] = {
+	unsigned char v512[64] __attribute__((aligned(64))) = {
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x02,0x00
-	};
-	unsigned char v0[64] = {
+   };
+	unsigned char v0[64]  __attribute__((aligned(64))) = {
+		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
+   };
+	unsigned char Sigma[64] __attribute__((aligned(64))) = {
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
 	};
-	unsigned char Sigma[64] = {
+	unsigned char N[64] __attribute__((aligned(64))) = {
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
 	};
-	unsigned char N[64] = {
-		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
-		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
-		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
-		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
-	};
-	unsigned char m[64], *hash = IV;
+	unsigned char m[64] __attribute__((aligned(64)));
+   unsigned char *hash = IV;
 	unsigned long long len = length;

 	// Stage 2
@@ -952,7 +964,7 @@ static void hash_X(unsigned char *IV,const unsigned char *message,unsigned long

 static void hash_512(const unsigned char *message, unsigned long long length, unsigned char *out)
 {
-	unsigned char IV[64] = {
+	unsigned char IV[64] __attribute__((aligned(64))) = {
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
 		0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
--- a/algo/gost/sph_gost.h
+++ b/algo/gost/sph_gost.h
@@ -81,9 +81,9 @@ typedef struct {
 */
 typedef struct {
 #ifndef DOXYGEN_IGNORE
-	unsigned char buf[64];    /* first field, for alignment */
+	unsigned char buf[64] __attribute__((aligned(64))); 
+   sph_u32 V[5][8] __attribute__((aligned(64)));
 	size_t ptr;
-	sph_u32 V[5][8];
 #endif
 } sph_gost512_context;

--- a/algo/groestl/aes_ni/groestl-asm-aes.h
+++ b/algo/groestl/aes_ni/groestl-asm-aes.h
--- a/algo/groestl/aes_ni/groestl-asm-avx.h
+++ b/algo/groestl/aes_ni/groestl-asm-avx.h
--- a/algo/groestl/aes_ni/groestl-asm-vperm.h
+++ b/algo/groestl/aes_ni/groestl-asm-vperm.h
--- a/algo/groestl/aes_ni/groestl-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl-intr-aes.h
@@ -1,3 +1,6 @@
+#if !defined GROESTL_INTR_AES_H__
+#define GROESTL_INTR_AES_H__
+
 /* groestl-intr-aes.h     Aug 2011
 *
 * Groestl implementation with intrinsics using ssse3, sse4.1, and aes
@@ -11,16 +14,51 @@
 #include <wmmintrin.h>
 #include "hash-groestl.h"

-/* global constants  */
-__m128i ROUND_CONST_Lx;
-//__m128i ROUND_CONST_L0[ROUNDS512];
-//__m128i ROUND_CONST_L7[ROUNDS512];
-__m128i ROUND_CONST_P[ROUNDS1024];
-__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_1B;
-__m128i ALL_FF;
+static const __m128i round_const_p[] __attribute__ ((aligned (64))) =
+{
+   { 0x7060504030201000, 0xf0e0d0c0b0a09080 },
+   { 0x7161514131211101, 0xf1e1d1c1b1a19181 },
+   { 0x7262524232221202, 0xf2e2d2c2b2a29282 },
+   { 0x7363534333231303, 0xf3e3d3c3b3a39383 },
+   { 0x7464544434241404, 0xf4e4d4c4b4a49484 },
+   { 0x7565554535251505, 0xf5e5d5c5b5a59585 },
+   { 0x7666564636261606, 0xf6e6d6c6b6a69686 },
+   { 0x7767574737271707, 0xf7e7d7c7b7a79787 },
+   { 0x7868584838281808, 0xf8e8d8c8b8a89888 },
+   { 0x7969594939291909, 0xf9e9d9c9b9a99989 },
+   { 0x7a6a5a4a3a2a1a0a, 0xfaeadacabaaa9a8a },
+   { 0x7b6b5b4b3b2b1b0b, 0xfbebdbcbbbab9b8b },
+   { 0x7c6c5c4c3c2c1c0c, 0xfcecdcccbcac9c8c },
+   { 0x7d6d5d4d3d2d1d0d, 0xfdedddcdbdad9d8d }
+};
+
+static const __m128i round_const_q[] __attribute__ ((aligned (64))) =
+{
+   { 0x8f9fafbfcfdfefff, 0x0f1f2f3f4f5f6f7f },
+   { 0x8e9eaebecedeeefe, 0x0e1e2e3e4e5e6e7e },
+   { 0x8d9dadbdcdddedfd, 0x0d1d2d3d4d5d6d7d },
+   { 0x8c9cacbcccdcecfc, 0x0c1c2c3c4c5c6c7c },
+   { 0x8b9babbbcbdbebfb, 0x0b1b2b3b4b5b6b7b },
+   { 0x8a9aaabacadaeafa, 0x0a1a2a3a4a5a6a7a },
+   { 0x8999a9b9c9d9e9f9, 0x0919293949596979 },
+   { 0x8898a8b8c8d8e8f8, 0x0818283848586878 },
+   { 0x8797a7b7c7d7e7f7, 0x0717273747576777 },
+   { 0x8696a6b6c6d6e6f6, 0x0616263646566676 },
+   { 0x8595a5b5c5d5e5f5, 0x0515253545556575 },
+   { 0x8494a4b4c4d4e4f4, 0x0414243444546474 },
+   { 0x8393a3b3c3d3e3f3, 0x0313233343536373 },
+   { 0x8292a2b2c2d2e2f2, 0x0212223242526272 }
+};
+
+static const __m128i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02 };
+static const __m128i SUBSH_MASK0 = { 0x0b0e0104070a0d00, 0x0306090c0f020508 };
+static const __m128i SUBSH_MASK1 = { 0x0c0f0205080b0e01, 0x04070a0d00030609 };
+static const __m128i SUBSH_MASK2 = { 0x0d000306090c0f02, 0x05080b0e0104070a };
+static const __m128i SUBSH_MASK3 = { 0x0e0104070a0d0003, 0x06090c0f0205080b };
+static const __m128i SUBSH_MASK4 = { 0x0f0205080b0e0104, 0x070a0d000306090c };
+static const __m128i SUBSH_MASK5 = { 0x000306090c0f0205, 0x080b0e0104070a0d };
+static const __m128i SUBSH_MASK6 = { 0x0104070a0d000306, 0x090c0f0205080b0e };
+static const __m128i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003 };

 #define tos(a)    #a
 #define tostr(a)  tos(a)
@@ -73,7 +111,7 @@ __m128i ALL_FF;
  b5 = a7;\
  a6 = _mm_xor_si128(a6, a7);\
  a7 = _mm_xor_si128(a7, b6);\
-  \
+   \
  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
  b0 = _mm_xor_si128(b0, a4);\
  b6 = _mm_xor_si128(b6, a4);\
@@ -111,7 +149,7 @@ __m128i ALL_FF;
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = ALL_1B;\
+  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -152,25 +190,6 @@ __m128i ALL_FF;
 }/*MixBytes*/


-#define SET_CONSTANTS(){\
-  ALL_FF = _mm_set_epi32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff);\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x0306090c, 0x0f020508, 0x0b0e0104, 0x070a0d00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070a0d, 0x00030609, 0x0c0f0205, 0x080b0e01);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05080b0e, 0x0104070a, 0x0d000306, 0x090c0f02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x06090c0f, 0x0205080b, 0x0e010407, 0x0a0d0003);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x070a0d00, 0x0306090c, 0x0f020508, 0x0b0e0104);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x080b0e01, 0x04070a0d, 0x00030609, 0x0c0f0205);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x090c0f02, 0x05080b0e, 0x0104070a, 0x0d000306);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x0e010407, 0x0a0d0003, 0x06090c0f, 0x0205080b);\
-  for(i = 0; i < ROUNDS1024; i++)\
-  {\
-    ROUND_CONST_P[i] = _mm_set_epi32(0xf0e0d0c0 ^ (i * 0x01010101), 0xb0a09080 ^ (i * 0x01010101), 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_Q[i] = _mm_set_epi32(0x0f1f2f3f ^ (i * 0x01010101), 0x4f5f6f7f ^ (i * 0x01010101), 0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101));\
-  }\
-}while(0);\
-
 /* one round
 * a0-a7 = input rows
 * b0-b7 = output rows
@@ -194,32 +213,34 @@ __m128i ALL_FF;
  u8 round_counter = 0;\
  for(round_counter = 0; round_counter < 14; round_counter+=2) {\
    /* AddRoundConstant P1024 */\
-    xmm8 = _mm_xor_si128(xmm8, (ROUND_CONST_P[round_counter]));\
-    /* ShiftBytes P1024 + pre-AESENCLAST */\
-    xmm8  = _mm_shuffle_epi8(xmm8,  (SUBSH_MASK[0]));\
-    xmm9  = _mm_shuffle_epi8(xmm9,  (SUBSH_MASK[1]));\
-    xmm10 = _mm_shuffle_epi8(xmm10, (SUBSH_MASK[2]));\
-    xmm11 = _mm_shuffle_epi8(xmm11, (SUBSH_MASK[3]));\
-    xmm12 = _mm_shuffle_epi8(xmm12, (SUBSH_MASK[4]));\
-    xmm13 = _mm_shuffle_epi8(xmm13, (SUBSH_MASK[5]));\
-    xmm14 = _mm_shuffle_epi8(xmm14, (SUBSH_MASK[6]));\
-    xmm15 = _mm_shuffle_epi8(xmm15, (SUBSH_MASK[7]));\
+    xmm8 = _mm_xor_si128( xmm8, \
+             casti_m128i( round_const_p, round_counter ) ); \
+     /* ShiftBytes P1024 + pre-AESENCLAST */\
+    xmm8  = _mm_shuffle_epi8( xmm8,  SUBSH_MASK0 ); \
+    xmm9  = _mm_shuffle_epi8( xmm9,  SUBSH_MASK1 ); \
+    xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK2 ); \
+    xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK3 ); \
+    xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK4 ); \
+    xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK5 ); \
+    xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK6 ); \
+    xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK7 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
+    SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
+            xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7 ); \
    \
    /* AddRoundConstant P1024 */\
-    xmm0 = _mm_xor_si128(xmm0, (ROUND_CONST_P[round_counter+1]));\
-    /* ShiftBytes P1024 + pre-AESENCLAST */\
-    xmm0 = _mm_shuffle_epi8(xmm0, (SUBSH_MASK[0]));\
-    xmm1 = _mm_shuffle_epi8(xmm1, (SUBSH_MASK[1]));\
-    xmm2 = _mm_shuffle_epi8(xmm2, (SUBSH_MASK[2]));\
-    xmm3 = _mm_shuffle_epi8(xmm3, (SUBSH_MASK[3]));\
-    xmm4 = _mm_shuffle_epi8(xmm4, (SUBSH_MASK[4]));\
-    xmm5 = _mm_shuffle_epi8(xmm5, (SUBSH_MASK[5]));\
-    xmm6 = _mm_shuffle_epi8(xmm6, (SUBSH_MASK[6]));\
-    xmm7 = _mm_shuffle_epi8(xmm7, (SUBSH_MASK[7]));\
-    /* SubBytes + MixBytes */\
-    SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
+    xmm0 = _mm_xor_si128( xmm0, \
+             casti_m128i( round_const_p, round_counter+1 ) ); \
+    xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK0 ); \
+    xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK1 ); \
+    xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK2 ); \
+    xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK3 ); \
+    xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK4 ); \
+    xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK5 ); \
+    xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK6 ); \
+    xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK7 ); \
+    SUBMIX( xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7, \
+            xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
  }\
 }

@@ -227,48 +248,52 @@ __m128i ALL_FF;
  u8 round_counter = 0;\
  for(round_counter = 0; round_counter < 14; round_counter+=2) {\
    /* AddRoundConstant Q1024 */\
-    xmm1 = ALL_FF;\
-    xmm8  = _mm_xor_si128(xmm8,  xmm1);\
-    xmm9  = _mm_xor_si128(xmm9,  xmm1);\
-    xmm10 = _mm_xor_si128(xmm10, xmm1);\
-    xmm11 = _mm_xor_si128(xmm11, xmm1);\
-    xmm12 = _mm_xor_si128(xmm12, xmm1);\
-    xmm13 = _mm_xor_si128(xmm13, xmm1);\
-    xmm14 = _mm_xor_si128(xmm14, xmm1);\
-    xmm15 = _mm_xor_si128(xmm15, (ROUND_CONST_Q[round_counter]));\
+    xmm1 = m128_neg1;\
+    xmm8  = _mm_xor_si128( xmm8,  xmm1 ); \
+    xmm9  = _mm_xor_si128( xmm9,  xmm1 ); \
+    xmm10 = _mm_xor_si128( xmm10, xmm1 ); \
+    xmm11 = _mm_xor_si128( xmm11, xmm1 ); \
+    xmm12 = _mm_xor_si128( xmm12, xmm1 ); \
+    xmm13 = _mm_xor_si128( xmm13, xmm1 ); \
+    xmm14 = _mm_xor_si128( xmm14, xmm1 ); \
+    xmm15 = _mm_xor_si128( xmm15, \
+              casti_m128i( round_const_q, round_counter ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm8  = _mm_shuffle_epi8(xmm8,  (SUBSH_MASK[1]));\
-    xmm9  = _mm_shuffle_epi8(xmm9,  (SUBSH_MASK[3]));\
-    xmm10 = _mm_shuffle_epi8(xmm10, (SUBSH_MASK[5]));\
-    xmm11 = _mm_shuffle_epi8(xmm11, (SUBSH_MASK[7]));\
-    xmm12 = _mm_shuffle_epi8(xmm12, (SUBSH_MASK[0]));\
-    xmm13 = _mm_shuffle_epi8(xmm13, (SUBSH_MASK[2]));\
-    xmm14 = _mm_shuffle_epi8(xmm14, (SUBSH_MASK[4]));\
-    xmm15 = _mm_shuffle_epi8(xmm15, (SUBSH_MASK[6]));\
+    xmm8  = _mm_shuffle_epi8( xmm8,  SUBSH_MASK1 ); \
+    xmm9  = _mm_shuffle_epi8( xmm9,  SUBSH_MASK3 ); \
+    xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK5 ); \
+    xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK7 ); \
+    xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK0 ); \
+    xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK2 ); \
+    xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK4 ); \
+    xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK6 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
+    SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
+            xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6 , xmm7 ); \
    \
    /* AddRoundConstant Q1024 */\
-    xmm9 = ALL_FF;\
-    xmm0 = _mm_xor_si128(xmm0,  xmm9);\
-    xmm1 = _mm_xor_si128(xmm1,  xmm9);\
-    xmm2 = _mm_xor_si128(xmm2,  xmm9);\
-    xmm3 = _mm_xor_si128(xmm3,  xmm9);\
-    xmm4 = _mm_xor_si128(xmm4,  xmm9);\
-    xmm5 = _mm_xor_si128(xmm5,  xmm9);\
-    xmm6 = _mm_xor_si128(xmm6,  xmm9);\
-    xmm7 = _mm_xor_si128(xmm7,  (ROUND_CONST_Q[round_counter+1]));\
+    xmm9 = m128_neg1;\
+    xmm0 = _mm_xor_si128( xmm0, xmm9 ); \
+    xmm1 = _mm_xor_si128( xmm1, xmm9 ); \
+    xmm2 = _mm_xor_si128( xmm2, xmm9 ); \
+    xmm3 = _mm_xor_si128( xmm3, xmm9 ); \
+    xmm4 = _mm_xor_si128( xmm4, xmm9 ); \
+    xmm5 = _mm_xor_si128( xmm5, xmm9 ); \
+    xmm6 = _mm_xor_si128( xmm6, xmm9 ); \
+    xmm7 = _mm_xor_si128( xmm7, \
+             casti_m128i( round_const_q, round_counter+1 ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm0 = _mm_shuffle_epi8(xmm0, (SUBSH_MASK[1]));\
-    xmm1 = _mm_shuffle_epi8(xmm1, (SUBSH_MASK[3]));\
-    xmm2 = _mm_shuffle_epi8(xmm2, (SUBSH_MASK[5]));\
-    xmm3 = _mm_shuffle_epi8(xmm3, (SUBSH_MASK[7]));\
-    xmm4 = _mm_shuffle_epi8(xmm4, (SUBSH_MASK[0]));\
-    xmm5 = _mm_shuffle_epi8(xmm5, (SUBSH_MASK[2]));\
-    xmm6 = _mm_shuffle_epi8(xmm6, (SUBSH_MASK[4]));\
-    xmm7 = _mm_shuffle_epi8(xmm7, (SUBSH_MASK[6]));\
+    xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK1 ); \
+    xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK3 ); \
+    xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK5 ); \
+    xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK7 ); \
+    xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK0 ); \
+    xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK2 ); \
+    xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK4 ); \
+    xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK6 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
+    SUBMIX( xmm0,  xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7, \
+            xmm8,  xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
  }\
 }

@@ -280,7 +305,7 @@ __m128i ALL_FF;
 * clobbers: t0-t7
 */
 #define Matrix_Transpose(i0, i1, i2, i3, i4, i5, i6, i7, t0, t1, t2, t3, t4, t5, t6, t7){\
-  t0 = TRANSP_MASK;\
+  t0 = TRANSP_MASK; \
 \
  i6 = _mm_shuffle_epi8(i6, t0);\
  i0 = _mm_shuffle_epi8(i0, t0);\
@@ -368,7 +393,7 @@ __m128i ALL_FF;
  i4 = _mm_unpacklo_epi64(i4, i5);\
  t1 = _mm_unpackhi_epi64(t1, i5);\
  t2 = i6;\
-  o0 = TRANSP_MASK;\
+  o0 = TRANSP_MASK; \
  i6 = _mm_unpacklo_epi64(i6, i7);\
  t2 = _mm_unpackhi_epi64(t2, i7);\
  /* load transpose mask into a register, because it will be used 8 times */\
@@ -609,3 +634,4 @@ void OF1024( __m128i* chaining )
  return;
 }

+#endif
--- a/algo/groestl/aes_ni/groestl-intr-avx.h
+++ b/algo/groestl/aes_ni/groestl-intr-avx.h
--- a/algo/groestl/aes_ni/groestl-intr-vperm.h
+++ b/algo/groestl/aes_ni/groestl-intr-vperm.h
--- a/algo/groestl/aes_ni/groestl-version.h
+++ b/algo/groestl/aes_ni/groestl-version.h
@@ -1,16 +0,0 @@
-// specify assembly or intrinsics implementation
-//#define TASM
-#define TINTR
-
-//#define AES_NI
-
-//#ifdef AES_NI
-// specify AES-NI, AVX (with AES-NI) or vector-permute implementation
-
-//#ifndef NO_AES_NI
-
-#define VAES
-// #define VAVX
-// #define VVPERM
-
-//#endif
--- a/algo/groestl/aes_ni/groestl256-asm-aes.h
+++ b/algo/groestl/aes_ni/groestl256-asm-aes.h
@@ -1,529 +0,0 @@
-/* groestl-asm-aes.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3, sse4.1, and aes
- * instructions.
- * Authors: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-/* global constants  */
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (16))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (16))) unsigned char ALL_1B[16];
-__attribute__ ((aligned (16))) unsigned char ALL_FF[16];
-
-/* temporary variables  */
-__attribute__ ((aligned (16))) unsigned char QTEMP[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP[3*16];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b */
-#define MUL2(i, j, k){\
-  asm("pxor xmm"tostr(j)", xmm"tostr(j)"");\
-  asm("pcmpgtb xmm"tostr(j)", xmm"tostr(i)"");\
-  asm("paddb xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("pand xmm"tostr(j)", xmm"tostr(k)"");\
-  asm("pxor xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* t_i = a_i + a_{i+1} */\
-  asm("movdqa xmm"tostr(b6)", xmm"tostr(a0)"");\
-  asm("movdqa xmm"tostr(b7)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(a0)", xmm"tostr(a1)"");\
-  asm("movdqa xmm"tostr(b0)", xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(a1)", xmm"tostr(a2)"");\
-  asm("movdqa xmm"tostr(b1)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(a3)"");\
-  asm("movdqa xmm"tostr(b2)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(a4)"");\
-  asm("movdqa xmm"tostr(b3)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(a5)"");\
-  asm("movdqa xmm"tostr(b4)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(a6)"");\
-  asm("movdqa xmm"tostr(b5)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a7)", xmm"tostr(b6)"");\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a6)"");\
-  /* spill values y_4, y_5 to memory */\
-  asm("movaps [TEMP+0*16], xmm"tostr(b0)"");\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a7)"");\
-  asm("movaps [TEMP+1*16], xmm"tostr(b1)"");\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a0)"");\
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  asm("movdqa xmm"tostr(b0)", xmm"tostr(a0)"");\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a1)"");\
-  asm("movdqa xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a2)"");\
-  asm("movaps [TEMP+2*16], xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  asm("pxor xmm"tostr(a0)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(a1)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm("pxor xmm"tostr(a7)", [TEMP+2*16]");\
-  \
-  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  /* compute w_i : add y_{i+4} */\
-  asm("movaps xmm"tostr(b1)", [ALL_1B]");\
-  MUL2(a0, b0, b1);\
-  asm("pxor xmm"tostr(a0)", [TEMP+0*16]");\
-  MUL2(a1, b0, b1);\
-  asm("pxor xmm"tostr(a1)", [TEMP+1*16]");\
-  MUL2(a2, b0, b1);\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(b2)"");\
-  MUL2(a3, b0, b1);\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(b3)"");\
-  MUL2(a4, b0, b1);\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(b4)"");\
-  MUL2(a5, b0, b1);\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(b5)"");\
-  MUL2(a6, b0, b1);\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(b6)"");\
-  MUL2(a7, b0, b1);\
-  asm("pxor xmm"tostr(a7)", xmm"tostr(b7)"");\
-  \
-  /* compute v_i : double w_i      */\
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  MUL2(a0, b0, b1);\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a0)"");\
-  MUL2(a1, b0, b1);\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a1)"");\
-  MUL2(a2, b0, b1);\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a2)"");\
-  MUL2(a5, b0, b1);\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a5)"");\
-  MUL2(a6, b0, b1);\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a6)"");\
-  MUL2(a7, b0, b1);\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a7)"");\
-  MUL2(a3, b0, b1);\
-  MUL2(a4, b0, b1);\
-  asm("movaps xmm"tostr(b0)", [TEMP+0*16]");\
-  asm("movaps xmm"tostr(b1)", [TEMP+1*16]");\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a4)"");\
-}/*MixBytes*/
-
-#define SET_CONSTANTS(){\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0c0f0104070b0e00ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x03060a0d08020509ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0e090205000d0801ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x04070c0f0a03060bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x080b0306010f0a02ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x05000e090c04070dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0a0d040702090c03ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0601080b0e05000fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0b0e0500030a0d04ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0702090c0f060108ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0d080601040c0f05ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x00030b0e0907020aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0f0a0702050e0906ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x01040d080b00030cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x090c000306080b07ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x02050f0a0d01040eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}while(0);
-
-#define Push_All_Regs() do{\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}while(0);
-
-#define Pop_All_Regs() do{\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}while(0);
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant */\
-  asm ("movaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("pxor   xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  /* ShiftBytes + SubBytes (interleaved) */\
-  asm ("pxor xmm"tostr(b0)",  xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("aesenclast xmm"tostr(a0)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("aesenclast xmm"tostr(a1)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("aesenclast xmm"tostr(a2)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("aesenclast xmm"tostr(a3)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("aesenclast xmm"tostr(a4)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("aesenclast xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("aesenclast xmm"tostr(a6)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  asm ("aesenclast xmm"tostr(a7)", xmm"tostr(b0)"");\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("movaps xmm"tostr(t0)", [TRANSP_MASK]");\
-  \
-  asm ("pshufb xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i3)", xmm"tostr(t0)"");\
-  \
-  asm ("movdqa xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(i2)"");\
-  \
-  asm ("punpcklwd xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhwd xmm"tostr(o1)", xmm"tostr(i1)"");\
-  asm ("punpcklwd xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhwd xmm"tostr(t0)", xmm"tostr(i3)"");\
-  \
-  asm ("pshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("pshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("pshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("pshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-  \
-  asm ("movdqa xmm"tostr(o2)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(o3)", xmm"tostr(o1)"");\
-  \
-  asm ("punpckldq xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("punpckldq xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("punpckhdq xmm"tostr(o2)", xmm"tostr(i2)"");\
-  asm ("punpckhdq xmm"tostr(o3)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i4)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o4)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o5)", xmm"tostr(i2)"");\
-  asm ("movdqa     xmm"tostr(o6)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o4)", xmm"tostr(i6)"");\
-  asm ("punpckhqdq xmm"tostr(o5)", xmm"tostr(i6)"");\
-  asm ("movdqa     xmm"tostr(o7)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o7)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("movdqa     xmm"tostr(o0)", xmm"tostr(i0)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhqdq xmm"tostr(o0)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i3)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i4)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("pxor xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(i1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(i3)", xmm"tostr(i2)"");\
-  asm ("movdqa xmm"tostr(i5)", xmm"tostr(i4)"");\
-  asm ("movdqa xmm"tostr(i7)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i3)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i5)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i7)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("movaps xmm12, [rdi+0*16]");
-  asm ("movaps xmm13, [rdi+1*16]");
-  asm ("movaps xmm14, [rdi+2*16]");
-  asm ("movaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("movaps [rdi+0*16], xmm12");
-  asm ("movaps [rdi+1*16], xmm2");
-  asm ("movaps [rdi+2*16], xmm6");
-  asm ("movaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("movaps xmm12, [rsi+0*16]");
-  asm ("movaps xmm13, [rsi+1*16]");
-  asm ("movaps xmm14, [rsi+2*16]");
-  asm ("movaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (2x64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  asm ("movaps xmm8, [rdi+0*16]");
-  asm ("movaps xmm0, [rdi+1*16]");
-  asm ("movaps xmm4, [rdi+2*16]");
-  asm ("movaps xmm5, [rdi+3*16]");
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("pxor xmm8, xmm12");
-  asm ("pxor xmm0, xmm2");
-  asm ("pxor xmm4, xmm6");
-  asm ("pxor xmm5, xmm7");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("pxor xmm0, xmm8");
-  asm ("pxor xmm1, xmm10");
-  asm ("pxor xmm2, xmm12");
-  asm ("pxor xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("pxor xmm0, [rdi+0*16]");
-  asm ("pxor xmm1, [rdi+1*16]");
-  asm ("pxor xmm2, [rdi+2*16]");
-  asm ("pxor xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("movaps [rdi+0*16], xmm0");
-  asm ("movaps [rdi+1*16], xmm1");
-  asm ("movaps [rdi+2*16], xmm2");
-  asm ("movaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("movaps xmm8,  [rdi+0*16]");
-  asm ("movaps xmm10, [rdi+1*16]");
-  asm ("movaps xmm12, [rdi+2*16]");
-  asm ("movaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("pxor xmm8,  [rdi+0*16]");
-  asm ("pxor xmm10, [rdi+1*16]");
-  asm ("pxor xmm12, [rdi+2*16]");
-  asm ("pxor xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-
-  /* we only need to return the truncated half of the state */
-  asm ("movaps [rdi+2*16], xmm9");
-  asm ("movaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
--- a/algo/groestl/aes_ni/groestl256-asm-avx.h
+++ b/algo/groestl/aes_ni/groestl256-asm-avx.h
@@ -1,519 +0,0 @@
-/* groestl-asm-avx.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3, sse4.1, aes and avx
- * instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-
-/* global variables  */
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (32))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (32))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (32))) unsigned char ALL_1B[32];
-__attribute__ ((aligned (32))) unsigned char ALL_FF[32];
-
-/* temporary variables  */
-__attribute__ ((aligned (32))) unsigned char TEMP[6*32];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_CONSTANTS(){\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0c0f0104070b0e00ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x03060a0d08020509ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0e090205000d0801ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x04070c0f0a03060bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x080b0306010f0a02ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x05000e090c04070dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0a0d040702090c03ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0601080b0e05000fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0b0e0500030a0d04ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0702090c0f060108ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0d080601040c0f05ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x00030b0e0907020aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0f0a0702050e0906ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x01040d080b00030cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x090c000306080b07ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x02050f0a0d01040eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}while(0);
-
-#define Push_All_Regs() do{\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}while(0);
-
-#define Pop_All_Regs() do{\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}while(0);
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2(i, j, k, z){\
-  asm("vpcmpgtb xmm"tostr(j)", xmm"tostr(z)", xmm"tostr(i)"");\
-  asm("vpaddb xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("vpand xmm"tostr(j)", xmm"tostr(j)", xmm"tostr(k)"");\
-  asm("vpxor xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2v2(i, j, k, z){\
-  asm("vpblendvb xmm"tostr(j)", xmm"tostr(z)", xmm"tostr(k)", xmm"tostr(i)"");\
-  asm("vpaddb xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("vpxor xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* xmm"tostr(8..xmm"tostr(15 = a2 a3... a0 a1 */\
-  asm("vmovdqa xmm"tostr(b0)", xmm"tostr(a2)"");\
-  asm("vmovdqa xmm"tostr(b1)", xmm"tostr(a3)"");\
-  asm("vmovdqa xmm"tostr(b2)", xmm"tostr(a4)"");\
-  asm("vmovdqa xmm"tostr(b3)", xmm"tostr(a5)"");\
-  asm("vmovdqa xmm"tostr(b4)", xmm"tostr(a6)"");\
-  asm("vmovdqa xmm"tostr(b5)", xmm"tostr(a7)"");\
-  asm("vmovdqa xmm"tostr(b6)", xmm"tostr(a0)"");\
-  asm("vmovdqa xmm"tostr(b7)", xmm"tostr(a1)"");\
-  \
-  /* t_i = a_i + a_{i+1} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b6)"");\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a3)"");\
-  \
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(b0)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(b1)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a5)"");\
-  \
-  /* spill values y_4, y_5 to memory */\
-  asm("vmovaps [TEMP+0*16], xmm"tostr(b0)"");\
-  asm("vmovaps [TEMP+1*16], xmm"tostr(b1)"");\
-  asm("vmovaps [TEMP+2*16], xmm"tostr(b2)"");\
-  \
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  asm("vmovdqa xmm"tostr(b0)", xmm"tostr(a0)"");\
-  asm("vmovdqa xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm("vmovaps [TEMP+3*16], xmm"tostr(a2)"");\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", [TEMP+3*16]");\
-  \
-  /*compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  asm("vmovaps xmm"tostr(b1)", [ALL_1B]");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(b2)"");\
-  VMUL2(a7, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a0, b0, b1, b2);\
-  \
-  /* compute w_i :  add y_{i+4} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", [TEMP+0*16]");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", [TEMP+1*16]");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", [TEMP+2*16]");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b3)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b4)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b5)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b6)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b7)"");\
-  \
-  /*compute v_i: double w_i */\
-  VMUL2(a0, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  \
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(a3)", [TEMP+0*16]");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(a4)", [TEMP+1*16]");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(a5)", [TEMP+2*16]");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a2)"");\
-}/*MixBytes*/
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant */\
-  asm ("vmovaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("vpxor   xmm"tostr(a0)", xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("vpxor   xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a7)", xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  /* ShiftBytes + SubBytes (interleaved) */\
-  asm ("vpxor xmm"tostr(b0)",  xmm"tostr(b0)",  xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a0)", xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("vaesenclast xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a1)", xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("vaesenclast xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a2)", xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("vaesenclast xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a3)", xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("vaesenclast xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a4)", xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("vaesenclast xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a5)", xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("vaesenclast xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a6)", xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("vaesenclast xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a7)", xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  asm ("vaesenclast xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b0)"");\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
-
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("vmovaps xmm"tostr(t0)", [TRANSP_MASK]");\
-\
-  asm ("vpshufb xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i1)", xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i3)", xmm"tostr(i3)", xmm"tostr(t0)"");\
-\
-  asm ("vpunpckhwd xmm"tostr(o1)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklwd xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpckhwd xmm"tostr(t0)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklwd xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-\
-  asm ("vpshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("vpshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("vpshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("vpshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-\
-  asm ("vpunpckhdq xmm"tostr(o2)", xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("vpunpckhdq xmm"tostr(o3)", xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("vpunpckldq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("vpunpckldq xmm"tostr(o1)", xmm"tostr(o1)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("vpunpckhqdq xmm"tostr(o1)", xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("vpunpcklqdq xmm"tostr(o2)", xmm"tostr(i1)", xmm"tostr(i5)"");\
-  asm ("vpunpckhqdq xmm"tostr(o3)", xmm"tostr(i1)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(o4)", xmm"tostr(i2)", xmm"tostr(i6)"");\
-  asm ("vpunpckhqdq xmm"tostr(o5)", xmm"tostr(i2)", xmm"tostr(i6)"");\
-  asm ("vpunpcklqdq xmm"tostr(o6)", xmm"tostr(i3)", xmm"tostr(i7)"");\
-  asm ("vpunpckhqdq xmm"tostr(o7)", xmm"tostr(i3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("vpunpckhqdq xmm"tostr(o0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpckhqdq xmm"tostr(o1)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpckhqdq xmm"tostr(o2)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpckhqdq xmm"tostr(o3)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("vpxor xmm"tostr(t0)", xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i1)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i3)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i5)", xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i7)", xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("vmovaps xmm12, [rdi+0*16]");
-  asm ("vmovaps xmm13, [rdi+1*16]");
-  asm ("vmovaps xmm14, [rdi+2*16]");
-  asm ("vmovaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("vmovaps [rdi+0*16], xmm12");
-  asm ("vmovaps [rdi+1*16], xmm2");
-  asm ("vmovaps [rdi+2*16], xmm6");
-  asm ("vmovaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("vmovaps xmm12, [rsi+0*16]");
-  asm ("vmovaps xmm13, [rsi+1*16]");
-  asm ("vmovaps xmm14, [rsi+2*16]");
-  asm ("vmovaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value and xor message to CV to get input of P */
-  /* we first put two rows (2x64 bit) of the CV into one 128-bit xmm register */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("vpxor xmm8, xmm12, [rdi+0*16]");
-  asm ("vpxor xmm0, xmm2,  [rdi+1*16]");
-  asm ("vpxor xmm4, xmm6,  [rdi+2*16]");
-  asm ("vpxor xmm5, xmm7,  [rdi+3*16]");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("vpxor xmm0, xmm0, xmm8");
-  asm ("vpxor xmm1, xmm1, xmm10");
-  asm ("vpxor xmm2, xmm2, xmm12");
-  asm ("vpxor xmm3, xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("vpxor xmm0, xmm0, [rdi+0*16]");
-  asm ("vpxor xmm1, xmm1, [rdi+1*16]");
-  asm ("vpxor xmm2, xmm2, [rdi+2*16]");
-  asm ("vpxor xmm3, xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("vmovaps [rdi+0*16], xmm0");
-  asm ("vmovaps [rdi+1*16], xmm1");
-  asm ("vmovaps [rdi+2*16], xmm2");
-  asm ("vmovaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("vmovaps xmm8,  [rdi+0*16]");
-  asm ("vmovaps xmm10, [rdi+1*16]");
-  asm ("vmovaps xmm12, [rdi+2*16]");
-  asm ("vmovaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("vpxor xmm8,  xmm8,  [rdi+0*16]");
-  asm ("vpxor xmm10, xmm10, [rdi+1*16]");
-  asm ("vpxor xmm12, xmm12, [rdi+2*16]");
-  asm ("vpxor xmm14, xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-
-  /* we only need to return the truncated half of the state */
-  asm ("vmovaps [rdi+2*16], xmm9");
-  asm ("vmovaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
--- a/algo/groestl/aes_ni/groestl256-asm-vperm.h
+++ b/algo/groestl/aes_ni/groestl256-asm-vperm.h
@@ -1,856 +0,0 @@
-/* groestl-asm-vperm.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3 instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * Based on the vperm and aes_ni implementations of the hash function Groestl
- * by Cagdas Calik <ccalik@metu.edu.tr> http://www.metu.edu.tr/~ccalik/
- * Institute of Applied Mathematics, Middle East Technical University, Turkey
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-
-/* global constants  */
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (16))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (16))) unsigned char ALL_0F[16];
-__attribute__ ((aligned (16))) unsigned char ALL_15[16];
-__attribute__ ((aligned (16))) unsigned char ALL_1B[16];
-__attribute__ ((aligned (16))) unsigned char ALL_63[16];
-__attribute__ ((aligned (16))) unsigned char ALL_FF[16];
-__attribute__ ((aligned (16))) unsigned char VPERM_IPT[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_OPT[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_INV[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB1[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB2[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB4[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SBO[2*16];
-
-/* temporary variables  */
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL1[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL2[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL4[1*16];
-__attribute__ ((aligned (16))) unsigned char QTEMP[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP[8*16];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_SHARED_CONSTANTS(){\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_63)[ 0] = 0x6363636363636363ULL;\
-  ((u64*)ALL_63)[ 1] = 0x6363636363636363ULL;\
-  ((u64*)ALL_0F)[ 0] = 0x0F0F0F0F0F0F0F0FULL;\
-  ((u64*)ALL_0F)[ 1] = 0x0F0F0F0F0F0F0F0FULL;\
-  ((u64*)VPERM_IPT)[ 0] = 0x4C01307D317C4D00ULL;\
-  ((u64*)VPERM_IPT)[ 1] = 0xCD80B1FCB0FDCC81ULL;\
-  ((u64*)VPERM_IPT)[ 2] = 0xC2B2E8985A2A7000ULL;\
-  ((u64*)VPERM_IPT)[ 3] = 0xCABAE09052227808ULL;\
-  ((u64*)VPERM_OPT)[ 0] = 0x01EDBD5150BCEC00ULL;\
-  ((u64*)VPERM_OPT)[ 1] = 0xE10D5DB1B05C0CE0ULL;\
-  ((u64*)VPERM_OPT)[ 2] = 0xFF9F4929D6B66000ULL;\
-  ((u64*)VPERM_OPT)[ 3] = 0xF7974121DEBE6808ULL;\
-  ((u64*)VPERM_INV)[ 0] = 0x01040A060F0B0780ULL;\
-  ((u64*)VPERM_INV)[ 1] = 0x030D0E0C02050809ULL;\
-  ((u64*)VPERM_INV)[ 2] = 0x0E05060F0D080180ULL;\
-  ((u64*)VPERM_INV)[ 3] = 0x040703090A0B0C02ULL;\
-  ((u64*)VPERM_SB1)[ 0] = 0x3618D415FAE22300ULL;\
-  ((u64*)VPERM_SB1)[ 1] = 0x3BF7CCC10D2ED9EFULL;\
-  ((u64*)VPERM_SB1)[ 2] = 0xB19BE18FCB503E00ULL;\
-  ((u64*)VPERM_SB1)[ 3] = 0xA5DF7A6E142AF544ULL;\
-  ((u64*)VPERM_SB2)[ 0] = 0x69EB88400AE12900ULL;\
-  ((u64*)VPERM_SB2)[ 1] = 0xC2A163C8AB82234AULL;\
-  ((u64*)VPERM_SB2)[ 2] = 0xE27A93C60B712400ULL;\
-  ((u64*)VPERM_SB2)[ 3] = 0x5EB7E955BC982FCDULL;\
-  ((u64*)VPERM_SB4)[ 0] = 0x3D50AED7C393EA00ULL;\
-  ((u64*)VPERM_SB4)[ 1] = 0xBA44FE79876D2914ULL;\
-  ((u64*)VPERM_SB4)[ 2] = 0xE1E937A03FD64100ULL;\
-  ((u64*)VPERM_SB4)[ 3] = 0xA876DE9749087E9FULL;\
-/*((u64*)VPERM_SBO)[ 0] = 0xCFE474A55FBB6A00ULL;\
-  ((u64*)VPERM_SBO)[ 1] = 0x8E1E90D1412B35FAULL;\
-  ((u64*)VPERM_SBO)[ 2] = 0xD0D26D176FBDC700ULL;\
-  ((u64*)VPERM_SBO)[ 3] = 0x15AABF7AC502A878ULL;*/\
-  ((u64*)ALL_15)[ 0] = 0x1515151515151515ULL;\
-  ((u64*)ALL_15)[ 1] = 0x1515151515151515ULL;\
-}/**/
-
-/* VPERM
- * Transform w/o settings c*
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2){\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(t1)", xmm"tostr(c0)"");\
-  asm ("pandn  xmm"tostr(t0)", xmm"tostr(a0)"");\
-  asm ("pandn  xmm"tostr(t1)", xmm"tostr(a1)"");\
-  asm ("psrld  xmm"tostr(t0)", 4");\
-  asm ("psrld  xmm"tostr(t1)", 4");\
-  asm ("pand   xmm"tostr(a0)", xmm"tostr(c0)"");\
-  asm ("pand   xmm"tostr(a1)", xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(t2)", xmm"tostr(c2)"");\
-  asm ("movdqa xmm"tostr(t3)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(t2)", xmm"tostr(a0)"");\
-  asm ("pshufb xmm"tostr(t3)", xmm"tostr(a1)"");\
-  asm ("movdqa xmm"tostr(a0)", xmm"tostr(c1)"");\
-  asm ("movdqa xmm"tostr(a1)", xmm"tostr(c1)"");\
-  asm ("pshufb xmm"tostr(a0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(a1)", xmm"tostr(t1)"");\
-  asm ("pxor   xmm"tostr(a0)", xmm"tostr(t2)"");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(t3)"");\
-}/**/
-
-#define VPERM_Transform_Set_Const(table, c0, c1, c2){\
-  asm ("movaps xmm"tostr(c0)", [ALL_0F]");\
-  asm ("movaps xmm"tostr(c1)", ["tostr(table)"+0*16]");\
-  asm ("movaps xmm"tostr(c2)", ["tostr(table)"+1*16]");\
-}/**/
-
-/* VPERM
- * Transform
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform(a0, a1, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Transform State
- * inputs:
- * a0-a3 = state
- * table = transformation table to use
- * t* = clobbers
- * outputs:
- * a0-a3 = transformed state
- * */
-#define VPERM_Transform_State(a0, a1, a2, a3, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-  VPERM_Transform_No_Const(a2, a3, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Add Constant to State
- * inputs:
- * a0-a7 = state
- * constant = constant to add
- * t0 = clobber
- * outputs:
- * a0-a7 = state + constant
- * */
-#define VPERM_Add_Constant(a0, a1, a2, a3, a4, a5, a6, a7, constant, t0){\
-  asm ("movaps xmm"tostr(t0)", ["tostr(constant)"]");\
-  asm ("pxor   xmm"tostr(a0)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a1)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a2)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a3)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a4)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a5)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a6)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a7)",  xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * Set Substitute Core Constants
- * */
-#define VPERM_Substitute_Core_Set_Const(c0, c1, c2){\
-  VPERM_Transform_Set_Const(VPERM_INV, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Substitute Core
- * first part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0 = 1 row
- * t*, c* = clobbers
- * outputs:
- * b0a, b0b = inputs for lookup step
- * */
-#define VPERM_Substitute_Core(a0, b0a, b0b, t0, t1, c0, c1, c2){\
-  asm ("movdqa xmm"tostr(t0)",  xmm"tostr(c0)"");\
-  asm ("pandn  xmm"tostr(t0)",  xmm"tostr(a0)"");\
-  asm ("psrld  xmm"tostr(t0)",  4");\
-  asm ("pand   xmm"tostr(a0)",  xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(b0a)", "tostr(c1)"");\
-  asm ("pshufb xmm"tostr(b0a)", xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(a0)",  xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(b0b)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0b)", xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(b0b)", xmm"tostr(b0a)"");\
-  asm ("movdqa xmm"tostr(t1)",  xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(t1)",  xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(t1)",  xmm"tostr(b0a)"");\
-  asm ("movdqa xmm"tostr(b0a)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0a)", xmm"tostr(b0b)"");\
-  asm ("pxor   xmm"tostr(b0a)", xmm"tostr(a0)"");\
-  asm ("movdqa xmm"tostr(b0b)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0b)", xmm"tostr(t1)"");\
-  asm ("pxor   xmm"tostr(b0b)", xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * Lookup
- * second part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0a, a0b = output of Substitution Core
- * table = lookup table to use (*1 / *2 / *4)
- * t0 = clobber
- * outputs:
- * b0 = output of sbox + multiplication
- * */
-#define VPERM_Lookup(a0a, a0b, table, b0, t0){\
-  asm ("movaps xmm"tostr(b0)", ["tostr(table)"+0*16]");\
-  asm ("movaps xmm"tostr(t0)", ["tostr(table)"+1*16]");\
-  asm ("pshufb xmm"tostr(b0)", xmm"tostr(a0b)"");\
-  asm ("pshufb xmm"tostr(t0)", xmm"tostr(a0a)"");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * SubBytes and *2 / *4
- * this function is derived from:
- *   Constant-time SSSE3 AES core implementation
- *   by Mike Hamburg
- * and
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0-a7 = state
- * t*, c* = clobbers
- * outputs:
- * a0-a7 = state * 4
- * c2 = row0 * 2 -> b0
- * c1 = row7 * 2 -> b3
- * c0 = row7 * 1 -> b4
- * t2 = row4 * 1 -> b7
- * TEMP_MUL1 = row(i) * 1
- * TEMP_MUL2 = row(i) * 2
- *
- * call:VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7) */
-#define VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, t0, t1, t3, t4, c2, c1, c0, t2){\
-  /* set Constants */\
-  VPERM_Substitute_Core_Set_Const(c0, c1, c2);\
-  /* row 1 */\
-  VPERM_Substitute_Core(a1, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+1*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+1*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a1, t4);\
-  /* --- */\
-  /* row 2 */\
-  VPERM_Substitute_Core(a2, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+2*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+2*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a2, t4);\
-  /* --- */\
-  /* row 3 */\
-  VPERM_Substitute_Core(a3, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+3*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+3*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a3, t4);\
-  /* --- */\
-  /* row 5 */\
-  VPERM_Substitute_Core(a5, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+5*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+5*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a5, t4);\
-  /* --- */\
-  /* row 6 */\
-  VPERM_Substitute_Core(a6, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+6*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+6*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a6, t4);\
-  /* --- */\
-  /* row 7 */\
-  VPERM_Substitute_Core(a7, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+7*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c1, t4); /*c1 -> b3*/\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a7, t4);\
-  /* --- */\
-  /* row 4 */\
-  VPERM_Substitute_Core(a4, t0, t1, t3, t4, c0, [VPERM_INV+0*16], c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4); /*t2 -> b7*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+4*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a4, t4);\
-  /* --- */\
-  /* row 0 */\
-  VPERM_Substitute_Core(a0, t0, t1, t3, t4, c0, [VPERM_INV+0*16], c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, c0, t4); /*c0 -> b4*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c2, t4); /*c2 -> b0*/\
-  asm ("movaps [TEMP_MUL2+0*16], xmm"tostr(c2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a0, t4);\
-  /* --- */\
-}/**/
-
-
-/* Optimized MixBytes
- * inputs:
- * a0-a7 = (row0-row7) * 4
- * b0 = row0 * 2
- * b3 = row7 * 2
- * b4 = row7 * 1
- * b7 = row4 * 1
- * all *1 and *2 values must also be in TEMP_MUL1, TEMP_MUL2
- * output: b0-b7
- * */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* save one value */\
-  asm ("movaps [TEMP_MUL4], xmm"tostr(a3)"");\
-  /* 1 */\
-  asm ("movdqa xmm"tostr(b1)", xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(b4)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b1)", [TEMP_MUL2+3*16]");\
-  asm ("movdqa xmm"tostr(b2)", xmm"tostr(b1)"");\
-  \
-  /* 2 */\
-  asm ("movdqa xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a4)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(b7)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(b3)""); /* -> helper! */\
-  asm ("movdqa xmm"tostr(b6)", xmm"tostr(b5)"");\
-  \
-  /* 4 */\
-  asm ("pxor   xmm"tostr(b7)", xmm"tostr(a6)"");\
-  /*asm ("pxor   xmm"tostr(b7)", [TEMP_MUL1+4*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b7)", [TEMP_MUL1+6*16]");\
-  asm ("pxor   xmm"tostr(b7)", [TEMP_MUL2+1*16]");\
-  asm ("pxor   xmm"tostr(b7)", xmm"tostr(b3)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b2)", xmm"tostr(b7)"");\
-  \
-  /* 3 */\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(a7)"");\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL1+5*16]");\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL1+7*16]");\
-  /*asm ("pxor   xmm"tostr(b0)", [TEMP_MUL2+0*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL2+2*16]");\
-  asm ("movdqa xmm"tostr(b3)", xmm"tostr(b0)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(b0)"");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(b7)""); /* moved from 4 */\
-  \
-  /* 5 */\
-  asm ("pxor   xmm"tostr(b4)", xmm"tostr(a2)"");\
-  /*asm ("pxor   xmm"tostr(b4)", [TEMP_MUL1+0*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL1+2*16]");\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL2+3*16]");\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL2+5*16]");\
-  asm ("pxor   xmm"tostr(b3)", xmm"tostr(b4)"");\
-  asm ("pxor   xmm"tostr(b6)", xmm"tostr(b4)"");\
-  \
-  /* 6 */\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL1+1*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL1+3*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL2+4*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL2+6*16]");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a3)"");\
-  asm ("pxor xmm"tostr(b5)", xmm"tostr(a3)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a3)"");\
-  \
-  /* 7 */\
-  asm ("pxor xmm"tostr(a1)", [TEMP_MUL1+1*16]");\
-  asm ("pxor xmm"tostr(a1)", [TEMP_MUL2+4*16]");\
-  asm ("pxor xmm"tostr(b2)", xmm"tostr(a1)"");\
-  asm ("pxor xmm"tostr(b3)", xmm"tostr(a1)"");\
-  \
-  /* 8 */\
-  asm ("pxor xmm"tostr(a5)", [TEMP_MUL1+5*16]");\
-  asm ("pxor xmm"tostr(a5)", [TEMP_MUL2+0*16]");\
-  asm ("pxor xmm"tostr(b6)", xmm"tostr(a5)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a5)"");\
-  \
-  /* 9 */\
-  asm ("movaps xmm"tostr(a3)", [TEMP_MUL1+2*16]");\
-  asm ("pxor   xmm"tostr(a3)", [TEMP_MUL2+5*16]");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(a3)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* 10 */\
-  asm ("movaps xmm"tostr(a1)", [TEMP_MUL1+6*16]");\
-  asm ("pxor   xmm"tostr(a1)", [TEMP_MUL2+1*16]");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm ("pxor   xmm"tostr(b4)", xmm"tostr(a1)"");\
-  \
-  /* 11 */\
-  asm ("movaps xmm"tostr(a5)", [TEMP_MUL1+3*16]");\
-  asm ("pxor   xmm"tostr(a5)", [TEMP_MUL2+6*16]");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm ("pxor   xmm"tostr(b6)", xmm"tostr(a5)"");\
-  \
-  /* 12 */\
-  asm ("movaps xmm"tostr(a3)", [TEMP_MUL1+7*16]");\
-  asm ("pxor   xmm"tostr(a3)", [TEMP_MUL2+2*16]");\
-  asm ("pxor   xmm"tostr(b2)", xmm"tostr(a3)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* 13 */\
-  asm ("pxor xmm"tostr(b0)", [TEMP_MUL4]");\
-  asm ("pxor xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm ("pxor xmm"tostr(b1)", xmm"tostr(a4)"");\
-  asm ("pxor xmm"tostr(b3)", xmm"tostr(a6)"");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a7)"");\
-  asm ("pxor xmm"tostr(b5)", xmm"tostr(a0)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a2)"");\
-}/**/
-
-//#if (LENGTH <= 256)
-
-#define SET_CONSTANTS(){\
-  SET_SHARED_CONSTANTS();\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0706050403020100ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x080f0e0d0c0b0a09ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0007060504030201ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x0a09080f0e0d0c0bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x0100070605040302ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x0c0b0a09080f0e0dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0201000706050403ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0e0d0c0b0a09080fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0302010007060504ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0f0e0d0c0b0a0908ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0403020100070605ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x09080f0e0d0c0b0aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0504030201000706ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x0b0a09080f0e0d0cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x0605040302010007ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x0d0c0b0a09080f0eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}/**/
-
-#define Push_All_Regs(){\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}/**/
-
-#define Pop_All_Regs(){\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}/**/
-
-
-/* vperm:
- * transformation before rounds with ipt
- * first round add transformed constant
- * middle rounds: add constant XOR 0x15...15
- * last round: additionally add 0x15...15 after MB
- * transformation after rounds with opt
- */
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant + ShiftBytes (interleaved) */\
-  asm ("movaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("pxor   xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("pshufb xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("pxor   xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("pshufb xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("pxor   xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("pshufb xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("pxor   xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  asm ("pshufb xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("pshufb xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  /* SubBytes + Multiplication by 2 and 4 */\
-  VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7);\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}/**/
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  VPERM_Add_Constant(8, 9, 10, 11, 12, 13, 14, 15, ALL_15, 0);\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  VPERM_Add_Constant(8, 9, 10, 11, 12, 13, 14, 15, ALL_15, 0);\
-}
-
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("movaps xmm"tostr(t0)", [TRANSP_MASK]");\
-\
-  asm ("pshufb xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i3)", xmm"tostr(t0)"");\
-\
-  asm ("movdqa xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(i2)"");\
-\
-  asm ("punpcklwd xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhwd xmm"tostr(o1)", xmm"tostr(i1)"");\
-  asm ("punpcklwd xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhwd xmm"tostr(t0)", xmm"tostr(i3)"");\
-\
-  asm ("pshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("pshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("pshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("pshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-\
-  asm ("movdqa xmm"tostr(o2)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(o3)", xmm"tostr(o1)"");\
-\
-  asm ("punpckldq xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("punpckldq xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("punpckhdq xmm"tostr(o2)", xmm"tostr(i2)"");\
-  asm ("punpckhdq xmm"tostr(o3)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i4)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o4)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o5)", xmm"tostr(i2)"");\
-  asm ("movdqa     xmm"tostr(o6)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o4)", xmm"tostr(i6)"");\
-  asm ("punpckhqdq xmm"tostr(o5)", xmm"tostr(i6)"");\
-  asm ("movdqa     xmm"tostr(o7)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o7)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("movdqa     xmm"tostr(o0)", xmm"tostr(i0)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhqdq xmm"tostr(o0)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i3)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i4)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("pxor xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(i1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(i3)", xmm"tostr(i2)"");\
-  asm ("movdqa xmm"tostr(i5)", xmm"tostr(i4)"");\
-  asm ("movdqa xmm"tostr(i7)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i3)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i5)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i7)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst_CNT2(i, j){\
-  asm ("movaps xmm0, [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("movaps xmm1, [ROUND_CONST_L7+"tostr(i)"*16]");\
-  asm ("movaps xmm2, [ROUND_CONST_L0+"tostr(j)"*16]");\
-  asm ("movaps xmm3, [ROUND_CONST_L7+"tostr(j)"*16]");\
-  VPERM_Transform_State(0, 1, 2, 3, VPERM_IPT, 4, 5, 6, 7, 8, 9, 10);\
-  asm ("pxor xmm0, [ALL_15]");\
-  asm ("pxor xmm1, [ALL_15]");\
-  asm ("pxor xmm2, [ALL_15]");\
-  asm ("pxor xmm3, [ALL_15]");\
-  asm ("movaps [ROUND_CONST_L0+"tostr(i)"*16], xmm0");\
-  asm ("movaps [ROUND_CONST_L7+"tostr(i)"*16], xmm1");\
-  asm ("movaps [ROUND_CONST_L0+"tostr(j)"*16], xmm2");\
-  asm ("movaps [ROUND_CONST_L7+"tostr(j)"*16], xmm3");\
-}/**/
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst(){\
-  asm ("movaps xmm0, [ROUND_CONST_Lx]");\
-  VPERM_Transform(0, 1, VPERM_IPT, 4, 5, 6, 7, 8, 9, 10);\
-  asm ("pxor xmm0, [ALL_15]");\
-  asm ("movaps [ROUND_CONST_Lx], xmm0");\
-  VPERM_Transform_RoundConst_CNT2(0, 1);\
-  VPERM_Transform_RoundConst_CNT2(2, 3);\
-  VPERM_Transform_RoundConst_CNT2(4, 5);\
-  VPERM_Transform_RoundConst_CNT2(6, 7);\
-  VPERM_Transform_RoundConst_CNT2(8, 9);\
-}/**/
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* transform round constants into VPERM mode */
-  VPERM_Transform_RoundConst();
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("movaps xmm12, [rdi+0*16]");
-  asm ("movaps xmm13, [rdi+1*16]");
-  asm ("movaps xmm14, [rdi+2*16]");
-  asm ("movaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  VPERM_Transform_State(12, 13, 14, 15, VPERM_IPT, 1, 2, 3, 4, 5, 6, 7);
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("movaps [rdi+0*16], xmm12");
-  asm ("movaps [rdi+1*16], xmm2");
-  asm ("movaps [rdi+2*16], xmm6");
-  asm ("movaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("movaps xmm12, [rsi+0*16]");
-  asm ("movaps xmm13, [rsi+1*16]");
-  asm ("movaps xmm14, [rsi+2*16]");
-  asm ("movaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  VPERM_Transform_State(12, 13, 14, 15, VPERM_IPT, 1, 2, 3, 4, 5, 6, 7);
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  asm ("movaps xmm8, [rdi+0*16]");
-  asm ("movaps xmm0, [rdi+1*16]");
-  asm ("movaps xmm4, [rdi+2*16]");
-  asm ("movaps xmm5, [rdi+3*16]");
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("pxor xmm8, xmm12");
-  asm ("pxor xmm0, xmm2");
-  asm ("pxor xmm4, xmm6");
-  asm ("pxor xmm5, xmm7");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("pxor xmm0, xmm8");
-  asm ("pxor xmm1, xmm10");
-  asm ("pxor xmm2, xmm12");
-  asm ("pxor xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("pxor xmm0, [rdi+0*16]");
-  asm ("pxor xmm1, [rdi+1*16]");
-  asm ("pxor xmm2, [rdi+2*16]");
-  asm ("pxor xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("movaps [rdi+0*16], xmm0");
-  asm ("movaps [rdi+1*16], xmm1");
-  asm ("movaps [rdi+2*16], xmm2");
-  asm ("movaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("movaps xmm8,  [rdi+0*16]");
-  asm ("movaps xmm10, [rdi+1*16]");
-  asm ("movaps xmm12, [rdi+2*16]");
-  asm ("movaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("pxor xmm8,  [rdi+0*16]");
-  asm ("pxor xmm10, [rdi+1*16]");
-  asm ("pxor xmm12, [rdi+2*16]");
-  asm ("pxor xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-  VPERM_Transform(9, 11, VPERM_OPT, 0, 1, 2, 3, 5, 6, 7);
-
-  /* we only need to return the truncated half of the state */
-  asm ("movaps [rdi+2*16], xmm9");
-  asm ("movaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
-
--- a/algo/groestl/aes_ni/groestl256-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl256-intr-aes.h
@@ -11,17 +11,44 @@
 #include <wmmintrin.h>
 #include "hash-groestl256.h"

-/* global constants  */
-__m128i ROUND_CONST_Lx;
-__m128i ROUND_CONST_L0[ROUNDS512];
-__m128i ROUND_CONST_L7[ROUNDS512];
-//__m128i ROUND_CONST_P[ROUNDS1024];
-//__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_1B;
-__m128i ALL_FF;
+static const __m128i round_const_l0[] __attribute__ ((aligned (64))) =
+{
+   { 0x7060504030201000, 0xffffffffffffffff },
+   { 0x7161514131211101, 0xffffffffffffffff },
+   { 0x7262524232221202, 0xffffffffffffffff },
+   { 0x7363534333231303, 0xffffffffffffffff },
+   { 0x7464544434241404, 0xffffffffffffffff },
+   { 0x7565554535251505, 0xffffffffffffffff },
+   { 0x7666564636261606, 0xffffffffffffffff },
+   { 0x7767574737271707, 0xffffffffffffffff },
+   { 0x7868584838281808, 0xffffffffffffffff },
+   { 0x7969594939291909, 0xffffffffffffffff }
+};

+static const __m128i round_const_l7[] __attribute__ ((aligned (64))) =
+{
+   { 0x0000000000000000, 0x8f9fafbfcfdfefff },
+   { 0x0000000000000000, 0x8e9eaebecedeeefe },
+   { 0x0000000000000000, 0x8d9dadbdcdddedfd },
+   { 0x0000000000000000, 0x8c9cacbcccdcecfc },
+   { 0x0000000000000000, 0x8b9babbbcbdbebfb },
+   { 0x0000000000000000, 0x8a9aaabacadaeafa },
+   { 0x0000000000000000, 0x8999a9b9c9d9e9f9 },
+   { 0x0000000000000000, 0x8898a8b8c8d8e8f8 },
+   { 0x0000000000000000, 0x8797a7b7c7d7e7f7 },
+   { 0x0000000000000000, 0x8696a6b6c6d6e6f6 }
+};
+
+static const __m128i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02 };
+
+static const __m128i SUBSH_MASK0 = { 0x0c0f0104070b0e00, 0x03060a0d08020509 };
+static const __m128i SUBSH_MASK1 = { 0x0e090205000d0801, 0x04070c0f0a03060b };
+static const __m128i SUBSH_MASK2 = { 0x080b0306010f0a02, 0x05000e090c04070d };
+static const __m128i SUBSH_MASK3 = { 0x0a0d040702090c03, 0x0601080b0e05000f };
+static const __m128i SUBSH_MASK4 = { 0x0b0e0500030a0d04, 0x0702090c0f060108 };
+static const __m128i SUBSH_MASK5 = { 0x0d080601040c0f05, 0x00030b0e0907020a };
+static const __m128i SUBSH_MASK6 = { 0x0f0a0702050e0906, 0x01040d080b00030c };
+static const __m128i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };

 #define tos(a)    #a
 #define tostr(a)  tos(a)
@@ -38,8 +65,6 @@ __m128i ALL_FF;
  i = _mm_xor_si128(i, j);\
 } 

- /**/
-
 /* Yet another implementation of MixBytes.
   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
   Input: a0, ..., a7
@@ -113,7 +138,7 @@ __m128i ALL_FF;
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = ALL_1B;\
+  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -153,25 +178,6 @@ __m128i ALL_FF;
  b1 = _mm_xor_si128(b1, a4);\
 }/*MixBytes*/

-#define SET_CONSTANTS(){\
-   ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}while(0); \
-
 /* one round
 * i = round number
 * a0-a7 = input rows
@@ -179,34 +185,34 @@ __m128i ALL_FF;
 */
 #define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
  /* AddRoundConstant */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a4 = _mm_xor_si128(a4, b1);\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
+  b1 = m128_const_64( 0xffffffffffffffff, 0 ); \
+  a0 = _mm_xor_si128( a0, casti_m128i( round_const_l0, i ) ); \
+  a1 = _mm_xor_si128( a1, b1 ); \
+  a2 = _mm_xor_si128( a2, b1 ); \
+  a3 = _mm_xor_si128( a3, b1 ); \
+  a4 = _mm_xor_si128( a4, b1 ); \
+  a5 = _mm_xor_si128( a5, b1 ); \
+  a6 = _mm_xor_si128( a6, b1 ); \
+  a7 = _mm_xor_si128( a7, casti_m128i( round_const_l7, i ) ); \
  \
  /* ShiftBytes + SubBytes (interleaved) */\
  b0 = _mm_xor_si128(b0,  b0);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a0 = _mm_aesenclast_si128(a0, b0);\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a1 = _mm_aesenclast_si128(a1, b0);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a2 = _mm_aesenclast_si128(a2, b0);\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a3 = _mm_aesenclast_si128(a3, b0);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a4 = _mm_aesenclast_si128(a4, b0);\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a5 = _mm_aesenclast_si128(a5, b0);\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a6 = _mm_aesenclast_si128(a6, b0);\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  a7 = _mm_aesenclast_si128(a7, b0);\
+  a0 = _mm_shuffle_epi8( a0, SUBSH_MASK0 ); \
+  a0 = _mm_aesenclast_si128( a0, b0 );\
+  a1 = _mm_shuffle_epi8( a1, SUBSH_MASK1 ); \
+  a1 = _mm_aesenclast_si128( a1, b0 );\
+  a2 = _mm_shuffle_epi8( a2, SUBSH_MASK2 ); \
+  a2 = _mm_aesenclast_si128( a2, b0 );\
+  a3 = _mm_shuffle_epi8( a3, SUBSH_MASK3 ); \
+  a3 = _mm_aesenclast_si128( a3, b0 );\
+  a4 = _mm_shuffle_epi8( a4, SUBSH_MASK4 ); \
+  a4 = _mm_aesenclast_si128( a4, b0 );\
+  a5 = _mm_shuffle_epi8( a5, SUBSH_MASK5 ); \
+  a5 = _mm_aesenclast_si128( a5, b0 );\
+  a6 = _mm_shuffle_epi8( a6, SUBSH_MASK6 ); \
+  a6 = _mm_aesenclast_si128( a6, b0 );\
+  a7 = _mm_shuffle_epi8( a7, SUBSH_MASK7 ); \
+  a7 = _mm_aesenclast_si128( a7, b0 );\
  \
  /* MixBytes */\
  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
@@ -234,8 +240,9 @@ __m128i ALL_FF;
 * outputs: i0, o1-o3
 * clobbers: t0
 */
+
 #define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
+  t0 = TRANSP_MASK; \
  \
  i0 = _mm_shuffle_epi8(i0, t0);\
  i1 = _mm_shuffle_epi8(i1, t0);\
--- a/algo/groestl/aes_ni/groestl256-intr-avx.h
+++ b/algo/groestl/aes_ni/groestl256-intr-avx.h
@@ -1,482 +0,0 @@
-/* groestl-intr-avx.h     Aug 2011
- *
- * Groestl implementation with intrinsics using ssse3, sse4.1, aes and avx
- * instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include <smmintrin.h>
-#include <wmmintrin.h>
-#include <immintrin.h>
-#include "hash-groestl256.h"
-
-/* global constants  */
-__m128i ROUND_CONST_Lx;
-__m128i ROUND_CONST_L0[ROUNDS512];
-__m128i ROUND_CONST_L7[ROUNDS512];
-__m128i ROUND_CONST_P[ROUNDS1024];
-__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_FF;
-//#if LENGTH <= 256
-__m128i ALL_1B;
-//#else
-//__m256d ALL_1B;
-//#endif
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define insert_m128i_in_m256d(ymm, xmm, pos) (_mm256_castsi256_pd(_mm256_insertf128_si256(_mm256_castpd_si256(ymm), xmm, pos)))
-#define extract_m128i_from_m256d(ymm, pos) (_mm256_extractf128_si256(_mm256_castpd_si256(ymm), pos))
-
-#define SET_CONSTANTS(){\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  ALL_FF = _mm_set_epi32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}while(0);
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2(i, j, k, z){\
-  j = _mm_cmpgt_epi8(z, i);\
-  i = _mm_add_epi8(i, i);\
-  j = _mm_and_si128(j, k);\
-  i = _mm_xor_si128(i, j);\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* xmm"tostr(8..xmm"tostr(15 = a2 a3... a0 a1 */\
-  b0 = a2;\
-  b1 = a3;\
-  b2 = a4;\
-  b3 = a5;\
-  b4 = a6;\
-  b5 = a7;\
-  b6 = a0;\
-  b7 = a1;\
-  \
-  /* t_i = a_i + a_{i+1} */\
-  a0 = _mm_xor_si128(a0, a1);\
-  a1 = _mm_xor_si128(a1, a2);\
-  a2 = _mm_xor_si128(a2, a3);\
-  a3 = _mm_xor_si128(a3, a4);\
-  a4 = _mm_xor_si128(a4, a5);\
-  a5 = _mm_xor_si128(a5, a6);\
-  a6 = _mm_xor_si128(a6, a7);\
-  a7 = _mm_xor_si128(a7, b6);\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  b0 = _mm_xor_si128(b0, a4);\
-  b1 = _mm_xor_si128(b1, a5);\
-  b2 = _mm_xor_si128(b2, a6);\
-  b3 = _mm_xor_si128(b3, a7);\
-  b4 = _mm_xor_si128(b4, a0);\
-  b5 = _mm_xor_si128(b5, a1);\
-  b6 = _mm_xor_si128(b6, a2);\
-  b7 = _mm_xor_si128(b7, a3);\
-  \
-  b0 = _mm_xor_si128(b0, a6);\
-  b1 = _mm_xor_si128(b1, a7);\
-  b2 = _mm_xor_si128(b2, a0);\
-  b3 = _mm_xor_si128(b3, a1);\
-  b4 = _mm_xor_si128(b4, a2);\
-  b5 = _mm_xor_si128(b5, a3);\
-  b6 = _mm_xor_si128(b6, a4);\
-  b7 = _mm_xor_si128(b7, a5);\
-  \
-  /* spill values y_4, y_5 to memory */\
-  TEMP0 = b0;\
-  TEMP1 = b1;\
-  TEMP2 = b2;\
-  \
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  b0 = a0;\
-  b1 = a1;\
-  TEMP3 = a2;\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  a0 = _mm_xor_si128(a0, a3);\
-  a1 = _mm_xor_si128(a1, a4);\
-  a2 = _mm_xor_si128(a2, a5);\
-  a3 = _mm_xor_si128(a3, a6);\
-  a4 = _mm_xor_si128(a4, a7);\
-  a5 = _mm_xor_si128(a5, b0);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, TEMP3);\
-  \
-  /*compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  b1 = ALL_1B;\
-  b2 = _mm_xor_si128(b2, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a0, b0, b1, b2);\
-  \
-  /* compute w_i :  add y_{i+4} */\
-  a0 = _mm_xor_si128(a0, TEMP0);\
-  a1 = _mm_xor_si128(a1, TEMP1);\
-  a2 = _mm_xor_si128(a2, TEMP2);\
-  a3 = _mm_xor_si128(a3, b3);\
-  a4 = _mm_xor_si128(a4, b4);\
-  a5 = _mm_xor_si128(a5, b5);\
-  a6 = _mm_xor_si128(a6, b6);\
-  a7 = _mm_xor_si128(a7, b7);\
-  \
-  /*compute v_i: double w_i */\
-  VMUL2(a0, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  \
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  b0 = _mm_xor_si128(a3, TEMP0);\
-  b1 = _mm_xor_si128(a4, TEMP1);\
-  b2 = _mm_xor_si128(a5, TEMP2);\
-  b3 = _mm_xor_si128(b3, a6);\
-  b4 = _mm_xor_si128(b4, a7);\
-  b5 = _mm_xor_si128(b5, a0);\
-  b6 = _mm_xor_si128(b6, a1);\
-  b7 = _mm_xor_si128(b7, a2);\
-}/*MixBytes*/
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* Add Round Constant */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a4 = _mm_xor_si128(a4, b1);\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
-  \
-  /* ShiftBytes + SubBytes (interleaved) */\
-  b0 = _mm_xor_si128(b0,  b0);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a0 = _mm_aesenclast_si128(a0, b0);\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a1 = _mm_aesenclast_si128(a1, b0);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a2 = _mm_aesenclast_si128(a2, b0);\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a3 = _mm_aesenclast_si128(a3, b0);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a4 = _mm_aesenclast_si128(a4, b0);\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a5 = _mm_aesenclast_si128(a5, b0);\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a6 = _mm_aesenclast_si128(a6, b0);\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  a7 = _mm_aesenclast_si128(a7, b0);\
-  \
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(1, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(2, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(3, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(4, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(5, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(6, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(7, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(8, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(9, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
-  \
-  i0 = _mm_shuffle_epi8(i0, t0);\
-  i1 = _mm_shuffle_epi8(i1, t0);\
-  i2 = _mm_shuffle_epi8(i2, t0);\
-  i3 = _mm_shuffle_epi8(i3, t0);\
-  \
-  o1 = _mm_unpackhi_epi16(i0, i1);\
-  i0 = _mm_unpacklo_epi16(i0, i1);\
-  t0 = _mm_unpackhi_epi16(i2, i3);\
-  i2 = _mm_unpacklo_epi16(i2, i3);\
-  \
-  i0 = _mm_shuffle_epi32(i0, 216);\
-  o1 = _mm_shuffle_epi32(o1, 216);\
-  i2 = _mm_shuffle_epi32(i2, 216);\
-  t0 = _mm_shuffle_epi32(t0, 216);\
-  \
-  o2 = _mm_unpackhi_epi32(i0, i2);\
-  o3 = _mm_unpackhi_epi32(o1, t0);\
-  i0 = _mm_unpacklo_epi32(i0, i2);\
-  o1 = _mm_unpacklo_epi32(o1, t0);\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  o1 = _mm_unpackhi_epi64(i0, i4);\
-  i0 = _mm_unpacklo_epi64(i0, i4);\
-  o2 = _mm_unpacklo_epi64(i1, i5);\
-  o3 = _mm_unpackhi_epi64(i1, i5);\
-  o4 = _mm_unpacklo_epi64(i2, i6);\
-  o5 = _mm_unpackhi_epi64(i2, i6);\
-  o6 = _mm_unpacklo_epi64(i3, i7);\
-  o7 = _mm_unpackhi_epi64(i3, i7);\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  o0 = _mm_unpackhi_epi64(i0, i1);\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  o1 = _mm_unpackhi_epi64(i2, i3);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  o2 = _mm_unpackhi_epi64(i4, i5);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  o3 = _mm_unpackhi_epi64(i6, i7);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  t0 = _mm_xor_si128(t0, t0);\
-  i1 = _mm_unpackhi_epi64(i0, t0);\
-  i0 = _mm_unpacklo_epi64(i0, t0);\
-  i3 = _mm_unpackhi_epi64(i2, t0);\
-  i2 = _mm_unpacklo_epi64(i2, t0);\
-  i5 = _mm_unpackhi_epi64(i4, t0);\
-  i4 = _mm_unpacklo_epi64(i4, t0);\
-  i7 = _mm_unpackhi_epi64(i6, t0);\
-  i6 = _mm_unpacklo_epi64(i6, t0);\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, /*xmm1,*/ xmm2, /*xmm3, xmm4, xmm5,*/ xmm6, xmm7;
-  static __m128i /*xmm8, xmm9, xmm10, xmm11,*/ xmm12, xmm13, xmm14, xmm15;
-
-  /* load IV into registers xmm12 - xmm15 */
-  xmm12 = chaining[0];
-  xmm13 = chaining[1];
-  xmm14 = chaining[2];
-  xmm15 = chaining[3];
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* store transposed IV */
-  chaining[0] = xmm12;
-  chaining[1] = xmm2;
-  chaining[2] = xmm6;
-  chaining[3] = xmm7;
-}
-
-void TF512(u64* h, u64* m)
-{
-  __m128i* const chaining = (__m128i*) h;
-  __m128i* const message = (__m128i*) m;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP0;
-  static __m128i TEMP1;
-  static __m128i TEMP2;
-  static __m128i TEMP3;
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  /* load message into registers xmm12 - xmm15 */
-  xmm12 = message[0];
-  xmm13 = message[1];
-  xmm14 = message[2];
-  xmm15 = message[3];
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* load previous chaining value and xor message to CV to get input of P */
-  /* we first put two rows (2x64 bit) of the CV into one 128-bit xmm register */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  xmm8 = _mm_xor_si128(xmm12, chaining[0]);
-  xmm0 = _mm_xor_si128(xmm2,  chaining[1]);
-  xmm4 = _mm_xor_si128(xmm6,  chaining[2]);
-  xmm5 = _mm_xor_si128(xmm7,  chaining[3]);
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(xmm8, xmm0, xmm4, xmm5, xmm12, xmm2, xmm6, xmm7, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, xmm8);
-  xmm1 = _mm_xor_si128(xmm1, xmm10);
-  xmm2 = _mm_xor_si128(xmm2, xmm12);
-  xmm3 = _mm_xor_si128(xmm3, xmm14);
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, chaining[0]);
-  xmm1 = _mm_xor_si128(xmm1, chaining[1]);
-  xmm2 = _mm_xor_si128(xmm2, chaining[2]);
-  xmm3 = _mm_xor_si128(xmm3, chaining[3]);
-
-  /* store CV */
-  chaining[0] = xmm0;
-  chaining[1] = xmm1;
-  chaining[2] = xmm2;
-  chaining[3] = xmm3;
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP0;
-  static __m128i TEMP1;
-  static __m128i TEMP2;
-  static __m128i TEMP3;
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = chaining[0];
-  xmm10 = chaining[1];
-  xmm12 = chaining[2];
-  xmm14 = chaining[3];
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = _mm_xor_si128(xmm8,  (chaining[0]));
-  xmm10 = _mm_xor_si128(xmm10, (chaining[1]));
-  xmm12 = _mm_xor_si128(xmm12, (chaining[2]));
-  xmm14 = _mm_xor_si128(xmm14, (chaining[3]));
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(xmm8, xmm10, xmm12, xmm14, xmm4, xmm9, xmm11, xmm0);
-
-  /* we only need to return the truncated half of the state */
-  chaining[2] = xmm9;
-  chaining[3] = xmm11;
-}
-
-
--- a/algo/groestl/aes_ni/groestl256-intr-vperm.h
+++ b/algo/groestl/aes_ni/groestl256-intr-vperm.h
@@ -1,793 +0,0 @@
-/* groestl-intr-vperm.h     Aug 2011
- *
- * Groestl implementation with intrinsics using ssse3 instructions.
- * Author: Günther A. Roland, Martin Schläffer
- *
- * Based on the vperm and aes_ni implementations of the hash function Groestl
- * by Cagdas Calik <ccalik@metu.edu.tr> http://www.metu.edu.tr/~ccalik/
- * Institute of Applied Mathematics, Middle East Technical University, Turkey
- *
- * This code is placed in the public domain
- */
-
-#include <tmmintrin.h>
-#include "hash-groestl256.h"
-
-/* global constants  */
-__m128i ROUND_CONST_Lx;
-__m128i ROUND_CONST_L0[ROUNDS512];
-__m128i ROUND_CONST_L7[ROUNDS512];
-__m128i ROUND_CONST_P[ROUNDS1024];
-__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_0F;
-__m128i ALL_15;
-__m128i ALL_1B;
-__m128i ALL_63;
-__m128i ALL_FF;
-__m128i VPERM_IPT[2];
-__m128i VPERM_OPT[2];
-__m128i VPERM_INV[2];
-__m128i VPERM_SB1[2];
-__m128i VPERM_SB2[2];
-__m128i VPERM_SB4[2];
-__m128i VPERM_SBO[2];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_SHARED_CONSTANTS(){\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  ALL_63 = _mm_set_epi32(0x63636363, 0x63636363, 0x63636363, 0x63636363);\
-  ALL_0F = _mm_set_epi32(0x0f0f0f0f, 0x0f0f0f0f, 0x0f0f0f0f, 0x0f0f0f0f);\
-  ALL_15 = _mm_set_epi32(0x15151515, 0x15151515, 0x15151515, 0x15151515);\
-  VPERM_IPT[0] = _mm_set_epi32(0xCD80B1FC, 0xB0FDCC81, 0x4C01307D, 0x317C4D00);\
-  VPERM_IPT[1] = _mm_set_epi32(0xCABAE090, 0x52227808, 0xC2B2E898, 0x5A2A7000);\
-  VPERM_OPT[0] = _mm_set_epi32(0xE10D5DB1, 0xB05C0CE0, 0x01EDBD51, 0x50BCEC00);\
-  VPERM_OPT[1] = _mm_set_epi32(0xF7974121, 0xDEBE6808, 0xFF9F4929, 0xD6B66000);\
-  VPERM_INV[0] = _mm_set_epi32(0x030D0E0C, 0x02050809, 0x01040A06, 0x0F0B0780);\
-  VPERM_INV[1] = _mm_set_epi32(0x04070309, 0x0A0B0C02, 0x0E05060F, 0x0D080180);\
-  VPERM_SB1[0] = _mm_set_epi32(0x3BF7CCC1, 0x0D2ED9EF, 0x3618D415, 0xFAE22300);\
-  VPERM_SB1[1] = _mm_set_epi32(0xA5DF7A6E, 0x142AF544, 0xB19BE18F, 0xCB503E00);\
-  VPERM_SB2[0] = _mm_set_epi32(0xC2A163C8, 0xAB82234A, 0x69EB8840, 0x0AE12900);\
-  VPERM_SB2[1] = _mm_set_epi32(0x5EB7E955, 0xBC982FCD, 0xE27A93C6, 0x0B712400);\
-  VPERM_SB4[0] = _mm_set_epi32(0xBA44FE79, 0x876D2914, 0x3D50AED7, 0xC393EA00);\
-  VPERM_SB4[1] = _mm_set_epi32(0xA876DE97, 0x49087E9F, 0xE1E937A0, 0x3FD64100);\
-}/**/
-
-/* VPERM
- * Transform w/o settings c*
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2){\
-  t0 = c0;\
-  t1 = c0;\
-  t0 = _mm_andnot_si128(t0, a0);\
-  t1 = _mm_andnot_si128(t1, a1);\
-  t0 = _mm_srli_epi32(t0, 4);\
-  t1 = _mm_srli_epi32(t1, 4);\
-  a0 = _mm_and_si128(a0, c0);\
-  a1 = _mm_and_si128(a1, c0);\
-  t2 = c2;\
-  t3 = c2;\
-  t2 = _mm_shuffle_epi8(t2, a0);\
-  t3 = _mm_shuffle_epi8(t3, a1);\
-  a0 = c1;\
-  a1 = c1;\
-  a0 = _mm_shuffle_epi8(a0, t0);\
-  a1 = _mm_shuffle_epi8(a1, t1);\
-  a0 = _mm_xor_si128(a0, t2);\
-  a1 = _mm_xor_si128(a1, t3);\
-}/**/
-
-#define VPERM_Transform_Set_Const(table, c0, c1, c2){\
-  c0 = ALL_0F;\
-  c1 = ((__m128i*) table )[0];\
-  c2 = ((__m128i*) table )[1];\
-}/**/
-
-/* VPERM
- * Transform
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform(a0, a1, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Transform State
- * inputs:
- * a0-a3 = state
- * table = transformation table to use
- * t* = clobbers
- * outputs:
- * a0-a3 = transformed state
- * */
-#define VPERM_Transform_State(a0, a1, a2, a3, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-  VPERM_Transform_No_Const(a2, a3, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Add Constant to State
- * inputs:
- * a0-a7 = state
- * constant = constant to add
- * t0 = clobber
- * outputs:
- * a0-a7 = state + constant
- * */
-#define VPERM_Add_Constant(a0, a1, a2, a3, a4, a5, a6, a7, constant, t0){\
-  t0 = constant;\
-  a0 = _mm_xor_si128(a0,  t0);\
-  a1 = _mm_xor_si128(a1,  t0);\
-  a2 = _mm_xor_si128(a2,  t0);\
-  a3 = _mm_xor_si128(a3,  t0);\
-  a4 = _mm_xor_si128(a4,  t0);\
-  a5 = _mm_xor_si128(a5,  t0);\
-  a6 = _mm_xor_si128(a6,  t0);\
-  a7 = _mm_xor_si128(a7,  t0);\
-}/**/
-
-/* VPERM
- * Set Substitute Core Constants
- * */
-#define VPERM_Substitute_Core_Set_Const(c0, c1, c2){\
-  VPERM_Transform_Set_Const(VPERM_INV, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Substitute Core
- * first part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0 = 1 row
- * t*, c* = clobbers
- * outputs:
- * b0a, b0b = inputs for lookup step
- * */
-#define VPERM_Substitute_Core(a0, b0a, b0b, t0, t1, c0, c1, c2){\
-  t0 = c0;\
-  t0 = _mm_andnot_si128(t0, a0);\
-  t0 = _mm_srli_epi32(t0, 4);\
-  a0 = _mm_and_si128(a0,  c0);\
-  b0a = c1;\
-  b0a = _mm_shuffle_epi8(b0a, a0);\
-  a0 = _mm_xor_si128(a0,  t0);\
-  b0b = c2;\
-  b0b = _mm_shuffle_epi8(b0b, t0);\
-  b0b = _mm_xor_si128(b0b, b0a);\
-  t1 = c2;\
-  t1 = _mm_shuffle_epi8(t1,  a0);\
-  t1 = _mm_xor_si128(t1,  b0a);\
-  b0a = c2;\
-  b0a = _mm_shuffle_epi8(b0a, b0b);\
-  b0a = _mm_xor_si128(b0a, a0);\
-  b0b = c2;\
-  b0b = _mm_shuffle_epi8(b0b, t1);\
-  b0b = _mm_xor_si128(b0b, t0);\
-}/**/
-
-/* VPERM
- * Lookup
- * second part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0a, a0b = output of Substitution Core
- * table = lookup table to use (*1 / *2 / *4)
- * t0 = clobber
- * outputs:
- * b0 = output of sbox + multiplication
- * */
-#define VPERM_Lookup(a0a, a0b, table, b0, t0){\
-  b0 = ((__m128i*) table )[0];\
-  t0 = ((__m128i*) table )[1];\
-  b0 = _mm_shuffle_epi8(b0, a0b);\
-  t0 = _mm_shuffle_epi8(t0, a0a);\
-  b0 = _mm_xor_si128(b0, t0);\
-}/**/
-
-/* VPERM
- * SubBytes and *2 / *4
- * this function is derived from:
- *   Constant-time SSSE3 AES core implementation
- *   by Mike Hamburg
- * and
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0-a7 = state
- * t*, c* = clobbers
- * outputs:
- * a0-a7 = state * 4
- * c2 = row0 * 2 -> b0
- * c1 = row7 * 2 -> b3
- * c0 = row7 * 1 -> b4
- * t2 = row4 * 1 -> b7
- * TEMP_MUL1 = row(i) * 1
- * TEMP_MUL2 = row(i) * 2
- *
- * call:VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7) */
-#define VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, t0, t1, t3, t4, c2, c1, c0, t2){\
-  /* set Constants */\
-  VPERM_Substitute_Core_Set_Const(c0, c1, c2);\
-  /* row 1 */\
-  VPERM_Substitute_Core(a1, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[1] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[1] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a1, t4);\
-  /* --- */\
-  /* row 2 */\
-  VPERM_Substitute_Core(a2, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[2] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[2] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a2, t4);\
-  /* --- */\
-  /* row 3 */\
-  VPERM_Substitute_Core(a3, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[3] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[3] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a3, t4);\
-  /* --- */\
-  /* row 5 */\
-  VPERM_Substitute_Core(a5, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[5] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[5] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a5, t4);\
-  /* --- */\
-  /* row 6 */\
-  VPERM_Substitute_Core(a6, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[6] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[6] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a6, t4);\
-  /* --- */\
-  /* row 7 */\
-  VPERM_Substitute_Core(a7, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[7] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c1, t4); /*c1 -> b3*/\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a7, t4);\
-  /* --- */\
-  /* row 4 */\
-  VPERM_Substitute_Core(a4, t0, t1, t3, t4, c0, (VPERM_INV[0]), c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4); /*t2 -> b7*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[4] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a4, t4);\
-  /* --- */\
-  /* row 0 */\
-  VPERM_Substitute_Core(a0, t0, t1, t3, t4, c0, (VPERM_INV[0]), c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, c0, t4); /*c0 -> b4*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c2, t4); /*c2 -> b0*/\
-  TEMP_MUL2[0] = c2;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a0, t4);\
-  /* --- */\
-}/**/
-
-
-/* Optimized MixBytes
- * inputs:
- * a0-a7 = (row0-row7) * 4
- * b0 = row0 * 2
- * b3 = row7 * 2
- * b4 = row7 * 1
- * b7 = row4 * 1
- * all *1 and *2 values must also be in TEMP_MUL1, TEMP_MUL2
- * output: b0-b7
- * */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* save one value */\
-  TEMP_MUL4 = a3;\
-  /* 1 */\
-  b1 = a0;\
-  b1 = _mm_xor_si128(b1, a5);\
-  b1 = _mm_xor_si128(b1, b4); /* -> helper! */\
-  b1 = _mm_xor_si128(b1, (TEMP_MUL2[3]));\
-  b2 = b1;\
-  \
-  /* 2 */\
-  b5 = a1;\
-  b5 = _mm_xor_si128(b5, a4);\
-  b5 = _mm_xor_si128(b5, b7); /* -> helper! */\
-  b5 = _mm_xor_si128(b5, b3); /* -> helper! */\
-  b6 = b5;\
-  \
-  /* 4 */\
-  b7 = _mm_xor_si128(b7, a6);\
-  /*b7 = _mm_xor_si128(b7, (TEMP_MUL1[4])); -> helper! */\
-  b7 = _mm_xor_si128(b7, (TEMP_MUL1[6]));\
-  b7 = _mm_xor_si128(b7, (TEMP_MUL2[1]));\
-  b7 = _mm_xor_si128(b7, b3); /* -> helper! */\
-  b2 = _mm_xor_si128(b2, b7);\
-  \
-  /* 3 */\
-  b0 = _mm_xor_si128(b0, a7);\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL1[5]));\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL1[7]));\
-  /*b0 = _mm_xor_si128(b0, (TEMP_MUL2[0])); -> helper! */\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL2[2]));\
-  b3 = b0;\
-  b1 = _mm_xor_si128(b1, b0);\
-  b0 = _mm_xor_si128(b0, b7); /* moved from 4 */\
-  \
-  /* 5 */\
-  b4 = _mm_xor_si128(b4, a2);\
-  /*b4 = _mm_xor_si128(b4, (TEMP_MUL1[0])); -> helper! */\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL1[2]));\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL2[3]));\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL2[5]));\
-  b3 = _mm_xor_si128(b3, b4);\
-  b6 = _mm_xor_si128(b6, b4);\
-  \
-  /* 6 */\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL1[1]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL1[3]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[4]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[6]));\
-  b4 = _mm_xor_si128(b4, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  b7 = _mm_xor_si128(b7, a3);\
-  \
-  /* 7 */\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL1[1]));\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL2[4]));\
-  b2 = _mm_xor_si128(b2, a1);\
-  b3 = _mm_xor_si128(b3, a1);\
-  \
-  /* 8 */\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL1[5]));\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL2[0]));\
-  b6 = _mm_xor_si128(b6, a5);\
-  b7 = _mm_xor_si128(b7, a5);\
-  \
-  /* 9 */\
-  a3 = TEMP_MUL1[2];\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[5]));\
-  b0 = _mm_xor_si128(b0, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  \
-  /* 10 */\
-  a1 = TEMP_MUL1[6];\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL2[1]));\
-  b1 = _mm_xor_si128(b1, a1);\
-  b4 = _mm_xor_si128(b4, a1);\
-  \
-  /* 11 */\
-  a5 = TEMP_MUL1[3];\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL2[6]));\
-  b1 = _mm_xor_si128(b1, a5);\
-  b6 = _mm_xor_si128(b6, a5);\
-  \
-  /* 12 */\
-  a3 = TEMP_MUL1[7];\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[2]));\
-  b2 = _mm_xor_si128(b2, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  \
-  /* 13 */\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL4));\
-  b0 = _mm_xor_si128(b0, a4);\
-  b1 = _mm_xor_si128(b1, a4);\
-  b3 = _mm_xor_si128(b3, a6);\
-  b4 = _mm_xor_si128(b4, a0);\
-  b4 = _mm_xor_si128(b4, a7);\
-  b5 = _mm_xor_si128(b5, a0);\
-  b7 = _mm_xor_si128(b7, a2);\
-}/**/
-
-#define SET_CONSTANTS(){\
-  SET_SHARED_CONSTANTS();\
-  SUBSH_MASK[0] = _mm_set_epi32(0x080f0e0d, 0x0c0b0a09, 0x07060504, 0x03020100);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x0a09080f, 0x0e0d0c0b, 0x00070605, 0x04030201);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x0c0b0a09, 0x080f0e0d, 0x01000706, 0x05040302);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0e0d0c0b, 0x0a09080f, 0x02010007, 0x06050403);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0f0e0d0c, 0x0b0a0908, 0x03020100, 0x07060504);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x09080f0e, 0x0d0c0b0a, 0x04030201, 0x00070605);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x0b0a0908, 0x0f0e0d0c, 0x05040302, 0x01000706);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x0d0c0b0a, 0x09080f0e, 0x06050403, 0x02010007);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}/**/
-
-/* vperm:
- * transformation before rounds with ipt
- * first round add transformed constant
- * middle rounds: add constant XOR 0x15...15
- * last round: additionally add 0x15...15 after MB
- * transformation after rounds with opt
- */
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant + ShiftBytes (interleaved) */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a4 = _mm_xor_si128(a4, b1);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  /* SubBytes + Multiplication by 2 and 4 */\
-  VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7);\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}/**/
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  VPERM_Add_Constant(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, ALL_15, xmm0);\
-  ROUND(0, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(1, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(2, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(3, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(4, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(5, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(6, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(7, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(8, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(9, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  VPERM_Add_Constant(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, ALL_15, xmm0);\
-}
-
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
-\
-  i0 = _mm_shuffle_epi8(i0, t0);\
-  i1 = _mm_shuffle_epi8(i1, t0);\
-  i2 = _mm_shuffle_epi8(i2, t0);\
-  i3 = _mm_shuffle_epi8(i3, t0);\
-\
-  o1 = i0;\
-  t0 = i2;\
-\
-  i0 = _mm_unpacklo_epi16(i0, i1);\
-  o1 = _mm_unpackhi_epi16(o1, i1);\
-  i2 = _mm_unpacklo_epi16(i2, i3);\
-  t0 = _mm_unpackhi_epi16(t0, i3);\
-\
-  i0 = _mm_shuffle_epi32(i0, 216);\
-  o1 = _mm_shuffle_epi32(o1, 216);\
-  i2 = _mm_shuffle_epi32(i2, 216);\
-  t0 = _mm_shuffle_epi32(t0, 216);\
-\
-  o2 = i0;\
-  o3 = o1;\
-\
-  i0 = _mm_unpacklo_epi32(i0, i2);\
-  o1 = _mm_unpacklo_epi32(o1, t0);\
-  o2 = _mm_unpackhi_epi32(o2, i2);\
-  o3 = _mm_unpackhi_epi32(o3, t0);\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  o1 = i0;\
-  o2 = i1;\
-  i0 = _mm_unpacklo_epi64(i0, i4);\
-  o1 = _mm_unpackhi_epi64(o1, i4);\
-  o3 = i1;\
-  o4 = i2;\
-  o2 = _mm_unpacklo_epi64(o2, i5);\
-  o3 = _mm_unpackhi_epi64(o3, i5);\
-  o5 = i2;\
-  o6 = i3;\
-  o4 = _mm_unpacklo_epi64(o4, i6);\
-  o5 = _mm_unpackhi_epi64(o5, i6);\
-  o7 = i3;\
-  o6 = _mm_unpacklo_epi64(o6, i7);\
-  o7 = _mm_unpackhi_epi64(o7, i7);\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  o0 = i0;\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  o0 = _mm_unpackhi_epi64(o0, i1);\
-  o1 = i2;\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  o1 = _mm_unpackhi_epi64(o1, i3);\
-  o2 = i4;\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  o2 = _mm_unpackhi_epi64(o2, i5);\
-  o3 = i6;\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-  o3 = _mm_unpackhi_epi64(o3, i7);\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  t0 = _mm_xor_si128(t0, t0);\
-  i1 = i0;\
-  i3 = i2;\
-  i5 = i4;\
-  i7 = i6;\
-  i0 = _mm_unpacklo_epi64(i0, t0);\
-  i1 = _mm_unpackhi_epi64(i1, t0);\
-  i2 = _mm_unpacklo_epi64(i2, t0);\
-  i3 = _mm_unpackhi_epi64(i3, t0);\
-  i4 = _mm_unpacklo_epi64(i4, t0);\
-  i5 = _mm_unpackhi_epi64(i5, t0);\
-  i6 = _mm_unpacklo_epi64(i6, t0);\
-  i7 = _mm_unpackhi_epi64(i7, t0);\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst_CNT2(i, j){\
-  xmm0 = ROUND_CONST_L0[i];\
-  xmm1 = ROUND_CONST_L7[i];\
-  xmm2 = ROUND_CONST_L0[j];\
-  xmm3 = ROUND_CONST_L7[j];\
-  VPERM_Transform_State(xmm0, xmm1, xmm2, xmm3, VPERM_IPT, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10);\
-  xmm0 = _mm_xor_si128(xmm0, (ALL_15));\
-  xmm1 = _mm_xor_si128(xmm1, (ALL_15));\
-  xmm2 = _mm_xor_si128(xmm2, (ALL_15));\
-  xmm3 = _mm_xor_si128(xmm3, (ALL_15));\
-  ROUND_CONST_L0[i] = xmm0;\
-  ROUND_CONST_L7[i] = xmm1;\
-  ROUND_CONST_L0[j] = xmm2;\
-  ROUND_CONST_L7[j] = xmm3;\
-}/**/
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst(){\
-  xmm0 = ROUND_CONST_Lx;\
-  VPERM_Transform(xmm0, xmm1, VPERM_IPT, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10);\
-  xmm0 = _mm_xor_si128(xmm0, (ALL_15));\
-  ROUND_CONST_Lx = xmm0;\
-  VPERM_Transform_RoundConst_CNT2(0, 1);\
-  VPERM_Transform_RoundConst_CNT2(2, 3);\
-  VPERM_Transform_RoundConst_CNT2(4, 5);\
-  VPERM_Transform_RoundConst_CNT2(6, 7);\
-  VPERM_Transform_RoundConst_CNT2(8, 9);\
-}/**/
-
-void INIT256(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, /*xmm11,*/ xmm12, xmm13, xmm14, xmm15;
-
-  /* transform round constants into VPERM mode */
-  VPERM_Transform_RoundConst();
-
-  /* load IV into registers xmm12 - xmm15 */
-  xmm12 = chaining[0];
-  xmm13 = chaining[1];
-  xmm14 = chaining[2];
-  xmm15 = chaining[3];
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  VPERM_Transform_State(xmm12, xmm13, xmm14, xmm15, VPERM_IPT, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* store transposed IV */
-  chaining[0] = xmm12;
-  chaining[1] = xmm2;
-  chaining[2] = xmm6;
-  chaining[3] = xmm7;
-}
-
-void TF512(u64* h, u64* m)
-{
-  __m128i* const chaining = (__m128i*) h;
-  __m128i* const message = (__m128i*) m;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP_MUL1[8];
-  static __m128i TEMP_MUL2[8];
-  static __m128i TEMP_MUL4;
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  /* load message into registers xmm12 - xmm15 */
-  xmm12 = message[0];
-  xmm13 = message[1];
-  xmm14 = message[2];
-  xmm15 = message[3];
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  VPERM_Transform_State(xmm12, xmm13, xmm14, xmm15, VPERM_IPT, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  xmm8 = chaining[0];
-  xmm0 = chaining[1];
-  xmm4 = chaining[2];
-  xmm5 = chaining[3];
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  xmm8 = _mm_xor_si128(xmm8, xmm12);
-  xmm0 = _mm_xor_si128(xmm0, xmm2);
-  xmm4 = _mm_xor_si128(xmm4, xmm6);
-  xmm5 = _mm_xor_si128(xmm5, xmm7);
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(xmm8, xmm0, xmm4, xmm5, xmm12, xmm2, xmm6, xmm7, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, xmm8);
-  xmm1 = _mm_xor_si128(xmm1, xmm10);
-  xmm2 = _mm_xor_si128(xmm2, xmm12);
-  xmm3 = _mm_xor_si128(xmm3, xmm14);
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, (chaining[0]));
-  xmm1 = _mm_xor_si128(xmm1, (chaining[1]));
-  xmm2 = _mm_xor_si128(xmm2, (chaining[2]));
-  xmm3 = _mm_xor_si128(xmm3, (chaining[3]));
-
-  /* store CV */
-  chaining[0] = xmm0;
-  chaining[1] = xmm1;
-  chaining[2] = xmm2;
-  chaining[3] = xmm3;
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-
-  return;
-}
-
-void OF512(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP_MUL1[8];
-  static __m128i TEMP_MUL2[8];
-  static __m128i TEMP_MUL4;
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = chaining[0];
-  xmm10 = chaining[1];
-  xmm12 = chaining[2];
-  xmm14 = chaining[3];
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = _mm_xor_si128(xmm8,  (chaining[0]));
-  xmm10 = _mm_xor_si128(xmm10, (chaining[1]));
-  xmm12 = _mm_xor_si128(xmm12, (chaining[2]));
-  xmm14 = _mm_xor_si128(xmm14, (chaining[3]));
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(xmm8, xmm10, xmm12, xmm14, xmm4, xmm9, xmm11, xmm0);
-  VPERM_Transform(xmm9, xmm11, VPERM_OPT, xmm0, xmm1, xmm2, xmm3, xmm5, xmm6, xmm7);
-
-  /* we only need to return the truncated half of the state */
-  chaining[2] = xmm9;
-  chaining[3] = xmm11;
-
-  return;
-}//OF512()
-
-
-
--- a/algo/groestl/aes_ni/hash-groestl.c
+++ b/algo/groestl/aes_ni/hash-groestl.c
@@ -14,50 +14,15 @@
 #include "miner.h"
 #include "simd-utils.h"

-#ifndef NO_AES_NI
+#ifdef __AES__

-#include "groestl-version.h"
-
-#ifdef TASM
-  #ifdef VAES
-    #include "groestl-asm-aes.h"
-  #else
-    #ifdef VAVX
-      #include "groestl-asm-avx.h"
-    #else
-      #ifdef VVPERM
-        #include "groestl-asm-vperm.h"
-      #else
-        #error NO VERSION SPECIFIED (-DV[AES/AVX/VVPERM])
-      #endif
-    #endif
-  #endif
-#else
-  #ifdef TINTR
-    #ifdef VAES
-      #include "groestl-intr-aes.h"
-    #else
-      #ifdef VAVX
-        #include "groestl-intr-avx.h"
-      #else
-        #ifdef VVPERM
-          #include "groestl-intr-vperm.h"
-        #else
-          #error NO VERSION SPECIFIED (-DV[AES/AVX/VVPERM])
-        #endif
-      #endif
-    #endif
-  #else
-    #error NO TYPE SPECIFIED (-DT[ASM/INTR])
-  #endif
-#endif
+#include "groestl-intr-aes.h"

 HashReturn_gr init_groestl( hashState_groestl* ctx, int hashlen )
 {
  int i;

  ctx->hashlen = hashlen;
-  SET_CONSTANTS();

  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return FAIL_GR;
@@ -67,8 +32,10 @@ HashReturn_gr init_groestl( hashState_groestl* ctx, int hashlen )
     ctx->chaining[i] = _mm_setzero_si128();
     ctx->buffer[i]   = _mm_setzero_si128();
  }
-  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
-  INIT(ctx->chaining);
+
+  // The only non-zero in the IV is len. It can be hard coded.
+  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+
  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;

@@ -87,8 +54,7 @@ HashReturn_gr reinit_groestl( hashState_groestl* ctx )
     ctx->chaining[i] = _mm_setzero_si128();
     ctx->buffer[i]   = _mm_setzero_si128();
  }
-  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
-  INIT(ctx->chaining);
+  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;

@@ -104,7 +70,7 @@ HashReturn_gr reinit_groestl( hashState_groestl* ctx )
 // 5. Midstate will work at reduced impact than full hash, if total hash
 //    (midstate + tail) is less than 1 block.
 //    This, unfortunately, is the case with all current users.
-// 6. the morefull blocks the bigger the gain
+// 6. the more full blocks the bigger the gain

 // use only for midstate precalc
 HashReturn_gr update_groestl( hashState_groestl* ctx, const void* input,
@@ -138,12 +104,11 @@ HashReturn_gr update_groestl( hashState_groestl* ctx, const void* input,
 // deprecated do not use
 HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
 {
-   const int len = (int)ctx->databitlen / 128;  // bits to __m128i 
-   const int blocks = ctx->blk_count + 1;       // adjust for final block
-
-   const int rem_ptr = ctx->rem_ptr;      // end of data start of padding
-   const int hashlen_m128i = ctx->hashlen / 16;  // bytes to __m128i
-   const int hash_offset = SIZE512 - hashlen_m128i;  // where in buffer
+   const int len = (int)ctx->databitlen / 128; // bits to __m128i 
+   const uint64_t blocks = ctx->blk_count + 1; // adjust for final block
+   const int rem_ptr = ctx->rem_ptr;           // end of data start of padding
+   const int hashlen_m128i = ctx->hashlen / 16;     // bytes to __m128i
+   const int hash_offset = SIZE512 - hashlen_m128i; // where in buffer
   int i;

   // first pad byte = 0x80, last pad byte = block count
@@ -152,21 +117,18 @@ HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
   if ( rem_ptr == len - 1 )
   {
       // only 128 bits left in buffer, all padding at once
-       ctx->buffer[rem_ptr] = _mm_set_epi8( blocks,0,0,0, 0,0,0,0,
-                                                  0,0,0,0, 0,0,0,0x80 );
+      ctx->buffer[rem_ptr] = _mm_set_epi64x( blocks << 56, 0x80 );
   }
   else
   {
       // add first padding
-       ctx->buffer[rem_ptr] = _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                            0,0,0,0, 0,0,0,0x80 );
+       ctx->buffer[rem_ptr] = m128_const_64( 0, 0x80 );
       // add zero padding
       for ( i = rem_ptr + 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();

       // add length padding, second last byte is zero unless blocks > 255
-       ctx->buffer[i] = _mm_set_epi8( blocks, blocks>>8, 0,0, 0,0,0,0,
-                                           0,         0 ,0,0, 0,0,0,0 );
+       ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0 );
   }

   // digest final padding block and do output transform
@@ -180,6 +142,75 @@ HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
   return SUCCESS_GR;
 }

+int groestl512_full( hashState_groestl* ctx, void* output,
+                                const void* input, uint64_t databitlen )
+{
+
+   int i;
+   ctx->hashlen = 64;
+
+   for ( i = 0; i < SIZE512; i++ )
+   {
+      ctx->chaining[i] = _mm_setzero_si128();
+      ctx->buffer[i]   = _mm_setzero_si128();
+   }
+   ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+   ctx->buf_ptr = 0;
+   ctx->rem_ptr = 0;
+
+   // --- update ---
+   
+   const int len = (int)databitlen / 128;
+   const int hashlen_m128i = ctx->hashlen / 16;   // bytes to __m128i
+   const int hash_offset = SIZE512 - hashlen_m128i;
+   int rem = ctx->rem_ptr;
+   uint64_t blocks = len / SIZE512;
+   __m128i* in = (__m128i*)input;
+
+   // digest any full blocks, process directly from input 
+   for ( i = 0; i < blocks; i++ )
+      TF1024( ctx->chaining, &in[ i * SIZE512 ] );
+   ctx->buf_ptr = blocks * SIZE512;
+
+   // copy any remaining data to buffer, it may already contain data
+   // from a previous update for a midstate precalc
+   for ( i = 0; i < len % SIZE512; i++ )
+       ctx->buffer[ rem + i ] = in[ ctx->buf_ptr + i ];
+   i += rem;    // use i as rem_ptr in final
+
+   //--- final ---
+
+   blocks++;      // adjust for final block
+
+   if ( i == len -1 )
+   {
+       // only 128 bits left in buffer, all padding at once
+      ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0x80 );
+   }
+   else
+   {
+       // add first padding
+       ctx->buffer[i] = m128_const_64( 0, 0x80 );
+       // add zero padding
+       for ( i += 1; i < SIZE512 - 1; i++ )
+           ctx->buffer[i] = _mm_setzero_si128();
+
+       // add length padding, second last byte is zero unless blocks > 255
+       ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0 ); 
+   }
+
+   // digest final padding block and do output transform
+   TF1024( ctx->chaining, ctx->buffer );
+   OF1024( ctx->chaining );
+
+   // store hash result in output 
+   for ( i = 0; i < hashlen_m128i; i++ )
+      casti_m128i( output, i ) = ctx->chaining[ hash_offset + i ];
+
+   return 0;
+}
+   
+
 HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
                                const void* input, DataLength_gr databitlen )
 {
@@ -187,7 +218,7 @@ HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
   const int hashlen_m128i = ctx->hashlen / 16;   // bytes to __m128i
   const int hash_offset = SIZE512 - hashlen_m128i;
   int rem = ctx->rem_ptr;
-   int blocks = len / SIZE512;
+   uint64_t blocks = len / SIZE512;
   __m128i* in = (__m128i*)input;
   int i;

@@ -211,21 +242,18 @@ HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
   if ( i == len -1 )
   {        
       // only 128 bits left in buffer, all padding at once
-       ctx->buffer[i] = _mm_set_epi8( blocks,0,0,0, 0,0,0,0,
-                                           0,0,0,0, 0,0,0,0x80 );
+      ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0x80 );
   }   
   else
   {
       // add first padding
-       ctx->buffer[i] = _mm_set_epi8( 0,0,0,0, 0,0,0,0, 
-                                      0,0,0,0, 0,0,0,0x80 );
+       ctx->buffer[i] = m128_const_64( 0, 0x80 );
       // add zero padding
       for ( i += 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();

       // add length padding, second last byte is zero unless blocks > 255
-       ctx->buffer[i] = _mm_set_epi8( blocks, blocks>>8, 0,0, 0,0,0,0, 
-                                           0,         0 ,0,0, 0,0,0,0 );
+       ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0 );
   }

   // digest final padding block and do output transform
--- a/algo/groestl/aes_ni/hash-groestl.h
+++ b/algo/groestl/aes_ni/hash-groestl.h
@@ -87,5 +87,6 @@ HashReturn_gr final_groestl( hashState_groestl*, void* );

 HashReturn_gr update_and_final_groestl( hashState_groestl*,  void*,
                                        const void*, DataLength_gr );
+int groestl512_full( hashState_groestl*,  void*, const void*, uint64_t );

 #endif /* __hash_h */
--- a/algo/groestl/aes_ni/hash-groestl256.c
+++ b/algo/groestl/aes_ni/hash-groestl256.c
@@ -11,43 +11,9 @@
 #include "miner.h"
 #include "simd-utils.h"

-#ifndef NO_AES_NI
+#ifdef __AES__

-#include "groestl-version.h"
-
-#ifdef TASM
-  #ifdef VAES
-    #include "groestl256-asm-aes.h"
-  #else
-    #ifdef VAVX
-      #include "groestl256-asm-avx.h"
-    #else
-      #ifdef VVPERM
-        #include "groestl256-asm-vperm.h"
-      #else
-        #error NO VERSION SPECIFIED (-DV[AES/AVX/VVPERM])
-      #endif
-    #endif
-  #endif
-#else
-  #ifdef TINTR
-    #ifdef VAES
-      #include "groestl256-intr-aes.h"
-    #else
-      #ifdef VAVX
-        #include "groestl256-intr-avx.h"
-      #else
-        #ifdef VVPERM
-          #include "groestl256-intr-vperm.h"
-        #else
-          #error NO VERSION SPECIFIED (-DV[AES/AVX/VVPERM])
-        #endif
-      #endif
-    #endif
-  #else
-    #error NO TYPE SPECIFIED (-DT[ASM/INTR])
-  #endif
-#endif
+#include "groestl256-intr-aes.h"

 /* initialise context */
 HashReturn_gr init_groestl256( hashState_groestl256* ctx, int hashlen )
@@ -55,7 +21,6 @@ HashReturn_gr init_groestl256( hashState_groestl256* ctx, int hashlen )
  int i;

  ctx->hashlen = hashlen;
-  SET_CONSTANTS();

  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return FAIL_GR;
@@ -86,8 +51,11 @@ HashReturn_gr reinit_groestl256(hashState_groestl256* ctx)
     ctx->chaining[i] = _mm_setzero_si128();
     ctx->buffer[i]   = _mm_setzero_si128();
  }
-  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
-  INIT256(ctx->chaining);
+
+  ctx->chaining[ 3 ] = m128_const_64( 0, 0x0100000000000000 );
+
+//  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
+//  INIT256(ctx->chaining);
  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;

@@ -246,6 +214,98 @@ HashReturn_gr update_and_final_groestl256( hashState_groestl256* ctx,
   return SUCCESS_GR;
 }

+int groestl256_full( hashState_groestl256* ctx,
+                   void* output, const void* input, DataLength_gr databitlen )
+{
+   int i;
+   ctx->hashlen = 32;
+  for ( i = 0; i < SIZE256; i++ )
+  {
+     ctx->chaining[i] = _mm_setzero_si128();
+     ctx->buffer[i]   = _mm_setzero_si128();
+  }
+  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
+  INIT256( ctx->chaining );
+  ctx->buf_ptr = 0;
+  ctx->rem_ptr = 0;
+
+   const int len = (int)databitlen / 128;
+   const int hashlen_m128i = ctx->hashlen / 16;   // bytes to __m128i
+   const int hash_offset = SIZE256 - hashlen_m128i;
+   int rem = ctx->rem_ptr;
+   int blocks = len / SIZE256;
+   __m128i* in = (__m128i*)input;
+
+   // --- update ---
+
+   // digest any full blocks, process directly from input
+   for ( i = 0; i < blocks; i++ )
+      TF512( ctx->chaining, &in[ i * SIZE256 ] );
+   ctx->buf_ptr = blocks * SIZE256;
+
+   // cryptonight has 200 byte input, an odd number of __m128i
+   // remainder is only 8 bytes, ie u64.
+   if ( databitlen % 128 !=0 )
+   {
+      // must be cryptonight, copy 64 bits of data
+      *(uint64_t*)(ctx->buffer) = *(uint64_t*)(&in[ ctx->buf_ptr ] );
+      i = -1; // signal for odd length
+   }
+   else
+   {
+      // Copy any remaining data to buffer for final transform
+      for ( i = 0; i < len % SIZE256; i++ )
+          ctx->buffer[ rem + i ] = in[ ctx->buf_ptr + i ];
+      i += rem;   // use i as rem_ptr in final
+   }
+
+   //--- final ---
+
+   // adjust for final block
+   blocks++;
+
+   if ( i == len - 1 )
+   {
+       // all padding at once
+       ctx->buffer[i] = _mm_set_epi8( blocks,blocks>>8,0,0, 0,0,0,0,
+                                           0,        0,0,0, 0,0,0,0x80 );
+   }
+   else
+   {
+      if ( i == -1 )
+      {
+         // cryptonight odd length
+         ((uint64_t*)ctx->buffer)[ 1 ] = 0x80ull;
+         // finish the block with zero and length padding as normal
+         i = 0;
+       }
+       else
+       {
+          // add first padding
+          ctx->buffer[i] = _mm_set_epi8( 0,0,0,0, 0,0,0,0,
+                                         0,0,0,0, 0,0,0,0x80 );
+       }
+       // add zero padding
+       for ( i += 1; i < SIZE256 - 1; i++ )
+           ctx->buffer[i] = _mm_setzero_si128();
+       // add length padding
+       // cheat since we know the block count is trivial, good if block < 256
+       ctx->buffer[i] = _mm_set_epi8( blocks,blocks>>8,0,0, 0,0,0,0,
+                                           0,        0,0,0, 0,0,0,0 );
+   }
+
+   // digest final padding block and do output transform
+   TF512( ctx->chaining, ctx->buffer );
+   OF512( ctx->chaining );
+
+   // store hash result in output 
+   for ( i = 0; i < hashlen_m128i; i++ )
+      casti_m128i( output, i ) = ctx->chaining[ hash_offset + i ];
+
+   return SUCCESS_GR;
+}
+
+
 /* hash bit sequence */
 HashReturn_gr hash_groestl256(int hashbitlen,
                const BitSequence_gr* data,
--- a/algo/groestl/aes_ni/hash-groestl256.h
+++ b/algo/groestl/aes_ni/hash-groestl256.h
@@ -93,9 +93,6 @@ typedef enum
 typedef struct {
  __attribute__ ((aligned (32))) __m128i chaining[SIZE256];
  __attribute__ ((aligned (32))) __m128i buffer[SIZE256];
-//  __attribute__ ((aligned (32))) u64 chaining[SIZE/8];      /* actual state */
-//  __attribute__ ((aligned (32))) BitSequence_gr buffer[SIZE];  /* data buffer */
-//  u64 block_counter;        /* message block counter */
  int hashlen;              // bytes
  int blk_count;
  int buf_ptr;              /* data buffer pointer */
@@ -118,4 +115,7 @@ HashReturn_gr hash_groestli256( int, const BitSequence_gr*, DataLength_gr,
 HashReturn_gr update_and_final_groestl256( hashState_groestl256*, void*,
                                           const void*, DataLength_gr );

+int groestl256_full( hashState_groestl256* ctx,
+                   void* output, const void* input, DataLength_gr databitlen );
+
 #endif /* __hash_h */
--- a/algo/groestl/groestl-4way.c
+++ b/algo/groestl/groestl-4way.c
@@ -0,0 +1,64 @@
+#include "groestl-gate.h"
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+
+#if defined(GROESTL_4WAY_VAES)
+
+#include "groestl512-hash-4way.h"
+
+void groestl_4way_hash( void *output, const void *input )
+{
+     uint32_t hash[16*4] __attribute__ ((aligned (128)));
+     groestl512_4way_context ctx;
+
+     groestl512_4way_init( &ctx, 64 );
+     groestl512_4way_update_close( &ctx, hash, input, 640 );
+
+     groestl512_4way_init( &ctx, 64 );
+     groestl512_4way_update_close( &ctx, hash, hash, 512 );
+
+     dintrlv_4x128( output, output+32, output+64, output+96, hash, 256 );
+ }
+
+int scanhash_groestl_4way( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
+{
+     uint32_t hash[8*4] __attribute__ ((aligned (128)));
+     uint32_t vdata[24*4] __attribute__ ((aligned (64)));
+     uint32_t *pdata = work->data;
+     uint32_t *ptarget = work->target;
+     uint32_t n = pdata[19];
+     const uint32_t first_nonce = pdata[19];
+     const uint32_t last_nonce = max_nonce - 4;
+     uint32_t *noncep = vdata + 64+3;   // 4*16 + 3
+     int thr_id = mythr->id;
+     const uint32_t Htarg = ptarget[7];
+
+     mm512_bswap32_intrlv80_4x128( vdata, pdata );
+
+     do
+     {
+        be32enc( noncep,    n   );
+        be32enc( noncep+ 4, n+1 );
+        be32enc( noncep+ 8, n+2 );
+        be32enc( noncep+12, n+3 );
+
+        groestl_4way_hash( hash, vdata );
+        pdata[19] = n;
+
+        for ( int lane = 0; lane < 4; lane++ )
+        if ( ( hash+(lane<<3) )[7] <= Htarg )
+        if ( fulltest( hash+(lane<<3), ptarget) && !opt_benchmark )
+        {
+           pdata[19] = n + lane;
+           submit_solution( work, hash+(lane<<3), mythr );
+        }
+        n += 4;
+     } while ( ( n < last_nonce ) && !work_restart[thr_id].restart );
+     *hashes_done = n - first_nonce;
+     return 0;
+}
+
+#endif
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Jay D Dee	19cc88d102	v3.16.5	2021-06-26 12:27:44 -04:00
Jay D Dee	a053690170	v3.16.4	2021-06-23 21:52:42 -04:00
Jay D Dee	3c5e8921b7	v3.16.3	2021-05-06 14:55:03 -04:00
Jay D Dee	f3333b0070	v3.16.2	2021-04-08 18:09:31 -04:00
Jay D Dee	902ec046dd	v3.16.1	2021-03-24 18:24:20 -04:00
Jay D Dee	d0b4941321	v3.16.0	2021-03-19 15:45:32 -04:00
Jay D Dee	40089428c5	v3.15.7	2021-03-08 22:44:44 -05:00
Jay D Dee	dc6b007a18	v3.15.6	2021-02-12 15:16:53 -05:00
Jay D Dee	06bfaa1249	v3.15.5	2020-12-21 13:25:33 -05:00
Jay D Dee	6566e99a13	v3.15.4	2020-12-15 13:15:02 -05:00
Jay D Dee	ccfccbadd5	v3.15.3	2020-12-10 18:23:49 -05:00
Jay D Dee	45ecd0de14	v3.15.2	2020-11-15 17:57:06 -05:00
Jay D Dee	4fa8fcea8b	v3.15.1	2020-11-09 13:19:05 -05:00
Jay D Dee	c85fb3842b	v3.15.0	2020-10-02 10:48:37 -04:00
Jay D Dee	cdd587537e	v3.14.3	2020-06-18 17:30:26 -04:00
Jay D Dee	51a1d91abd	v3.14.2	2020-05-30 21:20:44 -04:00
Jay D Dee	13563e2598	v3.14.1	2020-05-21 13:00:29 -04:00
Jay D Dee	9571f85d53	v3.14.0	2020-05-20 13:56:35 -04:00
Jay D Dee	0e69756634	v3.13.2-segwit-test	2020-05-18 18:17:27 -04:00
Jay D Dee	9653bca1e2	v3.13.1.1	2020-05-17 19:21:37 -04:00
Jay D Dee	1c0719e8a4	v3.13.1	2020-05-10 21:34:55 -04:00
Jay D Dee	8b4b4dc613	v3.13.0.1	2020-05-07 17:57:04 -04:00
Jay D Dee	e76feaced8	v3.13.0	2020-05-06 00:53:43 -04:00
Jay D Dee	5e088d00d0	v3.12.8.2	2020-04-24 21:18:56 -04:00
Jay D Dee	972d4d70db	v3.12.8.1	2020-04-17 16:12:45 -04:00
Jay D Dee	e96a6bd699	v3.12.8	2020-04-09 12:56:18 -04:00
Jay D Dee	fb9163185a	v3.12.7	2020-03-20 16:30:12 -04:00
Jay D Dee	6e8b8ed34f	v3.12.6.1	2020-03-07 14:11:06 -05:00
Jay D Dee	c0aadbcc99	v3.12.6	2020-03-05 18:43:20 -05:00
Jay D Dee	3da149418a	v3.12.5	2020-03-01 13:18:17 -05:00
Jay D Dee	720610cce5	v3.12.4.6	2020-02-28 18:20:32 -05:00
Jay D Dee	cedcf4d070	v3.12.4.5	2020-02-28 02:42:22 -05:00
Jay D Dee	81b50c3c71	v3.12.4.4	2020-02-25 14:07:32 -05:00
Jay D Dee	0e1e88f53e	v3.12.4.3	2020-02-24 21:35:19 -05:00
Jay D Dee	45c77a5c81	v3.12.4.2	2020-02-23 15:31:06 -05:00
Jay D Dee	dbce7e0721	v3.12.4.1	2020-02-22 18:06:39 -05:00
Jay D Dee	6d66051de6	v3.12.4	2020-02-21 16:34:53 -05:00
Jay D Dee	b93be8816a	v3.12.3.1	2020-02-18 12:05:47 -05:00
Jay D Dee	19b0ac6d5c	v3.12.3	2020-02-13 04:25:33 -05:00
Jay D Dee	3da2b958cf	v3.12.2	2020-02-09 13:30:40 -05:00
Jay D Dee	dc2f8d81d3	v3.12.1	2020-02-07 20:18:20 -05:00
Jay D Dee	fc97ef174a	v3.12.0.1	2020-02-06 22:50:20 -05:00
Jay D Dee	13523a12f9	v3.12.0	2020-02-05 22:50:58 -05:00
Jay D Dee	1b76cee239	v3.11.9	2020-02-04 01:31:59 -05:00
Jay D Dee	0681ca996d	v3.11.8	2020-01-30 03:47:11 -05:00
Jay D Dee	88f81fda0b	v3.11.7	2020-01-26 04:33:39 -05:00
Jay D Dee	103e6ad36c	v3.11.6	2020-01-23 00:11:08 -05:00
Jay D Dee	1a7a573675	v3.11.5	2020-01-18 15:14:27 -05:00
Jay D Dee	70089d1224	v3.11.2	2020-01-08 14:44:47 -05:00
Jay D Dee	3572cb53c4	v3.11.0	2020-01-02 23:54:08 -05:00
Jay D Dee	241bc26767	v3.10.6	2019-12-25 01:26:26 -05:00
Jay D Dee	c65b0ff7a6	v3.10.5	2019-12-21 13:19:29 -05:00
Jay D Dee	a17ff6f189	v3.10.2	2019-12-09 15:59:02 -05:00
Jay D Dee	73430b13b1	v3.10.1	2019-12-05 19:09:23 -05:00
Jay D Dee	40039386a0	v3.10.0	2019-12-03 12:26:11 -05:00
Jay D Dee	91ec6f1771	v3.9.11	2019-11-26 09:22:03 -05:00
Jay D Dee	a52c5eccf7	v3.9.10	2019-11-22 20:29:18 -05:00
Jay D Dee	86b889e1b0	v3.9.9.1	2019-10-24 14:11:26 -04:00
Jay D Dee	72330eb5a7	v3.9.9	2019-10-10 19:58:34 -04:00
Jay D Dee	789c8b70bc	v3.9.8.1	2019-10-01 14:17:36 -04:00
Jay D Dee	01550d94a2	v3.9.8	2019-09-26 22:37:26 -04:00
Jay D Dee	a042fb7612	v3.9.7	2019-08-03 10:39:54 -04:00