v3.21.2

AUTHORS

@@ -29,3 +29,10 @@ Wolf0
 Optiminer
 
 Jay D Dee
+
+xcouiz@gmail.com
+
+Cryply
+
+Colin Percival
+Alexander Peslyak

INSTALL_LINUX

@@ -0,0 +1,164 @@
+
+These instructions may be out of date, see the Wiki for the latest...
+https://github.com/JayDDee/cpuminer-opt/wiki/Compiling-from-source
+
+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.
+
+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
+beyond the scope of cpuminer-opt. Regardless compiling is trivial if you
+follow the instructions.
+
+Make sure you have the basic development packages installed.
+Here is a good start:
+
+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 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. Fedora and other distributions may have similar
+but different package names. 
+
+$ 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.
+
+znver1 and znver2 should be recognized on most recent version of GCC and
+znver3 is available with GCC 11. GCC 11 also includes rocketlake support.
+In the meantime here are some suggestions to compile with new CPUs:
+
+"-march=native" is usually the best choice, used by build.sh.
+
+"-march=znver2 -mvaes" can be used for Ryzen 5000 if znver3 is not recongized.
+
+"-mcascadelake -msha" or
+"-mcometlake -mavx512 -msha" can be used for Rocket Lake.
+
+Features can also be added individually:
+
+"-msha" adds support for HW accelerated sha256.
+
+"-mavx512" adds support for 512 bit vectors
+
+"-mvaes" add support for parallel AES
+
+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.
+
+3. Download cpuminer-opt
+------------------------
+
+Download the source code for the latest realease from the official repository.
+
+https://github.com/JayDDee/cpuminer-opt/releases
+
+Extract the source code.
+
+$ 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
+
+
+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.
+
+Building on Windows prerequisites:
+
+msys
+mingw_w64
+Visual C++ redistributable 2008 X64
+openssl
+
+Install msys and mingw_w64, only needed once.
+
+Unpack msys into C:\msys or your preferred directory.
+
+Install mingw_w64 from win-builds.
+Follow instructions, check "msys or cygwin" and "x86_64" and accept default
+existing msys instalation.
+
+Open a msys shell by double clicking on msys.bat.
+Note that msys shell uses linux syntax for file specifications, "C:\" is
+mounted at "/c/".
+
+Add mingw bin directory to PATH variable
+PATH="/c/msys/opt/windows_64/bin/:$PATH"
+
+Instalation complete, compile cpuminer-opt.
+
+Unpack cpuminer-opt source files using tar from msys shell, or using 7zip
+or similar Windows program.
+
+In msys shell cd to miner directory.
+cd /c/path/to/cpuminer-opt
+
+Run build.sh to build on Windows or execute the following commands.
+
+./autogen.sh
+CFLAGS="-O3 -march=native -Wall" ./configure --with-curl
+make
+
+Start mining
+
+cpuminer.exe -a algo -o url -u user -p password
+
+The following tips may be useful for older AMD CPUs.
+
+AMD CPUs older than Steamroller, including Athlon x2 and Phenom II x4, are
+not 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.
+
+Some users with AMD CPUs without AES_NI have reported problems compiling
+with build.sh or "-march=native". Problems have included compile errors
+and poor performance. These users are recommended to compile manually
+specifying "-march=btver1" on the configure command line.
+
+Support for even older x86_64 without AES_NI or SSE2 is not availble.
+

INSTALL_WINDOWS

@@ -0,0 +1,158 @@
+Instructions for compiling cpuminer-opt for Windows.
+
+These intructions are 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
+cpuminer-opt executable binaries for Windows.
+
+These instructions were written for Debian and Ubuntu compatible distributions
+but should work on other major distributions as well. However some of the
+package names or file paths may be different.
+
+It is assumed a Linux system is already available and running. And the user
+has enough Linux knowledge to find and install packages and follow these
+instructions.
+
+First it is a good idea to create new user specifically for cross compiling.
+It keeps all mingw stuff contained and isolated from the rest of the system.
+
+Step by step...
+
+1. Install necessary packages from the distribution's repositories.
+
+Refer to Linux compile instructions and install required packages.
+
+Additionally, 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. 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.
+
+Download the following source code packages from their respective and
+respected download locations, copy them to $HOME/usr/lib/ and uncompress them. 
+
+openssl: https://github.com/openssl/openssl/releases
+
+curl: https://github.com/curl/curl/releases
+
+gmp: https://gmplib.org/download/gmp/
+
+In most cases the latest version is ok but it's safest to download the same major and minor version as included in your distribution. The following uses versions from Ubuntu 20.04. Change version numbers as required.
+
+Run the following commands or follow the supplied instructions. Do not run "make install" unless you are using /usr/lib, which isn't recommended.
+
+Some instructions insist on running "make check". If make check fails it may still work, YMMV.
+
+You can speed up "make" by using all CPU cores available with "-j n" where n is the number of CPU threads you want to use.
+
+openssl:
+
+$ ./Configure mingw64 shared --cross-compile-prefix=x86_64-w64-mingw32-
+$ make
+
+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
+
+gmp:
+
+$ ./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.
+
+Define some local variables to point to local library.
+
+$ export LOCAL_LIB="$HOME/usr/lib"
+
+$ 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"
+
+Adjust for gcc version:
+
+$ 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 $LOCAL_LIB/curl/lib/.libs/libcurl-4.dll release/
+
+The following steps need to be done every time a new source package is
+opened.
+
+5. Download cpuminer-opt
+
+Download the latest source code package of cpumuner-opt to your desired
+location. .zip or .tar.gz, your choice.
+
+https://github.com/JayDDee/cpuminer-opt/releases
+
+Decompress and change to the cpuminer-opt directory.
+
+6. compile
+
+Create a link to the locally compiled version of gmp.h
+
+$ 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.
+
+

Makefile.am

@@ -18,19 +18,10 @@ 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 \
+  malloc-huge.c \
   algo/argon2/argon2a/argon2a.c \
   algo/argon2/argon2a/ar2/argon2.c \
   algo/argon2/argon2a/ar2/opt.c \
@@ -42,20 +33,24 @@ cpuminer_SOURCES = \
   algo/argon2/argon2d/argon2d/argon2.c \
   algo/argon2/argon2d/argon2d/core.c \
   algo/argon2/argon2d/argon2d/opt.c \
-  algo/argon2/argon2d/argon2d/thread.c \
+  algo/argon2/argon2d/argon2d/argon2d_thread.c \
   algo/argon2/argon2d/argon2d/encoding.c \
   algo/blake/sph_blake.c \
-  algo/blake/blake-hash-4way.c \
+  algo/blake/blake256-hash-4way.c \
+  algo/blake/blake512-hash-4way.c \
   algo/blake/blake-gate.c \
   algo/blake/blake.c \
   algo/blake/blake-4way.c \
   algo/blake/sph_blake2b.c \
-  algo/blake/blake2b.c \
   algo/blake/sph-blake2s.c \
   algo/blake/blake2s-hash-4way.c \
   algo/blake/blake2s.c \
   algo/blake/blake2s-gate.c \
   algo/blake/blake2s-4way.c \
+  algo/blake/blake2b-hash-4way.c \
+  algo/blake/blake2b.c \
+  algo/blake/blake2b-gate.c \
+  algo/blake/blake2b-4way.c \
   algo/blake/blakecoin-gate.c \
   algo/blake/mod_blakecoin.c \
   algo/blake/blakecoin.c \
@@ -67,33 +62,35 @@ cpuminer_SOURCES = \
   algo/blake/pentablake-4way.c \
   algo/blake/pentablake.c \
   algo/bmw/sph_bmw.c \
-  algo/bmw/bmw-hash-4way.c \
+  algo/bmw/bmw256-hash-4way.c \
+  algo/bmw/bmw512-hash-4way.c \
   algo/bmw/bmw256.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/sse2/cubehash_sse2.c\
+  algo/bmw/bmw512-gate.c \
+  algo/bmw/bmw512.c \
+  algo/bmw/bmw512-4way.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 \
@@ -109,40 +106,47 @@ 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/lyra2rev2-gate.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 \
+  algo/lyra2/lyra2rev3.c \
+  algo/lyra2/lyra2rev3-4way.c \
   algo/lyra2/lyra2re.c \
-  algo/lyra2/lyra2z-gate.c \
   algo/lyra2/lyra2z.c \
   algo/lyra2/lyra2z-4way.c \
   algo/lyra2/lyra2z330.c \
-  algo/lyra2/lyra2h-gate.c \
   algo/lyra2/lyra2h.c \
   algo/lyra2/lyra2h-4way.c \
-  algo/lyra2/allium-gate.c \
   algo/lyra2/allium-4way.c \
   algo/lyra2/allium.c \
-  algo/m7m.c \
-  algo/neoscrypt/neoscrypt.c \
+  algo/lyra2/phi2-4way.c \
+  algo/lyra2/phi2.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/pluck.c \
+  algo/panama/panama-hash-4way.c \
+  algo/panama/sph_panama.c \
   algo/quark/quark-gate.c \
   algo/quark/quark.c \
   algo/quark/quark-4way.c \
   algo/quark/anime-gate.c \
   algo/quark/anime.c \
   algo/quark/anime-4way.c \
+  algo/quark/hmq1725-gate.c \
+  algo/quark/hmq1725-4way.c \
+  algo/quark/hmq1725.c \
   algo/qubit/qubit-gate.c \
   algo/qubit/qubit.c \
   algo/qubit/qubit-2way.c \
@@ -154,21 +158,33 @@ cpuminer_SOURCES = \
   algo/ripemd/lbry-gate.c \
   algo/ripemd/lbry.c \
   algo/ripemd/lbry-4way.c \
-  algo/scrypt.c \
-  algo/scryptjane/scrypt-jane.c \
+  algo/scrypt/scrypt.c \
+  algo/scrypt/scrypt-core-4way.c \
+  algo/scrypt/neoscrypt.c \
+  algo/sha/sha256-hash.c \
   algo/sha/sph_sha2.c \
   algo/sha/sph_sha2big.c \
-  algo/sha/sha2-hash-4way.c \
+  algo/sha/sha256-hash-4way.c \
+  algo/sha/sha512-hash-4way.c \
+  algo/sha/sha256-hash-opt.c \
+  algo/sha/sha256-hash-2way-ni.c \
+  algo/sha/hmac-sha256-hash.c \
+  algo/sha/hmac-sha256-hash-4way.c \
+  algo/sha/sha256d.c \
   algo/sha/sha2.c \
+  algo/sha/sha256d-4way.c \
   algo/sha/sha256t-gate.c \
   algo/sha/sha256t-4way.c \
   algo/sha/sha256t.c \
+  algo/sha/sha256q-4way.c \
+  algo/sha/sha256q.c \
   algo/shabal/sph_shabal.c \
   algo/shabal/shabal-hash-4way.c \
   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 \
@@ -179,14 +195,17 @@ 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 \
-  algo/whirlpool/whirlpool-4way.c \
   algo/whirlpool/whirlpool.c \
   algo/whirlpool/whirlpoolx.c \
   algo/x11/x11-gate.c \
@@ -207,7 +226,6 @@ 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 \
@@ -226,7 +244,8 @@ 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 \
   algo/x14/x14.c \
   algo/x14/x14-4way.c \
@@ -240,19 +259,36 @@ cpuminer_SOURCES = \
   algo/x15/x15-gate.c \
   algo/x15/x15.c \
   algo/x15/x15-4way.c \
+  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 \
   algo/x17/xevan-gate.c \
   algo/x17/xevan.c \
   algo/x17/xevan-4way.c \
-  algo/x17/x16r-gate.c \
-  algo/x17/x16r.c \
-  algo/x17/x16r-4way.c \
-  algo/x17/hmq1725.c \
-  algo/yescrypt/yescrypt.c \
-  algo/yescrypt/sha256_Y.c \
-  algo/yescrypt/yescrypt-best.c
+  algo/x17/sonoa-gate.c \
+  algo/x17/sonoa-4way.c \
+  algo/x17/sonoa.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/yespower/yespower-gate.c \
+  algo/yespower/yespower-blake2b.c \
+  algo/yespower/crypto/hmac-blake2b.c \
+  algo/yespower/yescrypt-r8g.c \
+  algo/yespower/yespower-opt.c
 
 disable_flags =
 

README.md

@@ -7,119 +7,187 @@ All of the code is believed to be open and free. If anyone has a
 claim to any of it post your case in the cpuminer-opt Bitcoin Talk forum
 or by email.
 
+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.
+
+
+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
 
-See file RELEASE_NOTES for change log and compile instructions.
+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.
 
 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.
+32 bit CPUs are not supported.
+Other CPU architectures such as ARM, Raspberry Pi, RISC-V, Xeon Phi, etc,
+are not supported.
 
-ARM CPUs are not supported.
+Mobile CPUs like laptop computers are not recommended because they aren't
+designed for extreme heat of operating at full load for extended periods of
+time.
 
-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. 
-64 bit Windows OS is supported with mingw_w64 and msys or pre-built binaries.
+Older CPUs and ARM architecture may be supported by cpuminer-multi by TPruvot.
 
+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. 
+
+Windows 7 or newer is supported with mingw_w64 and msys or using the pre-built
+binaries. WindowsXP 64 bit is YMMV.
+
+FreeBSD is not actively tested but should work, YMMV.
 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
 --------------------
 
-                          allium       Garlicoin
-                          anime        Animecoin
-                          argon2       Argon2 coin (AR2)
-                          argon2d-crds Credits (CRDS)
-                          argon2d-dyn  Dynamic (DYN)
-                          axiom        Shabal-256 MemoHash
-                          bastion
-                          blake        Blake-256 (SFR)
-                          blakecoin    blake256r8
-                          blake2s      Blake-2 S
-                          bmw          BMW 256
-                          c11          Chaincoin
-                          cryptolight  Cryptonight-light
-                          cryptonight  cryptonote, Monero (XMR)
+                          allium        Garlicoin
+                          anime         Animecoin
+                          argon2        Argon2 coin (AR2)
+                          argon2d250    argon2d-crds, Credits (CRDS)
+                          argon2d500    argon2d-dyn,  Dynamic (DYN)
+                          argon2d4096   argon2d-uis, Unitus, (UIS)
+                          blake         Blake-256
+                          blake2b       Blake2-512
+                          blake2s       Blake2-256
+                          blakecoin     blake256r8
+                          bmw           BMW 256
+                          bmw512        BMW 512
+                          c11           
                           decred
-                          deep         Deepcoin (DCN)
-                          dmd-gr       Diamond-Groestl
-                          drop         Dropcoin
-                          fresh        Fresh
-                          groestl      Groestl coin
-                          heavy        Heavy
-                          hmq1725      Espers
-                          hodl         Hodlcoin
-                          jha          Jackpotcoin
-                          keccak       Maxcoin
-                          keccakc      Creative coin
-                          lbry         LBC, LBRY Credits
-                          luffa        Luffa
-                          lyra2h       Hppcoin
-                          lyra2re      lyra2
-                          lyra2rev2    lyra2v2, Vertcoin
-                          lyra2z       Zcoin (XZC)
-                          lyra2z330    Lyra2 330 rows, Zoin (ZOI)
-                          m7m          Magi (XMG)
-                          myr-gr       Myriad-Groestl
-                          neoscrypt    NeoScrypt(128, 2, 1)
-                          nist5        Nist5
-                          pentablake   Pentablake
-                          phi1612      phi, LUX coin
-                          pluck        Pluck:128 (Supcoin)
-                          polytimos    Ninja
-                          quark        Quark
-                          qubit        Qubit
-                          scrypt       scrypt(1024, 1, 1) (default)
-                          scrypt:N     scrypt(N, 1, 1)
-                          scryptjane:nf
-                          sha256d      Double SHA-256
-                          sha256t      Triple SHA-256, Onecoin (OC)
-                          shavite3     Shavite3
-                          skein        Skein+Sha (Skeincoin)
-                          skein2       Double Skein (Woodcoin)
-                          skunk        Signatum (SIGT)
-                          timetravel   Machinecoin (MAC)
-                          timetravel10 Bitcore
-                          tribus       Denarius (DNR)
-                          vanilla      blake256r8vnl (VCash)
-                          veltor       (VLT)
+                          deep          Deepcoin (DCN)
+                          dmd-gr        Diamond-Groestl
+                          groestl       Groestl coin
+                          hex           x16r-hex
+                          hmq1725       
+                          hodl          Hodlcoin
+                          jha           Jackpotcoin
+                          keccak        Maxcoin
+                          keccakc       Creative coin
+                          lbry          LBC, LBRY Credits
+                          lyra2h        
+                          lyra2re       lyra2
+                          lyra2rev2     lyra2v2
+                          lyra2rev3     lyrav2v3
+                          lyra2z        
+                          lyra2z330     
+                          m7m           
+                          minotaur 
+                          minotaurx
+                          myr-gr        Myriad-Groestl
+                          neoscrypt     NeoScrypt(128, 2, 1)
+                          nist5         Nist5
+                          pentablake    Pentablake
+                          phi1612       phi
+                          phi2          
+                          polytimos     Ninja
+                          power2b       MicroBitcoin (MBC)
+                          quark         Quark
+                          qubit         Qubit
+                          scrypt        scrypt(1024, 1, 1) (default)
+                          scrypt:N      scrypt(N, 1, 1)
+                          scryptn2      scrypt(1048576, 1, 1)
+                          sha256d       Double SHA-256
+                          sha256q       Quad SHA-256
+                          sha256t       Triple SHA-256
+                          sha3d         Double keccak256 (BSHA3)
+                          skein         Skein+Sha (Skeincoin)
+                          skein2        Double Skein (Woodcoin)
+                          skunk         Signatum (SIGT)
+                          sonoa         Sono
+                          timetravel    Machinecoin (MAC)
+                          timetravel10  Bitcore
+                          tribus        Denarius (DNR)
+                          vanilla       blake256r8vnl (VCash)
+                          veltor        (VLT)
+                          verthash      Vertcoin
                           whirlpool
                           whirlpoolx
-                          x11          Dash
-                          x11evo       Revolvercoin
-                          x11gost      sib (SibCoin)
-                          x12          Galaxie Cash (GCH)
-                          x13          X13
-                          x13sm3       hsr (Hshare)
-                          x14          X14
-                          x15          X15
-                          x16r         Ravencoin (RVN)
-                          x16s         pigeoncoin (PGN)
+                          x11           Dash
+                          x11evo        Revolvercoin
+                          x11gost       sib (SibCoin)
+                          x12           
+                          x13           
+                          x13bcd        bcd
+                          x13sm3        hsr (Hshare)
+                          x14           
+                          x15           
+                          x16r          
+                          x16rv2        
+                          x16rt         
+                          x16rt-veil    veil
+                          x16s          
                           x17
-                          xevan        Bitsend (BSD)
-                          yescrypt     Globalboost-Y (BSTY)
-                          yescryptr8   BitZeny (ZNY)
-                          yescryptr16  Yenten (YTN)
-                          yescryptr32  WAVI
-                          zr5          Ziftr
+                          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.
 
 AMD CPUs older than Piledriver, including Athlon x2 and Phenom II x4, are not
@@ -135,14 +203,17 @@ 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
 ---------
@@ -150,10 +221,6 @@ 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!
 

README.txt

@@ -1,36 +1,80 @@
 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.
+cpuminer-opt is open source and free of any fees. Many forks exist that are
+closed source and contain usage fees. support open source free software.
+
+This package is officially avalaible 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
+command prompt. There is no GUI and no mouse support.
+
+New users are encouraged to consult the cpuminer-opt Wiki for detailed
+information on usage:
+
+https://github.com/JayDDee/cpuminer-opt/wiki
+
+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.
 
 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 all the available features.
+optimum speed using the best available features.
 
-Architecture names and compile options used are only provided for Intel
-Core series. Pentium and Celeron often have fewer features.
+Architecture names and compile options used are only provided for 
+mainstream desktop CPUs. Budget CPUs like Pentium and Celeron are often
+missing some features. Check your CPU.
 
-AMD CPUs older than Piledriver, including Athlon x2 and Phenom II x4, are not
-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.
+Support for AMD CPUs older than Ryzen is incomplete and without specific 
+recommendations. Find the best fit. CPUs older than Piledriver, including
+Athlon x2 and Phenom II x4, are not supported by cpuminer-opt due to an
+incompatible implementation of SSE2 on these CPUs. 
 
-Exe name                Compile flags              Arch name
+More information for Intel and AMD CPU architectures and their features
+can be found on Wikipedia.
+
+https://en.wikipedia.org/wiki/List_of_Intel_CPU_microarchitectures
+
+https://en.wikipedia.org/wiki/List_of_AMD_CPU_microarchitectures
+
+File name                      Architecture name
+
+cpuminer-sse2.exe              Core2, Nehalem, generic x86_64 with SSE2   
+cpuminer-aes-sse42.exe         Westmere
+cpuminer-avx.exe               Sandybridge, Ivybridge
+cpuminer-avx2.exe              Haswell, Skylake, Kabylake, Coffeelake, Cometlake
+cpuminer-avx2-sha.exe          AMD Zen1, Zen2
+cpuminer-avx2-sha-vaes.exe     Intel Alderlake*, AMD Zen3
+cpuminer-avx512.exe            Intel HEDT Skylake-X, Cascadelake
+cpuminer-avx512-sha-vaes.exe   AMD Zen4, Intel Rocketlake, Icelake
+
+* Alderlake is a hybrid architecture with a mix of E-cores & P-cores. Although
+  the P-cores can support AVX512 the E-cores can't so Intel decided to disable
+  AVX512 on the the P-cores.
+
+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
+
+Some included DLL files may already be installed on the system by Windows or
+third party packages. They often will work and may be used instead of the
+included version of the files.
 
-cpuminer-sse2.exe      "-msse2"                    Core2, Nehalem   
-cpuminer-aes-sse42.exe "-maes -msse4.2"            Westmere
-cpuminer-aes-avx.exe   "-march=corei7-avx"         Sandybridge, Ivybridge
-cpuminer-avx2.exe      "-march=core-avx2"          Haswell...
-cpuminer-avx2-sha.exe  "-march=core-avx2 -msha"    Ryzen
 
 If you like this software feel free to donate:
 
 BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
-ETH: 0x72122edabcae9d3f57eab0729305a425f6fef6d0
-LTC: LdUwoHJnux9r9EKqFWNvAi45kQompHk6e8
-BCH: 1QKYkB6atn4P7RFozyziAXLEnurwnUM1cQ
-BTG: GVUyECtRHeC5D58z9F3nGGfVQndwnsPnHQ
 
 

aclocal.m4

@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.16.5 -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2021 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,13 +14,13 @@
 m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
-[m4_warning([this file was generated for autoconf 2.69.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.71],,
+[m4_warning([this file was generated for autoconf 2.71.
 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) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002-2021 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='1.14'
+[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], [1.14.1], [],
+m4_if([$1], [1.16.5], [],
       [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([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.16.5])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) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 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) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -123,15 +123,14 @@ _AM_IF_OPTION([no-dependencies],, [_AM_DEPENDENCIES([CCAS])])dnl
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -162,7 +161,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -353,13 +352,12 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2021 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],
@@ -367,49 +365,43 @@ 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.  If GNU make was not used, consider
+    re-running the configure script with MAKE="gmake" (or whatever is
+    necessary).  You can also 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
 
@@ -418,18 +410,17 @@ 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) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -457,6 +448,10 @@ m4_defn([AC_PROG_CC])
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
 [AC_PREREQ([2.65])dnl
+m4_ifdef([_$0_ALREADY_INIT],
+  [m4_fatal([$0 expanded multiple times
+]m4_defn([_$0_ALREADY_INIT]))],
+  [m4_define([_$0_ALREADY_INIT], m4_expansion_stack)])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -493,7 +488,7 @@ m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
 m4_if(
-  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
+  m4_ifset([AC_PACKAGE_NAME], [ok]):m4_ifset([AC_PACKAGE_VERSION], [ok]),
   [ok:ok],,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
@@ -516,11 +511,11 @@ AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
 AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # For better backward compatibility.  To be removed once Automake 1.9.x
 # dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+# <https://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <https://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -545,6 +540,20 @@ AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
 		  [m4_define([AC_PROG_OBJCXX],
 			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
 ])
+# Variables for tags utilities; see am/tags.am
+if test -z "$CTAGS"; then
+  CTAGS=ctags
+fi
+AC_SUBST([CTAGS])
+if test -z "$ETAGS"; then
+  ETAGS=etags
+fi
+AC_SUBST([ETAGS])
+if test -z "$CSCOPE"; then
+  CSCOPE=cscope
+fi
+AC_SUBST([CSCOPE])
+
 AC_REQUIRE([AM_SILENT_RULES])dnl
 dnl The testsuite driver may need to know about EXEEXT, so add the
 dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
@@ -584,7 +593,7 @@ 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
@@ -593,7 +602,11 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi])
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -622,7 +635,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -633,7 +646,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -643,7 +656,7 @@ if test x"${install_sh}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -665,7 +678,7 @@ AC_SUBST([am__leading_dot])])
 # Add --enable-maintainer-mode option to configure.         -*- Autoconf -*-
 # From Jim Meyering
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -700,7 +713,7 @@ AC_MSG_CHECKING([whether to enable maintainer-specific portions of Makefiles])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -708,49 +721,42 @@ 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) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -771,12 +777,7 @@ AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
 if test x"${MISSING+set}" != xset; then
-  case $am_aux_dir in
-  *\ * | *\	*)
-    MISSING="\${SHELL} \"$am_aux_dir/missing\"" ;;
-  *)
-    MISSING="\${SHELL} $am_aux_dir/missing" ;;
-  esac
+  MISSING="\${SHELL} '$am_aux_dir/missing'"
 fi
 # Use eval to expand $SHELL
 if eval "$MISSING --is-lightweight"; then
@@ -789,7 +790,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -818,7 +819,7 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -865,7 +866,7 @@ AC_LANG_POP([C])])
 # For backward compatibility.
 AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -884,7 +885,7 @@ AC_DEFUN([AM_RUN_LOG],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -965,7 +966,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1025,7 +1026,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1053,7 +1054,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -1072,7 +1073,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004-2021 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,

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.
@@ -89,39 +87,178 @@ 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. Not to be used when
+// prehashing can be done or when byte swapping the data can be avoided.
+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 );
+   // overwrite byte swapped nonce with original byte order for proper
+   // incrementing. The nonce only needs to byte swapped if it is to be
+   // sumbitted.
+   *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->display_extra_data      = (void*)&do_nothing;
-   gate->wait_for_diff           = (void*)&std_wait_for_diff;
-   gate->get_max64               = (void*)&get_max64_0x1fffffLL;
+   gate->work_decode             = (void*)&std_le_work_decode;
+   gate->decode_extra_data       = (void*)&do_nothing;
    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->work_decode             = (void*)&std_le_work_decode;
    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;
@@ -131,12 +268,11 @@ 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;
    gate->nbits_index             = STD_NBITS_INDEX;
    gate->nonce_index             = STD_NONCE_INDEX;
-   gate->work_data_size          = STD_WORK_DATA_SIZE;
    gate->work_cmp_size           = STD_WORK_CMP_SIZE;
 }
 
@@ -144,103 +280,118 @@ 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 )
 {
-   if ( NULL == gate )
-   {
-     applog(LOG_ERR,"FAIL: algo_gate registration failed, NULL gate\n");
-     return false;
-   }
+  bool rc = false;
 
-   init_algo_gate( gate );
-
-   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_ARGON2DCRDS:  register_argon2d_crds_algo( gate ); break;
-     case ALGO_ARGON2DDYN:   register_argon2d_dyn_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_BLAKECOIN:    register_blakecoin_algo   ( gate ); break;
-//     case ALGO_BLAKE2B:      register_blake2b_algo    ( gate ); break;
-     case ALGO_BLAKE2S:      register_blake2s_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_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_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_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_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_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_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_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_X16S:         register_x16s_algo        ( gate ); break;
-     case ALGO_X17:          register_x17_algo         ( gate ); break;
-     case ALGO_XEVAN:        register_xevan_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_ZR5:          register_zr5_algo         ( gate ); break;
-    default:
-        applog(LOG_ERR,"FAIL: algo_gate registration failed, unknown algo %s.\n", algo_names[opt_algo] );
-        return false;
-   } // switch
-
-  // ensure required functions were defined.
-  if (  gate->scanhash == (void*)&null_scanhash )
+  if ( NULL == gate )
   {
-    applog(LOG_ERR, "FAIL: Required algo_gate functions undefined\n");
+    applog(LOG_ERR,"FAIL: algo_gate registration failed, NULL gate\n");
+    return false;
+  }
+
+  init_algo_gate( gate );
+
+  switch ( algo )
+  {
+    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_MINOTAURX:    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_algo      ( gate ); break;
+    case ALGO_YESCRYPTR8:   rc = register_yescryptr8_algo    ( gate ); break;
+    case ALGO_YESCRYPTR8G:  rc = register_yescryptr8g_algo   ( gate ); break;
+    case ALGO_YESCRYPTR16:  rc = register_yescryptr16_algo   ( gate ); break;
+    case ALGO_YESCRYPTR32:  rc = 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,"BUG: unregistered algorithm %s.\n", algo_names[opt_algo] );
+      return false;
+  } // switch
+
+  if ( !rc )
+  {
+    applog(LOG_ERR, "FAIL: %s algorithm failed to initialize\n", algo_names[opt_algo] );
     return false;
   }
   return true;
@@ -249,26 +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 )
-{
-  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;   
@@ -288,42 +419,40 @@ void exec_hash_function( int algo, void *output, const void *pdata )
 const char* const algo_alias_map[][2] =
 {
 //   alias                proper
-  { "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"    },
-  { "lyra2zoin",         "lyra2z330"    },
-  { "myrgr",             "myr-gr"       },
-  { "myriad",            "myr-gr"       },
-  { "neo",               "neoscrypt"    },
-  { "phi",               "phi1612"      },
-//  { "sia",               "blake2b"      },
-  { "sib",               "x11gost"      },
-  { "timetravel8",       "timetravel"   },
-  { "ziftr",             "zr5"          },
-  { "yenten",            "yescryptr16"  },
-  { "yescryptr8k",       "yescrypt"     },
-  { "zcoin",             "lyra2z"       },
-  { "zoin",              "lyra2z330"    },
-  { 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 name.
-// No validation of the algo or alias is done, It is the responsinility of the
-// calling function to validate the algo after return.
+// if arg is a valid alias for a known algo it is updated with the proper
+// name. No validation of the algo or alias is done, It is the responsinility
+// of the calling function to validate the algo after return.
 void get_algo_alias( char** algo_or_alias )
 {
   int i;
@@ -331,8 +460,11 @@ 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;
     }
 }
 
+#undef ALIAS
+#undef PROPER
+

algo-gate-api.h

@@ -1,7 +1,11 @@
+#ifndef __ALGO_GATE_API_H__
+#define __ALGO_GATE_API_H__ 1
+
 #include <stdlib.h>
 #include <stdbool.h>
 #include <stdint.h>
 #include "miner.h"
+#include "simd-utils.h"
 
 /////////////////////////////
 ////
@@ -34,7 +38,7 @@
 //    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
@@ -47,10 +51,10 @@
 //    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
@@ -74,7 +78,7 @@
 
 // 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:
@@ -84,13 +88,16 @@
 
 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 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)
+
 
 // return set containing all elements from sets a & b
 inline set_t set_union ( set_t a, set_t b ) { return a | b; }
@@ -106,45 +113,68 @@ inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }
 
 typedef struct
 {
-// mandatory functions, must be overwritten
-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 ( *display_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;
 int  nbits_index;
 int  nonce_index;            // use with caution, see warning below
-int  work_data_size;
 int  work_cmp_size;
-
 } algo_gate_t;
 
 extern algo_gate_t algo_gate;
@@ -176,102 +206,119 @@ 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();
 
+// 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.
+
+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 );
 
+int std_get_work_data_size();
+
 // Gate admin functions
 
 // Called from main to initialize all gate functions and algo-specific data
 // 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 );
 
+#endif

algo/argon2/argon2a/ar2/sj/scrypt-jane-portable-x86.h

@@ -344,7 +344,7 @@ static size_t
 detect_cpu(void) {
 	//union { uint8_t s[12]; uint32_t i[3]; } vendor_string;
 	//cpu_vendors_x86 vendor = cpu_nobody;
-	x86_regs regs;
+	x86_regs regs; regs.eax = regs.ebx = regs.ecx = 0;
 	uint32_t max_level, max_ext_level;
 	size_t cpu_flags = 0;
 #if defined(X86ASM_AVX) || defined(X86_64ASM_AVX)
@@ -460,4 +460,4 @@ get_top_cpuflag_desc(size_t flag) {
 	#endif
 #endif
 
-#endif /* defined(CPU_X86) || defined(CPU_X86_64) */
+#endif /* defined(CPU_X86) || defined(CPU_X86_64) */

algo/argon2/argon2a/ar2/sj/scrypt-jane-romix-basic.h

@@ -4,11 +4,12 @@ typedef void (FASTCALL *scrypt_ROMixfn)(scrypt_mix_word_t *X/*[chunkWords]*/, sc
 #endif
 
 /* romix pre/post nop function */
+/*
 static void asm_calling_convention
 scrypt_romix_nop(scrypt_mix_word_t *blocks, size_t nblocks) {
 	(void)blocks; (void)nblocks;
 }
-
+*/
 /* romix pre/post endian conversion function */
 static void asm_calling_convention
 scrypt_romix_convert_endian(scrypt_mix_word_t *blocks, size_t nblocks) {

algo/argon2/argon2a/argon2a.c

@@ -42,12 +42,14 @@ void argon2hash(void *output, const void *input)
 		(unsigned char *)output);
 }
 
-int scanhash_argon2(int thr_id, struct work* work, uint32_t max_nonce, uint64_t *hashes_done)
+int scanhash_argon2( 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) 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];
@@ -60,9 +62,7 @@ int scanhash_argon2(int thr_id, struct work* work, uint32_t max_nonce, uint64_t
 		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);
@@ -72,19 +72,14 @@ int scanhash_argon2(int thr_id, struct work* work, uint32_t max_nonce, uint64_t
 	return 0;
 }
 
-int64_t argon2_get_max64 ()
-{
-  return 0x1ffLL;
-}
-
 bool register_argon2_algo( algo_gate_t* gate )
 {
-  gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT;
+  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;
 };
 

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)
@@ -28,49 +29,48 @@ void argon2d_crds_hash( void *output, const void *input )
 	context.lanes = 4;    // Degree of Parallelism
 	context.threads = 1;  // Threads
 	context.t_cost = 1;   // Iterations
+        context.version = ARGON2_VERSION_10;
 
 	argon2_ctx( &context, Argon2_d );
 }
 
-int scanhash_argon2d_crds( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+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) hash[8];
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
+   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];
+   uint32_t nonce = first_nonce;
 
-        const uint32_t first_nonce = pdata[19];
-        const uint32_t Htarg = ptarget[7];
+   swab32_array( edata, pdata, 20 );
 
-        uint32_t nonce = first_nonce;
+   do {
+      be32enc(&edata[19], nonce);
+      argon2d_crds_hash( hash, edata );
+      if ( hash[7] <= Htarg && fulltest( hash, ptarget ) && !opt_benchmark )
+      {
+          pdata[19] = nonce;
+          submit_solution( work, hash, mythr );
+      }
+      nonce++;
+   } while (nonce < max_nonce && !work_restart[thr_id].restart);
 
-        swab32_array( endiandata, pdata, 20 );
-
-        do {
-                be32enc(&endiandata[19], nonce);
-                argon2d_crds_hash( 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;
-                }
-                nonce++;
-        } while (nonce < max_nonce && !work_restart[thr_id].restart);
-
-        pdata[19] = nonce;
-        *hashes_done = pdata[19] - first_nonce + 1;
-        return 0;
+   pdata[19] = nonce;
+   *hashes_done = pdata[19] - first_nonce + 1;
+   return 0;
 }
 
 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 | AES_OPT | AVX2_OPT;
+        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
+        return true;
 }
 
 // Dynamic
@@ -96,48 +96,94 @@ void argon2d_dyn_hash( void *output, const void *input )
     context.lanes = 8;     // Degree of Parallelism
     context.threads = 1;   // Threads
     context.t_cost = 2;    // Iterations
+    context.version = ARGON2_VERSION_10;
 
     argon2_ctx( &context, Argon2_d );
 }
 
-int scanhash_argon2d_dyn( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+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) hash[8];
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
+   uint32_t _ALIGN(64) edata[20];
+   uint32_t _ALIGN(64) hash[8];
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   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;
 
-        const uint32_t first_nonce = pdata[19];
-        const uint32_t Htarg = ptarget[7];
+   mm128_bswap32_80( edata, pdata );
+   do
+   {
+      edata[19] = nonce;
+      argon2d_dyn_hash( hash, edata );
+      if ( unlikely( valid_hash( (uint64_t*)hash, (uint64_t*)ptarget )
+           && !bench ) )
+      {
+          pdata[19] = bswap_32( nonce );;
+          submit_solution( work, hash, mythr );
+      }
+      nonce++;
+  } while ( likely( nonce < last_nonce && !work_restart[thr_id].restart ) );
 
-        uint32_t nonce = first_nonce;
-
-        swab32_array( endiandata, pdata, 20 );
-
-        do {
-                be32enc(&endiandata[19], nonce);
-                argon2d_dyn_hash( 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;
-                }
-                nonce++;
-        } while (nonce < max_nonce && !work_restart[thr_id].restart);
-
-        pdata[19] = nonce;
-        *hashes_done = pdata[19] - first_nonce + 1;
-        return 0;
+   pdata[19] = nonce;
+   *hashes_done = pdata[19] - first_nonce;
+   return 0;
 }
 
 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 | AES_OPT | AVX2_OPT;
+        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
+        return true;
+}
+
+// Unitus
+
+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) edata[20];
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = (const uint32_t)max_nonce;
+   uint32_t n = first_nonce;
+   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;
+
+   mm128_bswap32_80( edata, pdata );
+
+   do {
+      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 ) )
+      {
+         be32enc( &pdata[19], n );
+         submit_solution( work, vhash, mythr );
+      }
+      n++;
+   } while ( likely( n < last_nonce && !work_restart[thr_id].restart ) );
+
+   *hashes_done = n - first_nonce;
+   pdata[19] = n;
+   return 0;
+}
+
+bool register_argon2d4096_algo( algo_gate_t* gate )
+{
+        gate->scanhash = (void*)&scanhash_argon2d4096;
+        gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
+        opt_target_factor = 65536.0;
+        return true;
 }
 

algo/argon2/argon2d/argon2d-gate.h

@@ -4,22 +4,28 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-// Credits
+// Credits: version = 0x10, m_cost = 250.
 bool register_argon2d_crds_algo( algo_gate_t* gate );
 
 void argon2d_crds_hash( void *state, const void *input );
 
-int scanhash_argon2d_crds( int thr_id, struct work *work, uint32_t max_nonce,
-                    uint64_t *hashes_done );
+int scanhash_argon2d_crds( struct work *work, uint32_t max_nonce,
+                    uint64_t *hashes_done, struct thr_info *mythr );
 
-// Dynamic
+// Dynamic: version = 0x10, m_cost = 500.
 bool register_argon2d_dyn_algo( algo_gate_t* gate );
 
 void argon2d_dyn_hash( void *state, const void *input );
 
-int scanhash_argon2d_dyn( int thr_id, struct work *work, uint32_t max_nonce,
-                    uint64_t *hashes_done );
+int scanhash_argon2d_dyn( struct work *work, uint32_t max_nonce,
+                    uint64_t *hashes_done, struct thr_info *mythr );
 
 
+// Unitus: version = 0x13, m_cost = 4096.
+bool register_argon2d4096_algo( algo_gate_t* gate );
+
+int scanhash_argon2d4096( struct work *work, uint32_t max_nonce,
+                    uint64_t *hashes_done, struct thr_info *mythr );
+
 #endif
 

algo/argon2/argon2d/argon2d/argon2.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -19,10 +19,6 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#ifdef _WIN32
-#include <malloc.h>
-#endif
-
 #include "argon2.h"
 #include "encoding.h"
 #include "core.h"
@@ -31,6 +27,10 @@ const char *argon2_type2string(argon2_type type, int uppercase) {
     switch (type) {
         case Argon2_d:
             return uppercase ? "Argon2d" : "argon2d";
+        case Argon2_i:
+            return uppercase ? "Argon2i" : "argon2i";
+        case Argon2_id:
+            return uppercase ? "Argon2id" : "argon2id";
     }
 
     return NULL;
@@ -46,7 +46,7 @@ int argon2_ctx(argon2_context *context, argon2_type type) {
         return result;
     }
 
-    if (Argon2_d != type) {
+    if (Argon2_d != type && Argon2_i != type && Argon2_id != type) {
         return ARGON2_INCORRECT_TYPE;
     }
 
@@ -62,18 +62,18 @@ int argon2_ctx(argon2_context *context, argon2_type type) {
     /* Ensure that all segments have equal length */
     memory_blocks = segment_length * (context->lanes * ARGON2_SYNC_POINTS);
 
+    instance.version = context->version;
     instance.memory = NULL;
     instance.passes = context->t_cost;
     instance.memory_blocks = memory_blocks;
     instance.segment_length = segment_length;
     instance.lane_length = segment_length * ARGON2_SYNC_POINTS;
     instance.lanes = context->lanes;
-    instance.limit = 1;
     instance.threads = context->threads;
     instance.type = type;
 
-    if (instance.threads > instance.limit) {
-        instance.threads = instance.limit;
+    if (instance.threads > instance.lanes) {
+        instance.threads = instance.lanes;
     }
 
     /* 3. Initialization: Hashing inputs, allocating memory, filling first
@@ -101,7 +101,8 @@ int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
                 const uint32_t parallelism, const void *pwd,
                 const size_t pwdlen, const void *salt, const size_t saltlen,
                 void *hash, const size_t hashlen, char *encoded,
-                const size_t encodedlen, argon2_type type){
+                const size_t encodedlen, argon2_type type,
+                const uint32_t version){
 
     argon2_context context;
     int result;
@@ -145,6 +146,7 @@ int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
     context.allocate_cbk = NULL;
     context.free_cbk = NULL;
     context.flags = ARGON2_DEFAULT_FLAGS;
+    context.version = version;
 
     result = argon2_ctx(&context, type);
 
@@ -174,23 +176,69 @@ int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
     return ARGON2_OK;
 }
 
+int argon2i_hash_encoded(const uint32_t t_cost, const uint32_t m_cost,
+                         const uint32_t parallelism, const void *pwd,
+                         const size_t pwdlen, const void *salt,
+                         const size_t saltlen, const size_t hashlen,
+                         char *encoded, const size_t encodedlen,
+                         const uint32_t version) {
+
+    return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
+                       NULL, hashlen, encoded, encodedlen, Argon2_i,
+                       version );
+}
+
+int argon2i_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
+                     const uint32_t parallelism, const void *pwd,
+                     const size_t pwdlen, const void *salt,
+                     const size_t saltlen, void *hash, const size_t hashlen,
+                     const uint32_t version ) {
+
+    return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
+                       hash, hashlen, NULL, 0, Argon2_i, version );
+}
+
 int argon2d_hash_encoded(const uint32_t t_cost, const uint32_t m_cost,
                          const uint32_t parallelism, const void *pwd,
                          const size_t pwdlen, const void *salt,
                          const size_t saltlen, const size_t hashlen,
-                         char *encoded, const size_t encodedlen) {
+                         char *encoded, const size_t encodedlen,
+                         const uint32_t version ) {
 
     return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
-                       NULL, hashlen, encoded, encodedlen, Argon2_d);
+                       NULL, hashlen, encoded, encodedlen, Argon2_d,
+                       version );
 }
 
 int argon2d_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
                      const uint32_t parallelism, const void *pwd,
                      const size_t pwdlen, const void *salt,
-                     const size_t saltlen, void *hash, const size_t hashlen) {
+                     const size_t saltlen, void *hash, const size_t hashlen,
+                     const uint32_t version ) {
 
     return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
-                       hash, hashlen, NULL, 0, Argon2_d);
+                       hash, hashlen, NULL, 0, Argon2_d, version );
+}
+
+int argon2id_hash_encoded(const uint32_t t_cost, const uint32_t m_cost,
+                          const uint32_t parallelism, const void *pwd,
+                          const size_t pwdlen, const void *salt,
+                          const size_t saltlen, const size_t hashlen,
+                          char *encoded, const size_t encodedlen,
+                          const uint32_t version ) {
+
+    return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
+                       NULL, hashlen, encoded, encodedlen, Argon2_id,
+                       version);
+}
+
+int argon2id_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
+                      const uint32_t parallelism, const void *pwd,
+                      const size_t pwdlen, const void *salt,
+                      const size_t saltlen, void *hash, const size_t hashlen,
+                      const uint32_t version ) {
+    return argon2_hash(t_cost, m_cost, parallelism, pwd, pwdlen, salt, saltlen,
+                       hash, hashlen, NULL, 0, Argon2_id, version );
 }
 
 static int argon2_compare(const uint8_t *b1, const uint8_t *b2, size_t len) {
@@ -269,15 +317,33 @@ fail:
     return ret;
 }
 
+int argon2i_verify(const char *encoded, const void *pwd, const size_t pwdlen) {
+
+    return argon2_verify(encoded, pwd, pwdlen, Argon2_i);
+}
+
 int argon2d_verify(const char *encoded, const void *pwd, const size_t pwdlen) {
 
     return argon2_verify(encoded, pwd, pwdlen, Argon2_d);
 }
 
+int argon2id_verify(const char *encoded, const void *pwd, const size_t pwdlen) {
+
+    return argon2_verify(encoded, pwd, pwdlen, Argon2_id);
+}
+
 int argon2d_ctx(argon2_context *context) {
     return argon2_ctx(context, Argon2_d);
 }
 
+int argon2i_ctx(argon2_context *context) {
+    return argon2_ctx(context, Argon2_i);
+}
+
+int argon2id_ctx(argon2_context *context) {
+    return argon2_ctx(context, Argon2_id);
+}
+
 int argon2_verify_ctx(argon2_context *context, const char *hash,
                       argon2_type type) {
     int ret = argon2_ctx(context, type);
@@ -296,6 +362,14 @@ int argon2d_verify_ctx(argon2_context *context, const char *hash) {
     return argon2_verify_ctx(context, hash, Argon2_d);
 }
 
+int argon2i_verify_ctx(argon2_context *context, const char *hash) {
+    return argon2_verify_ctx(context, hash, Argon2_i);
+}
+
+int argon2id_verify_ctx(argon2_context *context, const char *hash) {
+    return argon2_verify_ctx(context, hash, Argon2_id);
+}
+
 const char *argon2_error_message(int error_code) {
     switch (error_code) {
     case ARGON2_OK:
@@ -374,307 +448,11 @@ const char *argon2_error_message(int error_code) {
         return "Unknown error code";
     }
 }
-
+/*
 size_t argon2_encodedlen(uint32_t t_cost, uint32_t m_cost, uint32_t parallelism,
                          uint32_t saltlen, uint32_t hashlen, argon2_type type) {
   return strlen("$$v=$m=,t=,p=$$") + strlen(argon2_type2string(type, 0)) +
          numlen(t_cost) + numlen(m_cost) + numlen(parallelism) +
-         b64len(saltlen) + b64len(hashlen);
+         b64len(saltlen) + b64len(hashlen) + numlen(ARGON2_VERSION_NUMBER) + 1;
 }
-
-#ifdef __AVX2__
-
-///////////////////////////
-// Wolf's Additions
-///////////////////////////
-
-#include <stdbool.h>
-#include <pthread.h>
-#include <x86intrin.h>
-#include "../blake2/blake2.h"
-
-typedef struct _Argon2d_Block
-{
-	union
-	{
-		uint64_t data[1024 / 8] __attribute__((aligned(32)));
-		__m128i dqwords[1024 / 16] __attribute__((aligned(32)));
-		__m256i qqwords[1024 / 32] __attribute__((aligned(32)));
-	};
-} Argon2d_Block;
-
-typedef struct _Argon2ThreadData
-{
-	Argon2d_Block *Matrix;
-	uint32_t slice;
-	uint32_t lane;
-} Argon2ThreadData;
-
-#define SEGMENT_LENGTH			(250U / (4U * 4U))		// memory_blocks / (context->lanes * ARGON2_SYNC_POINTS);
-#define LANE_LENGTH				(SEGMENT_LENGTH * 4U)	// segment_length * ARGON2_SYNC_POINTS;
-#define CONCURRENT_THREADS		4
-
-static const uint64_t blake2b_IV[8] =
-{
-	0x6A09E667F3BCC908ULL, 0xBB67AE8584CAA73BULL,
-	0x3C6EF372FE94F82BULL, 0xA54FF53A5F1D36F1ULL,
-	0x510E527FADE682D1ULL, 0x9B05688C2B3E6C1FULL,
-	0x1F83D9ABFB41BD6BULL, 0x5BE0CD19137E2179ULL
-};
-
-static const unsigned int blake2b_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},
-};
-
-#define ROTL64(x, y)		(((x) << (y)) | ((x) >> (64 - (y))))
-
-#define G(r, i, a, b, c, d)                                                    \
-    do {                                                                       \
-        a = a + b + m[blake2b_sigma[r][2 * i + 0]];                            \
-        d = ROTL64(d ^ a, 32);                                                 \
-        c = c + d;                                                             \
-        b = ROTL64(b ^ c, 40);                                                 \
-        a = a + b + m[blake2b_sigma[r][2 * i + 1]];                            \
-        d = ROTL64(d ^ a, 48);                                                 \
-        c = c + d;                                                             \
-        b = ROTL64(b ^ c, 1);                                                 \
-    } while ((void)0, 0)
-
-#define ROUND(r)                                                               \
-    do {                                                                       \
-        G(r, 0, v[0], v[4], v[8], v[12]);                                      \
-        G(r, 1, v[1], v[5], v[9], v[13]);                                      \
-        G(r, 2, v[2], v[6], v[10], v[14]);                                     \
-        G(r, 3, v[3], v[7], v[11], v[15]);                                     \
-        G(r, 4, v[0], v[5], v[10], v[15]);                                     \
-        G(r, 5, v[1], v[6], v[11], v[12]);                                     \
-        G(r, 6, v[2], v[7], v[8], v[13]);                                      \
-        G(r, 7, v[3], v[4], v[9], v[14]);                                      \
-    } while ((void)0, 0)
-
-void CompressBlock(uint64_t *h, const uint64_t *m, uint64_t t, uint64_t f)
-{
-	uint64_t v[16];
-	
-	int i;
-	for(i = 0; i < 8; ++i) v[i] = h[i];
-	
-	for(i = 8; i < 16; ++i) v[i] = blake2b_IV[i - 8];
-	
-	v[12] ^= t;
-	v[14] ^= f;
-	
-	int r;
-	for(r = 0; r < 12; ++r)
-	{
-		ROUND(r);
-	}
-	
-	for(i = 0; i < 8; ++i) h[i] ^= v[i] ^ v[i + 8];
-}
-
-void Argon2dInitHash(void *HashOut, void *Input)
-{
-	blake2b_state BlakeHash;
-	uint32_t InBuf[64];							// Is only 50 uint32_t, but need more space for Blake2B
-	
-	memset(InBuf, 0x00, 200);
-	
-	InBuf[0] = 4UL;								// Lanes
-	InBuf[1] = 32UL;								// Output Length
-	InBuf[2] = 250UL;							// Memory Cost
-	InBuf[3] = 1UL;								// Time Cost
-	InBuf[4] = 16UL;								// Argon2 Version Number
-	InBuf[5] = 0UL;								// Type
-	InBuf[6] = 80UL;								// Password Length
-	
-	memcpy(InBuf + 7, Input, 80);				// Password
-	
-	InBuf[27] = 80UL;							// Salt Length
-	
-	memcpy(InBuf + 28, Input, 80);				// Salt
-	
-	InBuf[48] = 0UL;								// Secret Length
-	InBuf[49] = 0UL;								// Associated Data Length
-	
-	int i;
-	for(i = 50; i < 64; ++i) InBuf[i] = 0UL;
-		
-	uint64_t H[8];
-	
-	for(i = 0; i < 8; ++i) H[i] = blake2b_IV[i];
-	
-	H[0] ^= 0x0000000001010040;
-	
-	CompressBlock(H, (uint64_t *)InBuf, 128ULL, 0ULL);
-	CompressBlock(H, (uint64_t *)(InBuf + 32), 200ULL, 0xFFFFFFFFFFFFFFFFULL);
-	
-	memcpy(HashOut, H, 64U);
-}
-
-void Argon2dFillFirstBlocks(Argon2d_Block *Matrix, void *InitHash)
-{
-	uint32_t lane;
-	for(lane = 0; lane < 4; ++lane)
-	{
-		((uint32_t *)InitHash)[16] = 0;
-		((uint32_t *)InitHash)[17] = lane;
-		blake2b_long(Matrix[lane * LANE_LENGTH].data, 1024, InitHash, 72);
-		((uint32_t *)InitHash)[16] |= 1;
-		blake2b_long(Matrix[lane * LANE_LENGTH + 1].data, 1024, InitHash, 72);
-	}
-}
-
-#include "../blake2/blamka-round-opt.h"
-
-void Argon2dFillSingleBlock(Argon2d_Block *State, Argon2d_Block *RefBlock, Argon2d_Block *NextBlock)
-{	
-	__m256i XY[32];
-	
-	int i;
-	for(i = 0; i < 32; ++i)
-		XY[i] = State->qqwords[i] = _mm256_xor_si256(State->qqwords[i], RefBlock->qqwords[i]);
-	
-	for(i = 0; i < 8; ++i)
-	{
-		BLAKE2_ROUND(	State->dqwords[8 * i + 0], State->dqwords[8 * i + 1], State->dqwords[8 * i + 2], State->dqwords[8 * i + 3],
-						State->dqwords[8 * i + 4], State->dqwords[8 * i + 5], State->dqwords[8 * i + 6], State->dqwords[8 * i + 7]);
-	}
-	
-	for(i = 0; i < 8; ++i)
-	{
-		BLAKE2_ROUND(	State->dqwords[8 * 0 + i], State->dqwords[8 * 1 + i], State->dqwords[8 * 2 + i], State->dqwords[8 * 3 + i],
-						State->dqwords[8 * 4 + i], State->dqwords[8 * 5 + i], State->dqwords[8 * 6 + i], State->dqwords[8 * 7 + i]);
-	}
-	
-	for(i = 0; i < 32; ++i)
-	{
-		State->qqwords[i] = _mm256_xor_si256(State->qqwords[i], XY[i]);
-		_mm256_store_si256(NextBlock->qqwords + i, State->qqwords[i]);
-	}
-}
-
-void FillSegment(Argon2d_Block *Matrix, uint32_t slice, uint32_t lane)
-{			
-	uint32_t startidx, prevoff, curoff;
-	Argon2d_Block State;
-	
-	startidx = (!slice) ? 2 : 0;
-	curoff = lane * LANE_LENGTH + slice * SEGMENT_LENGTH + startidx;
-	
-	//if(!(curoff % LANE_LENGTH)) prevoff = curoff + LANE_LENGTH - 1;
-	//else prevoff = curoff - 1;
-	
-	prevoff = (!(curoff % LANE_LENGTH)) ? curoff + LANE_LENGTH - 1 : curoff - 1;
-	
-	memcpy(State.data, (Matrix + prevoff)->data, 1024);
-	
-	int i;
-	for(i = startidx; i < SEGMENT_LENGTH; ++i, ++curoff, ++prevoff)
-	{
-		if((curoff % LANE_LENGTH) == 1) prevoff = curoff - 1;
-		
-		uint64_t pseudorand = Matrix[prevoff].data[0];
-		uint64_t reflane = (!slice) ? lane : (pseudorand >> 32) & 3;		// mod lanes
-				
-		uint32_t index = i;
-		bool samelane = reflane == lane;
-		pseudorand &= 0xFFFFFFFFULL;
-		uint32_t refareasize = ((reflane == lane) ? slice * SEGMENT_LENGTH + index - 1 : slice * SEGMENT_LENGTH + ((!index) ? -1 : 0));
-		
-		
-		if(!slice) refareasize = index - 1;
-		
-		uint64_t relativepos = (pseudorand & 0xFFFFFFFFULL);
-		relativepos = relativepos * relativepos >> 32;
-		relativepos = refareasize - 1 - (refareasize * relativepos >> 32);
-		
-		uint32_t startpos = 0;
-				
-		uint32_t abspos = (startpos + relativepos) % LANE_LENGTH;
-		
-		uint32_t refidx = abspos;
-		
-		Argon2dFillSingleBlock(&State, Matrix + (LANE_LENGTH * reflane + refidx), Matrix + curoff);
-	}
-}
-
-void *ThreadedSegmentFill(void *ThrData)
-{
-	Argon2ThreadData *Data = (Argon2ThreadData *)ThrData;
-	
-	FillSegment(Data->Matrix, Data->slice, Data->lane);
-	return(NULL);
-}
-
-void Argon2dFillAllBlocks(Argon2d_Block *Matrix)
-{
-	pthread_t ThrHandles[CONCURRENT_THREADS];
-	Argon2ThreadData ThrData[CONCURRENT_THREADS];
-	
-	int s;
-	for(s = 0; s < 4; ++s)
-	{
-		// WARNING: Assumes CONCURRENT_THREADS == lanes == 4
-		int l;
-		for(l = 0; l < 4; ++l)
-		{
-			FillSegment(Matrix, s, l);
-		}		
-	}
-}
-
-void Argon2dFinalizeHash(void *OutputHash, Argon2d_Block *Matrix)
-{
-	int l;
-	for(l = 1; l < 4; ++l)
-	{
-		int i;
-		for(i = 0; i < 32; ++i)
-			Matrix[LANE_LENGTH - 1].qqwords[i] = _mm256_xor_si256(Matrix[LANE_LENGTH - 1].qqwords[i], Matrix[LANE_LENGTH * l + (LANE_LENGTH - 1)].qqwords[i]);
-	}
-	
-	blake2b_long(OutputHash, 32, Matrix[LANE_LENGTH - 1].data, 1024);
-}
-
-void WolfArgon2dPoWHash(void *Output, void *Matrix, const void *BlkHdr)
-{
-	uint8_t tmp[72];
-		
-	Argon2dInitHash(tmp, (uint8_t *)BlkHdr);
-		
-	Argon2dFillFirstBlocks(Matrix, tmp);
-	
-	Argon2dFillAllBlocks(Matrix);
-	
-	Argon2dFinalizeHash((uint8_t *)Output, Matrix);
-}
-
-void WolfArgon2dAllocateCtx(void **Matrix)
-{
-	#ifdef _WIN32
-	*((Argon2d_Block **)Matrix) = (Argon2d_Block *)_aligned_malloc(32, sizeof(Argon2d_Block) * (SEGMENT_LENGTH << 4));
-	#else
-	*((Argon2d_Block **)Matrix) = (Argon2d_Block *)malloc(sizeof(Argon2d_Block) * (SEGMENT_LENGTH << 4));
-	posix_memalign(Matrix, 32, sizeof(Argon2d_Block) * (SEGMENT_LENGTH << 4));
-	#endif
-}
-
-void WolfArgon2dFreeCtx(void *Matrix)
-{
-	free(Matrix);
-}
-
-#endif
+*/

algo/argon2/argon2d/argon2d/argon2.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -29,10 +29,13 @@ extern "C" {
 /* Symbols visibility control */
 #ifdef A2_VISCTL
 #define ARGON2_PUBLIC __attribute__((visibility("default")))
+#define ARGON2_LOCAL __attribute__ ((visibility ("hidden")))
 #elif _MSC_VER
 #define ARGON2_PUBLIC __declspec(dllexport)
+#define ARGON2_LOCAL
 #else
 #define ARGON2_PUBLIC
+#define ARGON2_LOCAL
 #endif
 
 /*
@@ -206,6 +209,8 @@ typedef struct Argon2_Context {
     uint32_t lanes;   /* number of lanes */
     uint32_t threads; /* maximum number of threads */
 
+    uint32_t version; /* version number */
+
     allocate_fptr allocate_cbk; /* pointer to memory allocator */
     deallocate_fptr free_cbk;   /* pointer to memory deallocator */
 
@@ -214,9 +219,15 @@ typedef struct Argon2_Context {
 
 /* Argon2 primitive type */
 typedef enum Argon2_type {
-  Argon2_d = 0
+  Argon2_d = 0,
+  Argon2_i = 1,
+  Argon2_id = 2
 } argon2_type;
 
+/* Version of the algorithm */
+#define ARGON2_VERSION_10 0x10
+#define ARGON2_VERSION_13 0x13
+
 /*
  * Function that gives the string representation of an argon2_type.
  * @param type The argon2_type that we want the string for
@@ -233,8 +244,31 @@ ARGON2_PUBLIC const char *argon2_type2string(argon2_type type, int uppercase);
 ARGON2_PUBLIC int argon2_ctx(argon2_context *context, argon2_type type);
 
 /**
- * Hashes a password with Argon2i, producing a raw hash by allocating memory at
- * @hash
+ * Hashes a password with Argon2i, producing an encoded hash
+ * @param t_cost Number of iterations
+ * @param m_cost Sets memory usage to m_cost kibibytes
+ * @param parallelism Number of threads and compute lanes
+ * @param pwd Pointer to password
+ * @param pwdlen Password size in bytes
+ * @param salt Pointer to salt
+ * @param saltlen Salt size in bytes
+ * @param hashlen Desired length of the hash in bytes
+ * @param encoded Buffer where to write the encoded hash
+ * @param encodedlen Size of the buffer (thus max size of the encoded hash)
+ * @pre   Different parallelism levels will give different results
+ * @pre   Returns ARGON2_OK if successful
+ */
+ARGON2_PUBLIC int argon2i_hash_encoded(const uint32_t t_cost,
+                                       const uint32_t m_cost,
+                                       const uint32_t parallelism,
+                                       const void *pwd, const size_t pwdlen,
+                                       const void *salt, const size_t saltlen,
+                                       const size_t hashlen, char *encoded,
+                                       const size_t encodedlen,
+                                       const uint32_t version );
+
+/**
+ * Hashes a password with Argon2i, producing a raw hash at @hash
  * @param t_cost Number of iterations
  * @param m_cost Sets memory usage to m_cost kibibytes
  * @param parallelism Number of threads and compute lanes
@@ -247,11 +281,12 @@ ARGON2_PUBLIC int argon2_ctx(argon2_context *context, argon2_type type);
  * @pre   Different parallelism levels will give different results
  * @pre   Returns ARGON2_OK if successful
  */
-ARGON2_PUBLIC int argon2d_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
+ARGON2_PUBLIC int argon2i_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
                                    const uint32_t parallelism, const void *pwd,
                                    const size_t pwdlen, const void *salt,
                                    const size_t saltlen, void *hash,
-                                   const size_t hashlen);
+                                   const size_t hashlen,
+                                   const uint32_t version );
 
 ARGON2_PUBLIC int argon2d_hash_encoded(const uint32_t t_cost,
                                        const uint32_t m_cost,
@@ -259,7 +294,32 @@ ARGON2_PUBLIC int argon2d_hash_encoded(const uint32_t t_cost,
                                        const void *pwd, const size_t pwdlen,
                                        const void *salt, const size_t saltlen,
                                        const size_t hashlen, char *encoded,
-                                       const size_t encodedlen);
+                                       const size_t encodedlen,
+                                       const uint32_t version );
+
+ARGON2_PUBLIC int argon2d_hash_raw(const uint32_t t_cost, const uint32_t m_cost,
+                                   const uint32_t parallelism, const void *pwd,
+                                   const size_t pwdlen, const void *salt,
+                                   const size_t saltlen, void *hash,
+                                   const size_t hashlen,
+                                   const uint32_t version );
+
+ARGON2_PUBLIC int argon2id_hash_encoded(const uint32_t t_cost,
+                                        const uint32_t m_cost,
+                                        const uint32_t parallelism,
+                                        const void *pwd, const size_t pwdlen,
+                                        const void *salt, const size_t saltlen,
+                                        const size_t hashlen, char *encoded,
+                                        const size_t encodedlen,
+                                        const uint32_t version );
+
+ARGON2_PUBLIC int argon2id_hash_raw(const uint32_t t_cost,
+                                    const uint32_t m_cost,
+                                    const uint32_t parallelism, const void *pwd,
+                                    const size_t pwdlen, const void *salt,
+                                    const size_t saltlen, void *hash,
+                                    const size_t hashlen,
+                                    const uint32_t version );
 
 /* generic function underlying the above ones */
 ARGON2_PUBLIC int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
@@ -267,7 +327,8 @@ ARGON2_PUBLIC int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
                               const size_t pwdlen, const void *salt,
                               const size_t saltlen, void *hash,
                               const size_t hashlen, char *encoded,
-                              const size_t encodedlen, argon2_type type);
+                              const size_t encodedlen, argon2_type type,
+                              const uint32_t version );
 
 /**
  * Verifies a password against an encoded string
@@ -276,9 +337,15 @@ ARGON2_PUBLIC int argon2_hash(const uint32_t t_cost, const uint32_t m_cost,
  * @param pwd Pointer to password
  * @pre   Returns ARGON2_OK if successful
  */
+ARGON2_PUBLIC int argon2i_verify(const char *encoded, const void *pwd,
+                                 const size_t pwdlen);
+
 ARGON2_PUBLIC int argon2d_verify(const char *encoded, const void *pwd,
                                  const size_t pwdlen);
 
+ARGON2_PUBLIC int argon2id_verify(const char *encoded, const void *pwd,
+                                  const size_t pwdlen);
+
 /* generic function underlying the above ones */
 ARGON2_PUBLIC int argon2_verify(const char *encoded, const void *pwd,
                                 const size_t pwdlen, argon2_type type);
@@ -293,6 +360,27 @@ ARGON2_PUBLIC int argon2_verify(const char *encoded, const void *pwd,
  */
 ARGON2_PUBLIC int argon2d_ctx(argon2_context *context);
 
+/**
+ * Argon2i: Version of Argon2 that picks memory blocks
+ * independent on the password and salt. Good for side-channels,
+ * but worse w.r.t. tradeoff attacks if only one pass is used.
+ *****
+ * @param  context  Pointer to current Argon2 context
+ * @return  Zero if successful, a non zero error code otherwise
+ */
+ARGON2_PUBLIC int argon2i_ctx(argon2_context *context);
+
+/**
+ * Argon2id: Version of Argon2 where the first half-pass over memory is
+ * password-independent, the rest are password-dependent (on the password and
+ * salt). OK against side channels (they reduce to 1/2-pass Argon2i), and
+ * better with w.r.t. tradeoff attacks (similar to Argon2d).
+ *****
+ * @param  context  Pointer to current Argon2 context
+ * @return  Zero if successful, a non zero error code otherwise
+ */
+ARGON2_PUBLIC int argon2id_ctx(argon2_context *context);
+
 /**
  * Verify if a given password is correct for Argon2d hashing
  * @param  context  Pointer to current Argon2 context
@@ -302,6 +390,25 @@ ARGON2_PUBLIC int argon2d_ctx(argon2_context *context);
  */
 ARGON2_PUBLIC int argon2d_verify_ctx(argon2_context *context, const char *hash);
 
+/**
+ * Verify if a given password is correct for Argon2i hashing
+ * @param  context  Pointer to current Argon2 context
+ * @param  hash  The password hash to verify. The length of the hash is
+ * specified by the context outlen member
+ * @return  Zero if successful, a non zero error code otherwise
+ */
+ARGON2_PUBLIC int argon2i_verify_ctx(argon2_context *context, const char *hash);
+
+/**
+ * Verify if a given password is correct for Argon2id hashing
+ * @param  context  Pointer to current Argon2 context
+ * @param  hash  The password hash to verify. The length of the hash is
+ * specified by the context outlen member
+ * @return  Zero if successful, a non zero error code otherwise
+ */
+ARGON2_PUBLIC int argon2id_verify_ctx(argon2_context *context,
+                                      const char *hash);
+
 /* generic function underlying the above ones */
 ARGON2_PUBLIC int argon2_verify_ctx(argon2_context *context, const char *hash,
                                     argon2_type type);
@@ -326,18 +433,6 @@ ARGON2_PUBLIC size_t argon2_encodedlen(uint32_t t_cost, uint32_t m_cost,
                                        uint32_t parallelism, uint32_t saltlen,
                                        uint32_t hashlen, argon2_type type);
 
-#ifdef __AVX2__
-
-///////////////////////////
-// Wolf's Additions
-///////////////////////////
-
-void WolfArgon2dPoWHash(void *Output, void *Matrix, const void *BlkHdr);
-void WolfArgon2dAllocateCtx(void **Matrix);
-void WolfArgon2dFreeCtx(void *Matrix);
-
-#endif
-
 #if defined(__cplusplus)
 }
 #endif

algo/argon2/argon2d/argon2d/argon2d_thread.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -17,7 +17,7 @@
 
 #if !defined(ARGON2_NO_THREADS)
 
-#include "thread.h"
+#include "argon2d_thread.h"
 #if defined(_WIN32)
 #include <windows.h>
 #endif

algo/argon2/argon2d/argon2d/argon2d_thread.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -46,7 +46,7 @@ typedef pthread_t argon2_thread_handle_t;
  * @param func A function pointer for the thread's entry point. Must not be
  * NULL.
  * @param args Pointer that is passed as an argument to @func. May be NULL.
- * @return 0 if @handle and @func are valid pointers and a thread is successfuly
+ * @return 0 if @handle and @func are valid pointers and a thread is successfully
  * created.
  */
 int argon2_thread_create(argon2_thread_handle_t *handle,

algo/argon2/argon2d/argon2d/core.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -25,16 +25,20 @@
 #endif
 #define VC_GE_2005(version) (version >= 1400)
 
-#include <inttypes.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <mm_malloc.h>
 
 #include "core.h"
-#include "thread.h"
+#include "argon2d_thread.h"
 #include "../blake2/blake2.h"
 #include "../blake2/blake2-impl.h"
 
+#ifdef GENKAT
+#include "genkat.h"
+#endif
+
 #if defined(__clang__)
 #if __has_attribute(optnone)
 #define NOT_OPTIMIZED __attribute__((optnone))
@@ -96,7 +100,8 @@ int allocate_memory(const argon2_context *context, uint8_t **memory,
     if (context->allocate_cbk) {
         (context->allocate_cbk)(memory, memory_size);
     } else {
-        *memory = malloc(memory_size);
+        *memory = _mm_malloc( memory_size, 64 );
+//        *memory = malloc(memory_size);
     }
 
     if (*memory == NULL) {
@@ -109,11 +114,12 @@ int allocate_memory(const argon2_context *context, uint8_t **memory,
 void free_memory(const argon2_context *context, uint8_t *memory,
                  size_t num, size_t size) {
     size_t memory_size = num*size;
-    clear_internal_memory(memory, memory_size);
+//    clear_internal_memory(memory, memory_size);
     if (context->free_cbk) {
         (context->free_cbk)(memory, memory_size);
     } else {
-        free(memory);
+//        free(memory);
+        _mm_free( memory );
     }
 }
 
@@ -131,10 +137,10 @@ void NOT_OPTIMIZED secure_wipe_memory(void *v, size_t n) {
 }
 
 /* Memory clear flag defaults to true. */
-int FLAG_clear_internal_memory = 1;
+int FLAG_clear_internal_memory = 0;
 void clear_internal_memory(void *v, size_t n) {
   if (FLAG_clear_internal_memory && v) {
-    secure_wipe_memory(v, n);
+//    secure_wipe_memory(v, n);
   }
 }
 
@@ -163,6 +169,10 @@ void finalize(const argon2_context *context, argon2_instance_t *instance) {
             clear_internal_memory(blockhash_bytes, ARGON2_BLOCK_SIZE);
         }
 
+#ifdef GENKAT
+        print_tag(context->out, context->outlen);
+#endif
+
         free_memory(context, (uint8_t *)instance->memory,
                     instance->memory_blocks, sizeof(block));
     }
@@ -249,6 +259,9 @@ static int fill_memory_blocks_st(argon2_instance_t *instance) {
                 fill_segment(instance, position);
             }
         }
+#ifdef GENKAT
+        internal_kat(instance, r); /* Print all memory blocks */
+#endif
     }
     return ARGON2_OK;
 }
@@ -331,6 +344,10 @@ static int fill_memory_blocks_mt(argon2_instance_t *instance) {
                 }
             }
         }
+
+#ifdef GENKAT
+        internal_kat(instance, r); /* Print all memory blocks */
+#endif
     }
 
 fail:
@@ -530,7 +547,8 @@ void initial_hash(uint8_t *blockhash, argon2_context *context,
     store32(&value, context->t_cost);
     blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
 
-    store32(&value, ARGON2_VERSION_NUMBER);        
+//    store32(&value, ARGON2_VERSION_NUMBER);
+    store32(&value, context->version);
     blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
 
     store32(&value, (uint32_t)type);
@@ -538,17 +556,17 @@ void initial_hash(uint8_t *blockhash, argon2_context *context,
 
     store32(&value, context->pwdlen);
     blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
-		
+
     if (context->pwd != NULL) {
         blake2b_update(&BlakeHash, (const uint8_t *)context->pwd,
                        context->pwdlen);
 
         if (context->flags & ARGON2_FLAG_CLEAR_PASSWORD) {
-            secure_wipe_memory(context->pwd, context->pwdlen);
+//            secure_wipe_memory(context->pwd, context->pwdlen);
             context->pwdlen = 0;
         }
     }
-		
+
     store32(&value, context->saltlen);
     blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value));
 
@@ -565,7 +583,7 @@ void initial_hash(uint8_t *blockhash, argon2_context *context,
                        context->secretlen);
 
         if (context->flags & ARGON2_FLAG_CLEAR_SECRET) {
-            secure_wipe_memory(context->secret, context->secretlen);
+//            secure_wipe_memory(context->secret, context->secretlen);
             context->secretlen = 0;
         }
     }
@@ -602,11 +620,14 @@ int initialize(argon2_instance_t *instance, argon2_context *context) {
     /* Hashing all inputs */
     initial_hash(blockhash, context, instance->type);
     /* Zeroing 8 extra bytes */
-
     clear_internal_memory(blockhash + ARGON2_PREHASH_DIGEST_LENGTH,
                           ARGON2_PREHASH_SEED_LENGTH -
                               ARGON2_PREHASH_DIGEST_LENGTH);
 
+#ifdef GENKAT
+    initial_kat(blockhash, context, instance->type);
+#endif
+
     /* 3. Creating first blocks, we always have at least two blocks in a slice
      */
     fill_first_blocks(blockhash, instance);

algo/argon2/argon2d/argon2d/core.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -25,12 +25,12 @@
 /**********************Argon2 internal constants*******************************/
 
 enum argon2_core_constants {
-    /* Version of the algorithm */   
-    ARGON2_VERSION_NUMBER = 0x10,
     /* Memory block size in bytes */
     ARGON2_BLOCK_SIZE = 1024,
     ARGON2_QWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 8,
     ARGON2_OWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 16,
+    ARGON2_HWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 32,
+    ARGON2_512BIT_WORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 64,
 
     /* Number of pseudo-random values generated by one call to Blake in Argon2i
        to
@@ -76,7 +76,6 @@ typedef struct Argon2_instance_t {
     uint32_t segment_length;
     uint32_t lane_length;
     uint32_t lanes;
-    uint32_t limit;
     uint32_t threads;
     argon2_type type;
     int print_internals; /* whether to print the memory blocks */

algo/argon2/argon2d/argon2d/encoding.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -326,6 +326,10 @@ int decode_string(argon2_context *ctx, const char *str, argon2_type type) {
     CC("$");
     CC(type_string);
 
+    /* Reading the version number if the default is suppressed */
+    ctx->version = ARGON2_VERSION_10;
+    CC_opt("$v=", DECIMAL_U32(ctx->version));
+
     CC("$m=");
     DECIMAL_U32(ctx->m_cost);
     CC(",t=");
@@ -411,6 +415,9 @@ int encode_string(char *dst, size_t dst_len, argon2_context *ctx,
     SS("$");
     SS(type_string);
 
+    SS("$v=");
+    SX(ctx->version);
+
     SS("$m=");
     SX(ctx->m_cost);
     SS(",t=");

algo/argon2/argon2d/argon2d/encoding.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *

algo/argon2/argon2d/argon2d/opt.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -21,7 +21,7 @@
 
 #include "argon2.h"
 #include "core.h"
-
+#include "simd-utils.h"
 #include "../blake2/blake2.h"
 #include "../blake2/blamka-round-opt.h"
 
@@ -34,6 +34,103 @@
  * @param with_xor Whether to XOR into the new block (1) or just overwrite (0)
  * @pre all block pointers must be valid
  */
+
+#if defined(__AVX512F__)
+
+static inline __m512i blamka( __m512i x, __m512i y )
+{
+    __m512i xy = _mm512_mul_epu32( x, y );
+    return _mm512_add_epi64( _mm512_add_epi64( x, y ),
+                             _mm512_add_epi64( xy, xy ) );
+}
+
+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_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] );
+    BLAKE2_ROUND_1( state[ 8], state[ 9], state[10], state[11],
+                    state[12], state[13], state[14], state[15] );
+
+    BLAKE2_ROUND_2( state[ 0], state[ 2], state[ 4], state[ 6],
+                    state[ 8], state[10], state[12], state[14] );
+    BLAKE2_ROUND_2( state[ 1], state[ 3], state[ 5], state[ 7],
+                    state[ 9], state[11], state[13], state[15] );
+
+    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] );
+    }
+}
+
+#elif defined(__AVX2__)
+
+static void fill_block(__m256i *state, const block *ref_block,
+                       block *next_block, int with_xor) {
+    __m256i block_XY[ARGON2_HWORDS_IN_BLOCK];
+    unsigned int i;
+
+    if (with_xor) {
+        for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
+            state[i] = _mm256_xor_si256(
+                state[i], _mm256_load_si256((const __m256i *)ref_block->v + i));
+            block_XY[i] = _mm256_xor_si256(
+                state[i], _mm256_load_si256((const __m256i *)next_block->v + i));
+        }
+    } else {
+        for (i = 0; i < ARGON2_HWORDS_IN_BLOCK; i++) {
+            block_XY[i] = state[i] = _mm256_xor_si256(
+                state[i], _mm256_load_si256((const __m256i *)ref_block->v + i));
+        }
+    }
+
+    BLAKE2_ROUND_1( state[ 0], state[ 4], state[ 1], state[ 5],
+                    state[ 2], state[ 6], state[ 3], state[ 7] );
+    BLAKE2_ROUND_1( state[ 8], state[12], state[ 9], state[13],
+                    state[10], state[14], state[11], state[15] );
+    BLAKE2_ROUND_1( state[16], state[20], state[17], state[21],
+                    state[18], state[22], state[19], state[23] );
+    BLAKE2_ROUND_1( state[24], state[28], state[25], state[29],
+                    state[26], state[30], state[27], state[31] );
+
+    BLAKE2_ROUND_2( state[ 0], state[ 4], state[ 8], state[12],
+                    state[16], state[20], state[24], state[28] );
+    BLAKE2_ROUND_2( state[ 1], state[ 5], state[ 9], state[13],
+                    state[17], state[21], state[25], state[29] );
+    BLAKE2_ROUND_2( state[ 2], state[ 6], state[10], state[14],
+                    state[18], state[22], state[26], state[30] );
+    BLAKE2_ROUND_2( state[ 3], state[ 7], state[11], state[15],
+                    state[19], state[23], state[27], state[31] );
+
+    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]);
+    }
+}
+
+#else  // SSE2
+
 static void fill_block(__m128i *state, const block *ref_block,
                        block *next_block, int with_xor) {
     __m128i block_XY[ARGON2_OWORDS_IN_BLOCK];
@@ -42,39 +139,72 @@ 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));
         }
     }
 
-    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]);
-    }
+    BLAKE2_ROUND( state[ 0], state[ 1], state[ 2], state[ 3],
+                  state[ 4], state[ 5], state[ 6], state[ 7] );
+    BLAKE2_ROUND( state[ 8], state[ 9], state[10], state[11], 
+                  state[12], state[13], state[14], state[15] );
+    BLAKE2_ROUND( state[16], state[17], state[18], state[19], 
+                  state[20], state[21], state[22], state[23] );
+    BLAKE2_ROUND( state[24], state[25], state[26], state[27], 
+                  state[28], state[29], state[30], state[31] );
+    BLAKE2_ROUND( state[32], state[33], state[34], state[35], 
+                  state[36], state[37], state[38], state[39] );
+    BLAKE2_ROUND( state[40], state[41], state[42], state[43], 
+                  state[44], state[45], state[46], state[47] );
+    BLAKE2_ROUND( state[48], state[49], state[50], state[51], 
+                  state[52], state[53], state[54], state[55] );
+    BLAKE2_ROUND( state[56], state[57], state[58], state[59], 
+                  state[60], state[61], state[62], state[63] );
 
-    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]);
-    }
+    BLAKE2_ROUND( state[ 0], state[ 8], state[16], state[24], 
+                  state[32], state[40], state[48], state[56] );
+    BLAKE2_ROUND( state[ 1], state[ 9], state[17], state[25],  
+                  state[33], state[41], state[49], state[57] );
+    BLAKE2_ROUND( state[ 2], state[10], state[18], state[26],  
+                  state[34], state[42], state[50], state[58] );
+    BLAKE2_ROUND( state[ 3], state[11], state[19], state[27],  
+                  state[35], state[43], state[51], state[59] );
+    BLAKE2_ROUND( state[ 4], state[12], state[20], state[28],  
+                  state[36], state[44], state[52], state[60] );
+    BLAKE2_ROUND( state[ 5], state[13], state[21], state[29],  
+                  state[37], state[45], state[53], state[61] );
+    BLAKE2_ROUND( state[ 6], state[14], state[22], state[30],  
+                  state[38], state[46], state[54], state[62] );
+    BLAKE2_ROUND( state[ 7], state[15], state[23], state[31],  
+                  state[39], state[47], state[55], state[63] );
 
     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]);
     }
 }
 
+#endif
+
+#if 0
 static void next_addresses(block *address_block, block *input_block) {
     /*Temporary zero-initialized blocks*/
+#if defined(__AVX512F__)
+    __m512i zero_block[ARGON2_512BIT_WORDS_IN_BLOCK];
+    __m512i zero2_block[ARGON2_512BIT_WORDS_IN_BLOCK];
+#elif defined(__AVX2__)
+    __m256i zero_block[ARGON2_HWORDS_IN_BLOCK];
+    __m256i zero2_block[ARGON2_HWORDS_IN_BLOCK];
+#else
     __m128i zero_block[ARGON2_OWORDS_IN_BLOCK];
     __m128i zero2_block[ARGON2_OWORDS_IN_BLOCK];
+#endif
 
     memset(zero_block, 0, sizeof(zero_block));
     memset(zero2_block, 0, sizeof(zero2_block));
@@ -88,30 +218,53 @@ static void next_addresses(block *address_block, block *input_block) {
     /*Second iteration of G*/
     fill_block(zero2_block, address_block, address_block, 0);
 }
+#endif
 
 void fill_segment(const argon2_instance_t *instance,
                   argon2_position_t position) {
     block *ref_block = NULL, *curr_block = NULL;
-    block address_block, input_block;
+//    block address_block, input_block;
     uint64_t pseudo_rand, ref_index, ref_lane;
     uint32_t prev_offset, curr_offset;
     uint32_t starting_index, i;
-    __m128i state[64];
-    int data_independent_addressing;
+#if defined(__AVX512F__)
+    __m512i state[ARGON2_512BIT_WORDS_IN_BLOCK];
+#elif defined(__AVX2__)
+    __m256i state[ARGON2_HWORDS_IN_BLOCK];
+#else
+    __m128i state[ARGON2_OWORDS_IN_BLOCK];
+#endif
+//    int data_independent_addressing;
 
     if (instance == NULL) {
         return;
     }
 
+    // data_independent_addressing =
+    //     (instance->type == Argon2_i) ||
+    //     (instance->type == Argon2_id && (position.pass == 0) &&
+    //      (position.slice < ARGON2_SYNC_POINTS / 2));
+
+    // if (data_independent_addressing) {
+    //     init_block_value(&input_block, 0);
+
+    //     input_block.v[0] = position.pass;
+    //     input_block.v[1] = position.lane;
+    //     input_block.v[2] = position.slice;
+    //     input_block.v[3] = instance->memory_blocks;
+    //     input_block.v[4] = instance->passes;
+    //     input_block.v[5] = instance->type;
+    // }
+
     starting_index = 0;
 
     if ((0 == position.pass) && (0 == position.slice)) {
         starting_index = 2; /* we have already generated the first two blocks */
 
         /* Don't forget to generate the first block of addresses: */
-        if (data_independent_addressing) {
-            next_addresses(&address_block, &input_block);
-        }
+//        if (data_independent_addressing) {
+//            next_addresses(&address_block, &input_block);
+//        }
     }
 
     /* Offset of the current block */
@@ -137,14 +290,14 @@ void fill_segment(const argon2_instance_t *instance,
 
         /* 1.2 Computing the index of the reference block */
         /* 1.2.1 Taking pseudo-random value from the previous block */
-        if (data_independent_addressing) {
-            if (i % ARGON2_ADDRESSES_IN_BLOCK == 0) {
-                next_addresses(&address_block, &input_block);
-            }
-            pseudo_rand = address_block.v[i % ARGON2_ADDRESSES_IN_BLOCK];
-        } else {
+//        if (data_independent_addressing) {
+//            if (i % ARGON2_ADDRESSES_IN_BLOCK == 0) {
+//                next_addresses(&address_block, &input_block);
+//            }
+//            pseudo_rand = address_block.v[i % ARGON2_ADDRESSES_IN_BLOCK];
+//        } else {
             pseudo_rand = instance->memory[prev_offset].v[0];
-        }
+//        }
 
         /* 1.2.2 Computing the lane of the reference block */
         ref_lane = ((pseudo_rand >> 32)) % instance->lanes;
@@ -165,8 +318,15 @@ void fill_segment(const argon2_instance_t *instance,
         ref_block =
             instance->memory + instance->lane_length * ref_lane + ref_index;
         curr_block = instance->memory + curr_offset;
-            
-        fill_block(state, ref_block, curr_block, 0);
-
+         if (ARGON2_VERSION_10 == instance->version) {
+             /* version 1.2.1 and earlier: overwrite, not XOR */
+             fill_block(state, ref_block, curr_block, 0);
+         } else {
+             if(0 == position.pass) {
+                fill_block(state, ref_block, curr_block, 0);
+             } else {
+                 fill_block(state, ref_block, curr_block, 1);
+             }
+         }
     }
 }

algo/argon2/argon2d/blake2/blake2-impl.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -153,4 +153,4 @@ static BLAKE2_INLINE uint64_t rotr64(const uint64_t w, const unsigned c) {
 
 void clear_internal_memory(void *v, size_t n);
 
-#endif
+#endif

algo/argon2/argon2d/blake2/blake2.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -78,7 +78,7 @@ int blake2b_final(blake2b_state *S, void *out, size_t outlen);
 
 /* Simple API */
 int blake2b(void *out, size_t outlen, const void *in, size_t inlen,
-            const void *key, size_t keylen);
+                         const void *key, size_t keylen);
 
 /* Argon2 Team - Begin Code */
 int blake2b_long(void *out, size_t outlen, const void *in, size_t inlen);
@@ -88,4 +88,4 @@ int blake2b_long(void *out, size_t outlen, const void *in, size_t inlen);
 }
 #endif
 
-#endif
+#endif

algo/argon2/argon2d/blake2/blake2b.c

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -387,4 +387,4 @@ fail:
     return ret;
 #undef TRY
 }
-/* Argon2 Team - End Code */
+/* Argon2 Team - End Code */

algo/argon2/argon2d/blake2/blamka-round-opt.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -29,6 +29,10 @@
 #include <x86intrin.h>
 #endif
 
+#include "simd-utils.h"
+
+#if !defined(__AVX512F__)
+#if !defined(__AVX2__)
 #if !defined(__XOP__)
 #if defined(__SSSE3__)
 #define r16                                                                    \
@@ -176,5 +180,286 @@ static BLAKE2_INLINE __m128i fBlaMka(__m128i x, __m128i y) {
                                                                                \
         UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1);                         \
     } while ((void)0, 0)
+#else /* __AVX2__ */
 
-#endif
+#include <immintrin.h>
+
+#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))
+//#define rotr16(x)   _mm256_shuffle_epi8(x, _mm256_setr_epi8(2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9, 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9))
+//#define rotr63(x)   _mm256_xor_si256(_mm256_srli_epi64((x), 63), _mm256_add_epi64((x), (x)))
+
+#define G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do { \
+        __m256i ml0, ml1; \
+        ml0 = _mm256_mul_epu32(A0, B0); \
+        ml1 = _mm256_mul_epu32(A1, B1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml0)); \
+        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml1)); \
+        D0 = _mm256_xor_si256(D0, A0); \
+        D1 = _mm256_xor_si256(D1, A1); \
+        D0 = rotr32(D0); \
+        D1 = rotr32(D1); \
+        ml0 = _mm256_mul_epu32(C0, D0); \
+        ml1 = _mm256_mul_epu32(C1, D1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml0)); \
+        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml1)); \
+        B0 = _mm256_xor_si256(B0, C0); \
+        B1 = _mm256_xor_si256(B1, C1); \
+        B0 = rotr24(B0); \
+        B1 = rotr24(B1); \
+    } while((void)0, 0);
+
+#define G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do { \
+        __m256i ml0, ml1; \
+        ml0 = _mm256_mul_epu32(A0, B0); \
+        ml1 = _mm256_mul_epu32(A1, B1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        A0 = _mm256_add_epi64(A0, _mm256_add_epi64(B0, ml0)); \
+        A1 = _mm256_add_epi64(A1, _mm256_add_epi64(B1, ml1)); \
+        D0 = _mm256_xor_si256(D0, A0); \
+        D1 = _mm256_xor_si256(D1, A1); \
+        D0 = rotr16(D0); \
+        D1 = rotr16(D1); \
+        ml0 = _mm256_mul_epu32(C0, D0); \
+        ml1 = _mm256_mul_epu32(C1, D1); \
+        ml0 = _mm256_add_epi64(ml0, ml0); \
+        ml1 = _mm256_add_epi64(ml1, ml1); \
+        C0 = _mm256_add_epi64(C0, _mm256_add_epi64(D0, ml0)); \
+        C1 = _mm256_add_epi64(C1, _mm256_add_epi64(D1, ml1)); \
+        B0 = _mm256_xor_si256(B0, C0); \
+        B1 = _mm256_xor_si256(B1, C1); \
+        B0 = rotr63(B0); \
+        B1 = rotr63(B1); \
+    } while((void)0, 0);
+
+#define DIAGONALIZE_1(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        B0 = _mm256_permute4x64_epi64(B0, _MM_SHUFFLE(0, 3, 2, 1)); \
+        C0 = _mm256_permute4x64_epi64(C0, _MM_SHUFFLE(1, 0, 3, 2)); \
+        D0 = _mm256_permute4x64_epi64(D0, _MM_SHUFFLE(2, 1, 0, 3)); \
+        \
+        B1 = _mm256_permute4x64_epi64(B1, _MM_SHUFFLE(0, 3, 2, 1)); \
+        C1 = _mm256_permute4x64_epi64(C1, _MM_SHUFFLE(1, 0, 3, 2)); \
+        D1 = _mm256_permute4x64_epi64(D1, _MM_SHUFFLE(2, 1, 0, 3)); \
+    } while((void)0, 0);
+
+#define DIAGONALIZE_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do { \
+        __m256i tmp1 = _mm256_blend_epi32(B0, B1, 0xCC); \
+        __m256i tmp2 = _mm256_blend_epi32(B0, B1, 0x33); \
+        B1 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+        tmp1 = C0; \
+        B0 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        C0 = C1; \
+        tmp2 = _mm256_blend_epi32(D0, D1, 0x33); \
+        C1 = tmp1; \
+        tmp1 = _mm256_blend_epi32(D0, D1, 0xCC); \
+        D1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+    } while(0);
+
+#define UNDIAGONALIZE_1(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        B0 = _mm256_permute4x64_epi64(B0, _MM_SHUFFLE(2, 1, 0, 3)); \
+        C0 = _mm256_permute4x64_epi64(C0, _MM_SHUFFLE(1, 0, 3, 2)); \
+        D0 = _mm256_permute4x64_epi64(D0, _MM_SHUFFLE(0, 3, 2, 1)); \
+        \
+        B1 = _mm256_permute4x64_epi64(B1, _MM_SHUFFLE(2, 1, 0, 3)); \
+        C1 = _mm256_permute4x64_epi64(C1, _MM_SHUFFLE(1, 0, 3, 2)); \
+        D1 = _mm256_permute4x64_epi64(D1, _MM_SHUFFLE(0, 3, 2, 1)); \
+    } while((void)0, 0);
+
+#define UNDIAGONALIZE_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do { \
+        __m256i tmp1 = _mm256_blend_epi32(B0, B1, 0xCC); \
+        __m256i tmp2 = _mm256_blend_epi32(B0, B1, 0x33); \
+        B0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+        tmp1 = C0; \
+        B1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        C0 = C1; \
+        tmp2 = _mm256_blend_epi32(D0, D1, 0xCC); \
+        C1 = tmp1; \
+        tmp1 = _mm256_blend_epi32(D0, D1, 0x33); \
+        D1 = _mm256_permute4x64_epi64(tmp2, _MM_SHUFFLE(2,3,0,1)); \
+        D0 = _mm256_permute4x64_epi64(tmp1, _MM_SHUFFLE(2,3,0,1)); \
+    } while((void)0, 0);
+
+#define BLAKE2_ROUND_1(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do{ \
+        G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        \
+        DIAGONALIZE_1(A0, B0, C0, D0, A1, B1, C1, D1) \
+        \
+        G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        \
+        UNDIAGONALIZE_1(A0, B0, C0, D0, A1, B1, C1, D1) \
+    } while((void)0, 0);
+
+#define BLAKE2_ROUND_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do{ \
+        G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        \
+        DIAGONALIZE_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        \
+        G1_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        G2_AVX2(A0, A1, B0, B1, C0, C1, D0, D1) \
+        \
+        UNDIAGONALIZE_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    } while((void)0, 0);
+
+#endif /* __AVX2__ */
+
+#else /* __AVX512F__ */
+
+#include <immintrin.h>
+
+static inline __m512i muladd(__m512i x, __m512i y)
+{
+    __m512i z = _mm512_mul_epu32(x, y);
+    return _mm512_add_epi64(_mm512_add_epi64(x, y), _mm512_add_epi64(z, z));
+}
+
+#define G1(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        A0 = muladd(A0, B0); \
+        A1 = muladd(A1, B1); \
+\
+        D0 = _mm512_xor_si512(D0, A0); \
+        D1 = _mm512_xor_si512(D1, A1); \
+\
+        D0 = _mm512_ror_epi64(D0, 32); \
+        D1 = _mm512_ror_epi64(D1, 32); \
+\
+        C0 = muladd(C0, D0); \
+        C1 = muladd(C1, D1); \
+\
+        B0 = _mm512_xor_si512(B0, C0); \
+        B1 = _mm512_xor_si512(B1, C1); \
+\
+        B0 = _mm512_ror_epi64(B0, 24); \
+        B1 = _mm512_ror_epi64(B1, 24); \
+    } while ((void)0, 0)
+
+#define G2(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        A0 = muladd(A0, B0); \
+        A1 = muladd(A1, B1); \
+\
+        D0 = _mm512_xor_si512(D0, A0); \
+        D1 = _mm512_xor_si512(D1, A1); \
+\
+        D0 = _mm512_ror_epi64(D0, 16); \
+        D1 = _mm512_ror_epi64(D1, 16); \
+\
+        C0 = muladd(C0, D0); \
+        C1 = muladd(C1, D1); \
+\
+        B0 = _mm512_xor_si512(B0, C0); \
+        B1 = _mm512_xor_si512(B1, C1); \
+\
+        B0 = _mm512_ror_epi64(B0, 63); \
+        B1 = _mm512_ror_epi64(B1, 63); \
+    } while ((void)0, 0)
+
+#define DIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        B0 = _mm512_permutex_epi64(B0, _MM_SHUFFLE(0, 3, 2, 1)); \
+        B1 = _mm512_permutex_epi64(B1, _MM_SHUFFLE(0, 3, 2, 1)); \
+\
+        C0 = _mm512_permutex_epi64(C0, _MM_SHUFFLE(1, 0, 3, 2)); \
+        C1 = _mm512_permutex_epi64(C1, _MM_SHUFFLE(1, 0, 3, 2)); \
+\
+        D0 = _mm512_permutex_epi64(D0, _MM_SHUFFLE(2, 1, 0, 3)); \
+        D1 = _mm512_permutex_epi64(D1, _MM_SHUFFLE(2, 1, 0, 3)); \
+    } while ((void)0, 0)
+
+#define UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        B0 = _mm512_permutex_epi64(B0, _MM_SHUFFLE(2, 1, 0, 3)); \
+        B1 = _mm512_permutex_epi64(B1, _MM_SHUFFLE(2, 1, 0, 3)); \
+\
+        C0 = _mm512_permutex_epi64(C0, _MM_SHUFFLE(1, 0, 3, 2)); \
+        C1 = _mm512_permutex_epi64(C1, _MM_SHUFFLE(1, 0, 3, 2)); \
+\
+        D0 = _mm512_permutex_epi64(D0, _MM_SHUFFLE(0, 3, 2, 1)); \
+        D1 = _mm512_permutex_epi64(D1, _MM_SHUFFLE(0, 3, 2, 1)); \
+    } while ((void)0, 0)
+
+#define BLAKE2_ROUND(A0, B0, C0, D0, A1, B1, C1, D1) \
+    do { \
+        G1(A0, B0, C0, D0, A1, B1, C1, D1); \
+        G2(A0, B0, C0, D0, A1, B1, C1, D1); \
+\
+        DIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1); \
+\
+        G1(A0, B0, C0, D0, A1, B1, C1, D1); \
+        G2(A0, B0, C0, D0, A1, B1, C1, D1); \
+\
+        UNDIAGONALIZE(A0, B0, C0, D0, A1, B1, C1, D1); \
+    } while ((void)0, 0)
+
+#define SWAP_HALVES(A0, A1) \
+    do { \
+        __m512i t; \
+        t = _mm512_shuffle_i64x2(A0, A1, _MM_SHUFFLE(1, 0, 1, 0)); \
+        A1 = _mm512_shuffle_i64x2(A0, A1, _MM_SHUFFLE(3, 2, 3, 2)); \
+        A0 = t; \
+    } while((void)0, 0)
+
+#define SWAP_QUARTERS(A0, A1) \
+    do { \
+        SWAP_HALVES(A0, 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_shuffle_i64x2( A0, A0, 0xd8 ); \
+        A1 = _mm512_shuffle_i64x2( A1, A1, 0xd8 ); \
+        SWAP_HALVES(A0, A1); \
+    } while((void)0, 0)
+
+#define BLAKE2_ROUND_1(A0, C0, B0, D0, A1, C1, B1, D1) \
+    do { \
+        SWAP_HALVES(A0, B0); \
+        SWAP_HALVES(C0, D0); \
+        SWAP_HALVES(A1, B1); \
+        SWAP_HALVES(C1, D1); \
+        BLAKE2_ROUND(A0, B0, C0, D0, A1, B1, C1, D1); \
+        SWAP_HALVES(A0, B0); \
+        SWAP_HALVES(C0, D0); \
+        SWAP_HALVES(A1, B1); \
+        SWAP_HALVES(C1, D1); \
+    } while ((void)0, 0)
+
+#define BLAKE2_ROUND_2(A0, A1, B0, B1, C0, C1, D0, D1) \
+    do { \
+        SWAP_QUARTERS(A0, A1); \
+        SWAP_QUARTERS(B0, B1); \
+        SWAP_QUARTERS(C0, C1); \
+        SWAP_QUARTERS(D0, D1); \
+        BLAKE2_ROUND(A0, B0, C0, D0, A1, B1, C1, D1); \
+        UNSWAP_QUARTERS(A0, A1); \
+        UNSWAP_QUARTERS(B0, B1); \
+        UNSWAP_QUARTERS(C0, C1); \
+        UNSWAP_QUARTERS(D0, D1); \
+    } while ((void)0, 0)
+
+#endif /* __AVX512F__ */
+#endif /* BLAKE_ROUND_MKA_OPT_H */

algo/argon2/argon2d/blake2/blamka-round-ref.h

@@ -4,7 +4,7 @@
  * Copyright 2015
  * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves
  *
- * You may use this work under the terms of a Creative Commons CC0 1.0 
+ * You may use this work under the terms of a Creative Commons CC0 1.0
  * License/Waiver or the Apache Public License 2.0, at your option. The terms of
  * these licenses can be found at:
  *
@@ -21,7 +21,7 @@
 #include "blake2.h"
 #include "blake2-impl.h"
 
-/*designed by the Lyra PHC team */
+/* designed by the Lyra PHC team */
 static BLAKE2_INLINE uint64_t fBlaMka(uint64_t x, uint64_t y) {
     const uint64_t m = UINT64_C(0xFFFFFFFF);
     const uint64_t xy = (x & m) * (y & m);
@@ -53,4 +53,4 @@ static BLAKE2_INLINE uint64_t fBlaMka(uint64_t x, uint64_t y) {
         G(v3, v4, v9, v14);                                                    \
     } while ((void)0, 0)
 
-#endif
+#endif

algo/blake/blake-4way.c

@@ -13,13 +13,13 @@ 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 );
-     mm_deinterleave_4x32( state, state+32, state+64, state+96, vhash, 256 );
+     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
 }
 
-int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done )
+int scanhash_blake_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*4] __attribute__ ((aligned (64)));
    uint32_t hash[8*4] __attribute__ ((aligned (32)));
@@ -27,43 +27,34 @@ int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t HTarget = ptarget[7];
-   uint32_t _ALIGN(32) edata[20];
+   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
    if (opt_benchmark)
       HTarget = 0x7f;
 
-   // we need big endian data...
-   swab32_array( edata, pdata, 20 );
-   mm_interleave_4x32( vdata, edata, edata, edata, edata, 640 );
+   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 );
 
-   uint32_t *noncep = vdata + 76;   // 19*4
    do {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
+      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
 
       blakehash_4way( hash, vdata );
 
       for ( int i = 0; i < 4; i++ )
-      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget ) )
+      if ( (hash+(i<<3))[7] <= HTarget )
+      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
       {
           pdata[19] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+          submit_solution( work, hash+(i<<3), mythr );
       }
       n += 4;
 
-   } while ( (num_found == 0) && (n < max_nonce) 
-             && !work_restart[thr_id].restart );
-
+   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #endif
@@ -79,13 +70,13 @@ void blakehash_8way( void *state, const void *input )
      memcpy( &ctx, &blake_8w_ctx, sizeof ctx );
      blake256r14_8way( &ctx, input + (64<<3), 16 );
      blake256r14_8way_close( &ctx, vhash );
-     mm256_deinterleave_8x32( state,     state+ 32, state+ 64, state+ 96,
-                              state+128, state+160, state+192, state+224,
-                              vhash, 256 );
+     _dintrlv_8x32( state,     state+ 32, state+ 64, state+ 96,
+                    state+128, state+160, state+192, state+224,
+                    vhash, 256 );
 }
 
-int scanhash_blake_8way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done )
+int scanhash_blake_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*8] __attribute__ ((aligned (64)));
    uint32_t hash[8*8] __attribute__ ((aligned (32)));
@@ -93,33 +84,21 @@ int scanhash_blake_8way( int thr_id, struct work *work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t HTarget = ptarget[7];
-   uint32_t _ALIGN(32) edata[20];
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
+   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
    if (opt_benchmark)
       HTarget = 0x7f;
 
-   // we need big endian data...
-   swab32_array( edata, pdata, 20 );
-
-   mm256_interleave_8x32( vdata, edata, edata, edata, edata,
-                                 edata, edata, edata, edata, 640 );
+   mm256_bswap32_intrlv80_8x32( vdata, pdata );
 
    blake256r14_8way_init( &blake_8w_ctx );
    blake256r14_8way( &blake_8w_ctx, vdata, 64 );
 
-   uint32_t *noncep = vdata + 152;   // 19*8
    do {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
-      be32enc( noncep +4, n+4 );
-      be32enc( noncep +5, n+5 );
-      be32enc( noncep +6, n+6 );
-      be32enc( noncep +7, n+7 );
+      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
+                                                  n+3, n+2, n+1, n ) );
       pdata[19] = n;
 
       blakehash_8way( hash, vdata );
@@ -128,17 +107,14 @@ int scanhash_blake_8way( int thr_id, struct work *work, uint32_t max_nonce,
       if ( (hash+i)[7] <= HTarget && fulltest( hash+i, ptarget ) )
       {
           pdata[19] = n+i;
-          num_found++;
-          nonces[i] = n+i;
-          work_set_target_ratio( work, hash+1 );
+          submit_solution( work, hash+(i<<3), mythr );
       }
       n += 8;
 
-   } while ( (num_found == 0) && (n < max_nonce)
-             && !work_restart[thr_id].restart );
+   } while ( (n < max_nonce) !work_restart[thr_id].restart );
 
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #endif

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;

algo/blake/blake-gate.h

@@ -10,12 +10,12 @@
 
 #if defined (BLAKE_4WAY)
 void blakehash_4way(void *state, const void *input);
-int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_blake_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 
 void blakehash( void *state, const void *input );
-int scanhash_blake( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done );
+int scanhash_blake( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
 
 #endif

algo/blake/blake-hash-4way.h

@@ -37,60 +37,71 @@
 #ifndef __BLAKE_HASH_4WAY__
 #define __BLAKE_HASH_4WAY__ 1
 
-#ifdef __SSE4_2__
-
 #ifdef __cplusplus
 extern "C"{
 #endif
 
 #include <stddef.h>
 #include "algo/sha/sph_types.h"
-#include "avxdefs.h"
+#include "simd-utils.h"
 
 #define SPH_SIZE_blake256   256
 
 #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 1 way SSE2
 
-// Blake-256 4 way
+void  blake256_transform_le( uint32_t *H, const uint32_t *buf,
+                             const uint32_t T0, const uint32_t T1 );
+
+/////////////////////////
+//
+//  Blake-512 1 way SSE2
+
+void  blake512_transform_le( uint64_t *H, const uint64_t *buf,
+                             const uint64_t T0, const uint64_t T1 );
+
+//////////////////////////
+//
+//   Blake-256 4 way SSE2
 
 typedef struct {
-   __m128i buf[16] __attribute__ ((aligned (64)));
-   __m128i H[8];
-   __m128i S[4];    
+   unsigned char buf[64<<2];
+   uint32_t H[8<<2];
    size_t ptr;
-   sph_u32 T0, T1;
+   uint32_t T0, T1;
    int rounds;   // 14 for blake, 8 for blakecoin & vanilla
-} blake_4way_small_context;
+} 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 *cc);
-void blake256_4way(void *cc, const void *data, size_t len);
-void blake256_4way_close(void *cc, void *dst);
+void blake256_4way_init(void *ctx);
+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
@@ -99,45 +110,122 @@ 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);
+void blake256_8way_update_le(void *cc, const void *data, size_t len);
+void blake256_8way_close_le(void *cc, void *dst);
+void blake256_8way_round0_prehash_le( void *midstate, const void *midhash,
+                                      void *data );
+void blake256_8way_final_rounds_le( void *final_hash, const void *midstate,
+                                     const void *midhash, const void *data );
 
 // 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 );
+void blake512_4way_full_le( blake_4way_big_context *sc, void * dst,
+                            const void *data, size_t len );
+void blake512_4way_prehash_le( blake_4way_big_context *sc, __m256i *midstate,
+                               const void *data );
+void blake512_4way_final_le( blake_4way_big_context *sc, void *hash,
+                             const __m256i nonce, const __m256i *midstate );
 
-#endif
+#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);
+// Expects data in little endian order, no byte swap needed
+void blake256_16way_update_le(void *cc, const void *data, size_t len);
+void blake256_16way_close_le(void *cc, void *dst);
+void blake256_16way_round0_prehash_le( void *midstate, const void *midhash,
+                                       void *data );
+void blake256_16way_final_rounds_le( void *final_hash, const void *midstate,
+                                     const void *midhash, const void *data );
+
+
+// 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 );
+void blake512_8way_full_le( blake_8way_big_context *sc, void * dst,
+                            const void *data, size_t len );
+void blake512_8way_prehash_le( blake_8way_big_context *sc, __m512i *midstate,
+                               const void *data );
+void blake512_8way_final_le( blake_8way_big_context *sc, void *hash,
+                             const __m512i nonce, const __m512i *midstate );
+
+#endif  // AVX512
+#endif  // AVX2
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif
-
-#endif
+#endif  // BLAKE_HASH_4WAY_H__

algo/blake/blake.c

@@ -39,8 +39,8 @@ void blakehash(void *state, const void *input)
 
 }
 
-int scanhash_blake( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+int scanhash_blake( 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;
@@ -49,6 +49,7 @@ int scanhash_blake( int thr_id, struct work *work, uint32_t max_nonce,
 	uint32_t _ALIGN(32) hash64[8];
 	uint32_t _ALIGN(32) endiandata[20];
 	uint32_t n = first_nonce;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	ctx_midstate_done = false;
 

algo/blake/blake2b-4way.c

@@ -0,0 +1,113 @@
+/**
+ * Blake2-B Implementation
+ * tpruvot@github 2015-2016
+ */
+
+#include "blake2b-gate.h"
+#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)
+{
+    blake2b_4way_ctx ctx;
+    blake2b_4way_init( &ctx );
+    blake2b_4way_update( &ctx, input, 80 );
+    blake2b_4way_final( &ctx, output );
+}
+
+int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
+                           uint64_t *hashes_done, struct thr_info *mythr )
+{
+	uint32_t hash[8*4] __attribute__ ((aligned (64)));;
+   uint32_t vdata[20*4] __attribute__ ((aligned (32)));;
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   blake2b_4way_ctx ctx __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[25]);   // 3*8+1
+	uint32_t *pdata = work->data;
+	uint32_t *ptarget = work->target;
+   int thr_id = mythr->id;
+   __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
+	const uint32_t Htarg = ptarget[7];
+	const uint32_t first_nonce = pdata[19];
+
+	uint32_t n = first_nonce;
+
+   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+
+	do {
+      *noncev = mm256_intrlv_blend_32( mm256_bswap_32(
+                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
+
+      blake2b_4way_init( &ctx ); 
+      blake2b_4way_update( &ctx, vdata, 80 );
+      blake2b_4way_final( &ctx, hash );
+
+      for ( int lane = 0; lane < 4; lane++ )
+      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_solution( work, lane_hash, mythr );
+          }
+      }
+      n += 4;
+   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
+
+   *hashes_done = n - first_nonce + 1;
+   return 0;
+}
+
+#endif

algo/blake/blake2b-gate.c

@@ -0,0 +1,20 @@
+#include "blake2b-gate.h"
+
+
+bool register_blake2b_algo( algo_gate_t* gate )
+{
+#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->optimizations =  AVX2_OPT | AVX512_OPT;
+  return true;
+};
+
+

algo/blake/blake2b-gate.h

@@ -0,0 +1,34 @@
+#ifndef __BLAKE2B_GATE_H__
+#define __BLAKE2B_GATE_H__ 1
+
+#include <stdint.h>
+#include "algo-gate-api.h"
+
+#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_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,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+#else
+
+void blake2b_hash( void *state, const void *input );
+int scanhash_blake2b( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
+
+#endif
+
+#endif

algo/blake/blake2b-hash-4way.c

@@ -0,0 +1,548 @@
+/*
+ * Copyright 2009 Colin Percival, 2014 savale
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * This file was originally written by Colin Percival as part of the Tarsnap
+ * online backup system.
+ */
+
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+
+#include "blake2b-hash-4way.h"
+
+#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 }
+};
+
+
+#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
+
+#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_xor3( ctx->h[0], v[0], v[ 8] );
+   ctx->h[1] = mm512_xor3( ctx->h[1], v[1], v[ 9] );
+   ctx->h[2] = mm512_xor3( ctx->h[2], v[2], v[10] );
+   ctx->h[3] = mm512_xor3( ctx->h[3], v[3], v[11] );
+   ctx->h[4] = mm512_xor3( ctx->h[4], v[4], v[12] );
+   ctx->h[5] = mm512_xor3( ctx->h[5], v[5], v[13] );
+   ctx->h[6] = mm512_xor3( ctx->h[6], v[6], v[14] );
+   ctx->h[7] = mm512_xor3( 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) \
+{ \
+   v[a] = _mm256_add_epi64( _mm256_add_epi64( v[a], v[b] ), x ); \
+	v[d] = mm256_swap64_32( _mm256_xor_si256( v[d], v[a] ) ); \
+	v[c] = _mm256_add_epi64( v[c], v[d] ); \
+	v[b] = mm256_shuflr64_24( _mm256_xor_si256( v[b], v[c] ) ); \
+	v[a] = _mm256_add_epi64( _mm256_add_epi64( v[a], v[b] ), y ); \
+	v[d] = mm256_shuflr64_16( _mm256_xor_si256( v[d], v[a] ) ); \
+	v[c] = _mm256_add_epi64( v[c], v[d] ); \
+	v[b] = mm256_ror_64( _mm256_xor_si256( v[b], v[c] ), 63 ); \
+}
+
+// Initialization Vector.
+/*
+static const uint64_t blake2b_iv[8] = {
+	0x6A09E667F3BCC908, 0xBB67AE8584CAA73B,
+	0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1,
+	0x510E527FADE682D1, 0x9B05688C2B3E6C1F,
+	0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179
+};
+*/
+
+static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
+{
+	__m256i 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] = m256_const1_64( 0x6A09E667F3BCC908 );
+   v[ 9] = m256_const1_64( 0xBB67AE8584CAA73B );
+   v[10] = m256_const1_64( 0x3C6EF372FE94F82B );
+   v[11] = m256_const1_64( 0xA54FF53A5F1D36F1 );
+   v[12] = m256_const1_64( 0x510E527FADE682D1 );
+   v[13] = m256_const1_64( 0x9B05688C2B3E6C1F );
+   v[14] = m256_const1_64( 0x1F83D9ABFB41BD6B );
+   v[15] = m256_const1_64( 0x5BE0CD19137E2179 );
+
+   v[12] = _mm256_xor_si256( v[12], _mm256_set1_epi64x( ctx->t[0] ) );
+   v[13] = _mm256_xor_si256( v[13], _mm256_set1_epi64x( ctx->t[1] ) );
+
+   if ( last )   
+		v[14] = mm256_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++ )
+   { 
+		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] ] );
+		B2B_G( 2, 6, 10, 14, m[ sigma[i][ 4] ], m[ sigma[i][ 5] ] );
+		B2B_G( 3, 7, 11, 15, m[ sigma[i][ 6] ], m[ sigma[i][ 7] ] );
+		B2B_G( 0, 5, 10, 15, m[ sigma[i][ 8] ], m[ sigma[i][ 9] ] );
+		B2B_G( 1, 6, 11, 12, m[ sigma[i][10] ], m[ sigma[i][11] ] );
+		B2B_G( 2, 7,  8, 13, m[ sigma[i][12] ], m[ sigma[i][13] ] );
+		B2B_G( 3, 4,  9, 14, m[ sigma[i][14] ], m[ sigma[i][15] ] );
+	}
+
+   ctx->h[0] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[0], v[0] ), v[ 8] );
+   ctx->h[1] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[1], v[1] ), v[ 9] );
+   ctx->h[2] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[2], v[2] ), v[10] );
+   ctx->h[3] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[3], v[3] ), v[11] );
+   ctx->h[4] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[4], v[4] ), v[12] );
+   ctx->h[5] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[5], v[5] ), v[13] );
+   ctx->h[6] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[6], v[6] ), v[14] );
+   ctx->h[7] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[7], v[7] ), v[15] );
+}
+
+int blake2b_4way_init( blake2b_4way_ctx *ctx ) 
+{
+	size_t i;
+
+   ctx->h[0] = m256_const1_64( 0x6A09E667F3BCC908 );
+   ctx->h[1] = m256_const1_64( 0xBB67AE8584CAA73B );
+   ctx->h[2] = m256_const1_64( 0x3C6EF372FE94F82B );
+   ctx->h[3] = m256_const1_64( 0xA54FF53A5F1D36F1 );
+   ctx->h[4] = m256_const1_64( 0x510E527FADE682D1 );
+   ctx->h[5] = m256_const1_64( 0x9B05688C2B3E6C1F );
+   ctx->h[6] = m256_const1_64( 0x1F83D9ABFB41BD6B );
+   ctx->h[7] = m256_const1_64( 0x5BE0CD19137E2179 );
+
+   ctx->h[0] = _mm256_xor_si256( ctx->h[0], m256_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] = m256_zero;
+
+	return 0;
+}
+
+void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+                          size_t inlen ) 
+{
+   __m256i* in =(__m256i*)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_4way_compress( ctx, 0 );
+			ctx->c = 0;
+		}
+      ctx->b[ c++ ] = in[i];
+      ctx->c += 8;
+   }
+}
+
+void blake2b_4way_final( blake2b_4way_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++] = m256_zero;
+      ctx->c += 8;
+   }
+
+   blake2b_4way_compress( ctx, 1 );           // final block flag = 1
+
+   casti_m256i( out, 0 ) = ctx->h[0];
+   casti_m256i( out, 1 ) = ctx->h[1];
+   casti_m256i( out, 2 ) = ctx->h[2];
+   casti_m256i( out, 3 ) = ctx->h[3];
+}
+
+#endif  // AVX2

algo/blake/blake2b-hash-4way.h

@@ -0,0 +1,53 @@
+#pragma once
+#ifndef __BLAKE2B_HASH_4WAY_H__
+#define __BLAKE2B_HASH_4WAY_H__
+
+#include "simd-utils.h"
+#include <stddef.h>
+#include <stdint.h>
+
+#if defined(_MSC_VER)
+#include <inttypes.h>
+#define inline __inline
+#define ALIGN(x) __declspec(align(x))
+#else
+#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
+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;
+
+int blake2b_4way_init( blake2b_4way_ctx *ctx );
+void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+                          size_t inlen );
+void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out );
+
+#endif
+
+#endif

algo/blake/blake2b.c

@@ -3,13 +3,14 @@
  * tpruvot@github 2015-2016
  */
 
-#include "algo-gate-api.h"
+#include "blake2b-gate.h"
+
+#if !defined(BLAKE2B_8WAY) && !defined(BLAKE2B_4WAY)
+
 #include <string.h>
 #include <stdint.h>
 #include "algo/blake/sph_blake2b.h"
 
-//static __thread sph_blake2b_ctx s_midstate;
-//static __thread sph_blake2b_ctx s_ctx;
 #define MIDLEN 76
 #define A 64
 
@@ -25,26 +26,17 @@ void blake2b_hash(void *output, const void *input)
 	memcpy(output, hash, 32);
 }
 
-/*
-static void blake2b_hash_end(uint32_t *output, const uint32_t *input)
-{
-	s_ctx.outlen = MIDLEN;
-	memcpy(&s_ctx, &s_midstate, 32 + 16 + MIDLEN);
-	sph_blake2b_update(&s_ctx, (uint8_t*) &input[MIDLEN/4], 80 - MIDLEN);
-	sph_blake2b_final(&s_ctx, (uint8_t*) output);
-}
-*/
-
-int scanhash_blake2b( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+int scanhash_blake2b( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
 {
 	uint32_t _ALIGN(A) vhashcpu[8];
 	uint32_t _ALIGN(A) endiandata[20];
 	uint32_t *pdata = work->data;
 	uint32_t *ptarget = work->target;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	const uint32_t Htarg = ptarget[7];
-	const uint32_t first_nonce = pdata[8];
+	const uint32_t first_nonce = pdata[19];
 
 	uint32_t n = first_nonce;
 
@@ -52,179 +44,21 @@ int scanhash_blake2b( int thr_id, struct work *work, uint32_t max_nonce,
 		be32enc(&endiandata[i], pdata[i]);
 	}
 
-	// midstate (untested yet)
-	//blake2b_init(&s_midstate, 32, NULL, 0);
-	//blake2b_update(&s_midstate, (uint8_t*) endiandata, MIDLEN);
-	//memcpy(&s_ctx, &s_midstate, sizeof(blake2b_ctx));
-
 	do {
-		be32enc(&endiandata[8], n);
-		//blake2b_hash_end(vhashcpu, endiandata);
+		be32enc(&endiandata[19], n);
 		blake2b_hash(vhashcpu, endiandata);
 
-		if (vhashcpu[7] < Htarg && fulltest(vhashcpu, ptarget)) {
-			work_set_target_ratio(work, vhashcpu);
-			*hashes_done = n - first_nonce + 1;
-			pdata[8] = n;
-			return 1;
-		}
-		n++;
-
+		if (vhashcpu[7] <= Htarg && fulltest(vhashcpu, ptarget))
+      {
+			pdata[19] = n;
+         submit_solution( work, vhashcpu, mythr );
+      }
+      n++;
 	} while (n < max_nonce && !work_restart[thr_id].restart);
 	*hashes_done = n - first_nonce + 1;
-	pdata[8] = n;
+	pdata[19] = n;
 
 	return 0;
 }
 
-static inline void swab256(void *dest_p, const void *src_p)
-{
-	uint32_t *dest = (uint32_t *)dest_p;
-	const uint32_t *src = (uint32_t *)src_p;
-
-	dest[0] = swab32(src[7]);
-	dest[1] = swab32(src[6]);
-	dest[2] = swab32(src[5]);
-	dest[3] = swab32(src[4]);
-	dest[4] = swab32(src[3]);
-	dest[5] = swab32(src[2]);
-	dest[6] = swab32(src[1]);
-	dest[7] = swab32(src[0]);
-}
-
-/* compute nbits to get the network diff */
-void blake2b_calc_network_diff(struct work *work)
-{
-        // sample for diff 43.281 : 1c05ea29
-        uint32_t nbits = work->data[11]; // unsure if correct
-        uint32_t bits = (nbits & 0xffffff);
-        int16_t shift = (swab32(nbits) & 0xff); // 0x1c = 28
-
-        double d = (double)0x0000ffff / (double)bits;
-        for (int m=shift; m < 29; m++) d *= 256.0;
-        for (int m=29; m < shift; m++) d /= 256.0;
-        if (opt_debug_diff)
-                applog(LOG_DEBUG, "net diff: %f -> shift %u, bits %08x", d, shift, bits);
-        net_diff = d;
-}
-
-void blake2b_be_build_stratum_request( char *req, struct work *work )
-{
-   unsigned char *xnonce2str;
-   uint32_t ntime,       nonce;
-   char     ntimestr[9], noncestr[9];
-   be32enc( &ntime, work->data[ algo_gate.ntime_index ] );
-   be32enc( &nonce, work->data[ algo_gate.nonce_index ] );
-   bin2hex( ntimestr, (char*)(&ntime), sizeof(uint32_t) );
-   bin2hex( noncestr, (char*)(&nonce), sizeof(uint32_t) );
-   uint16_t high_nonce = swab32(work->data[9]) >> 16;
-   xnonce2str = abin2hex((unsigned char*)(&high_nonce), 2);
-   snprintf( req, JSON_BUF_LEN,
-        "{\"method\": \"mining.submit\", \"params\": [\"%s\", \"%s\", \"%s\", \"%s\", \"%s\"], \"id\":4}",
-         rpc_user, work->job_id, xnonce2str, ntimestr, noncestr );
-   free( xnonce2str );
-}
-
-#define min(a,b) (a>b ? (b) :(a))
-
-// merkle root handled here, no need for gen_merkle_root gate target
-void blake2b_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
-{
-    uchar merkle_root[64] = { 0 };
-    uint32_t extraheader[32] = { 0 };
-    int headersize = 0;
-    size_t t;
-    int i;
-
-    // merkle root
-    memcpy( merkle_root, sctx->job.coinbase, 32 );
-    headersize = min( (int)sctx->job.coinbase_size - 32, 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 );
-//    for ( i = 0; i < 8; i++ )
-//       g_work->data[1 + i] = le32dec( (uint32_t *) sctx->job.prevhash + i );
-    for ( i = 0; i < 8; i++ )
-       g_work->data[i] = ((uint32_t*)sctx->job.prevhash)[7-i];
-//    for ( i = 0; i < 8; i++ )
-//       g_work->data[9 + i] = be32dec( (uint32_t *) merkle_root + i );
-    g_work->data[8]  = 0; // nonce
-    g_work->data[9]  = swab32( extraheader[0] ) | ( rand() & 0xf0 );
-    g_work->data[10] = be32dec( sctx->job.ntime );
-    g_work->data[11] = be32dec( sctx->job.nbits );
-    for ( i = 0; i < 8; i++ )
-       g_work->data[12+i] = ( (uint32_t*)merkle_root )[i];
-}
-
-#undef min
-
-void blake2b_get_new_work( struct work* work, struct work* g_work, int thr_id,
-                           uint32_t* end_nonce_ptr, bool clean_job )
-{
-   const int wkcmp_sz = 32;  // bytes
-   const int wkcmp_off = 32 + 16; 
-   uint32_t *nonceptr = algo_gate.get_nonceptr( work->data );
-
-   if ( memcmp( &work->data[ wkcmp_off ], &g_work->data[ wkcmp_off ], wkcmp_sz )
-      && ( clean_job || ( *nonceptr >= *end_nonce_ptr ) 
-      || strcmp( work->job_id, g_work->job_id ) ) )
-   {
-      work_free( work );
-      work_copy( work, g_work );
-      *nonceptr = ( 0xffffffffU / opt_n_threads ) * thr_id;
-      if ( opt_randomize )
-         *nonceptr += ( (rand() *4 ) & UINT32_MAX ) / opt_n_threads;
-      *end_nonce_ptr = ( 0xffffffffU / opt_n_threads ) * (thr_id+1) - 0x20;
-   }
-   else
-       ++(*nonceptr);
-
-   // suprnova job_id check without data/target/height change...
-   // we just may have copied new g_wwork to work so why this test here?
-//   if (  have_stratum && strcmp( work->job_id, g_work->job_id ) )
-      // exit thread loop
-//      continue;
-//   else
-//   {
-//      nonceptr[1] += 0x10;
-//      nonceptr[1] |= thr_id;
-//   }
-}
-
-bool blake2b_ready_to_mine( struct work* work, struct stratum_ctx* stratum,
-                           int thr_id )
-{
-   if ( have_stratum && strcmp( stratum->job.job_id, work->job_id ) )
-      // need to regen g_work..
-      return false;
-   // extradata: prevent duplicates
-   work->data[ 8     ] += 0x10;
-   work->data[ 8 + 1 ] |= thr_id;
-   return true;
-}
-
-double blake2b_get_max64() { return 0x1fffffLL; }
-
-bool register_blake2b_algo( algo_gate_t* gate )
-{
-  algo_not_tested();
-  gate->ntime_index   = 10;
-  gate->nbits_index   = 11;
-  gate->nonce_index   =  8;
-  gate->work_cmp_size = 32;
-  gate->scanhash              = (void*)&scanhash_blake2b;
-  gate->hash                  = (void*)&blake2b_hash;
-  gate->calc_network_diff     = (void*)&blake2b_calc_network_diff;
-  gate->build_stratum_request = (void*)&blake2b_be_build_stratum_request;
-  gate->work_decode           = (void*)&std_be_work_decode;
-  gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
-  gate->build_extraheader     = (void*)&blake2b_build_extraheader;
-  gate->get_new_work          = (void*)&blake2b_get_new_work;
-  gate->get_max64             = (void*)&blake2b_get_max64;
-  gate->ready_to_mine         = (void*)&blake2b_ready_to_mine;
-  have_gbt = false;
-  return true;
-}
+#endif

algo/blake/blake2s-4way.c

@@ -3,73 +3,115 @@
 #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 );
-
-   mm256_deinterleave_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( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
 {
    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;
-   uint32_t _ALIGN(64) edata[20];
    const uint32_t Htarg = ptarget[7];
    const uint32_t first_nonce = pdata[19];
+   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
-   uint32_t *noncep = vdata + 152;   // 19*8
+   int thr_id = mythr->id; 
 
-   swab32_array( edata, pdata, 20 );
-   mm256_interleave_8x32( vdata, edata, edata, edata, edata,
-                                 edata, edata, edata, edata, 640 );
+   mm256_bswap32_intrlv80_8x32( vdata, pdata );
    blake2s_8way_init( &blake2s_8w_ctx, BLAKE2S_OUTBYTES );
    blake2s_8way_update( &blake2s_8w_ctx, vdata, 64 );
 
    do {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
-      be32enc( noncep +4, n+4 );
-      be32enc( noncep +5, n+5 );
-      be32enc( noncep +6, n+6 );
-      be32enc( noncep +7, n+7 );
+      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
+                                                  n+3, n+2, n+1, n ) );
       pdata[19] = n;
 
       blake2s_8way_hash( hash, vdata );
 
-
-      for ( int i = 0; i < 8; i++ )
-      if (  (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget ) )
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[lane] <= Htarg ) )
       {
-          pdata[19] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+         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 ( (num_found == 0) && (n < max_nonce)
-             && !work_restart[thr_id].restart );
+   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
 
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #elif defined(BLAKE2S_4WAY)
@@ -78,59 +120,51 @@ 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 );
-
-   mm_deinterleave_4x32( output, output+32, output+64, output+96, vhash, 256 );
+   blake2s_4way_final( &ctx, output, BLAKE2S_OUTBYTES );
 }
 
-int scanhash_blake2s_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done )
+int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr )
 {
    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;
-   uint32_t _ALIGN(64) edata[20];
    const uint32_t Htarg = ptarget[7];
    const uint32_t first_nonce = pdata[19];
+   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
-   uint32_t *noncep = vdata + 76;   // 19*4
+   int thr_id = mythr->id; 
 
-   swab32_array( edata, pdata, 20 );
-   mm_interleave_4x32( vdata, edata, edata, edata, edata, 640 );
+   mm128_bswap32_intrlv80_4x32( vdata, pdata );
    blake2s_4way_init( &blake2s_4w_ctx, BLAKE2S_OUTBYTES );
    blake2s_4way_update( &blake2s_4w_ctx, vdata, 64 );
 
    do {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
+      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
       pdata[19] = n;
 
       blake2s_4way_hash( hash, vdata );
 
-      for ( int i = 0; i < 4; i++ )
-      if ( (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget ) )
+      for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
       {
-          pdata[19] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+         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 ( (num_found == 0) && (n < max_nonce)
-             && !work_restart[thr_id].restart );
+   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
 
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #endif

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 = SSE42_OPT | AVX2_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
   return true;
 };
 

algo/blake/blake2s-gate.h

@@ -4,31 +4,42 @@
 #include <stdint.h>
 #include "algo-gate-api.h"
 
-#if defined(__SSE4_2__)
+#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)
 
 void blake2s_8way_hash( void *state, const void *input );
-int scanhash_blake2s_8way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
 
 #elif defined (BLAKE2S_4WAY)
 
 void blake2s_4way_hash( void *state, const void *input );
-int scanhash_blake2s_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
 #else
 
 void blake2s_hash( void *state, const void *input );
-int scanhash_blake2s( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done );
+int scanhash_blake2s( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
 
 #endif
 

algo/blake/blake2s-hash-4way.c

@@ -17,13 +17,16 @@
 #include <string.h>
 #include <stdio.h>
 
-#if defined(__SSE4_2__)
+//#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] =
 {
@@ -39,6 +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 )
@@ -57,8 +61,18 @@ int blake2s_4way_init( blake2s_4way_state *S, const uint8_t outlen )
    memset( P->personal, 0, sizeof( P->personal ) );
 
    memset( S, 0, sizeof( blake2s_4way_state ) );
-   for( int i = 0; i < 8; ++i )
-      S->h[i] = _mm_set1_epi32( blake2s_IV[i] );
+
+   S->h[0] = m128_const1_64( 0x6A09E6676A09E667ULL );
+   S->h[1] = m128_const1_64( 0xBB67AE85BB67AE85ULL );
+   S->h[2] = m128_const1_64( 0x3C6EF3723C6EF372ULL );
+   S->h[3] = m128_const1_64( 0xA54FF53AA54FF53AULL );
+   S->h[4] = m128_const1_64( 0x510E527F510E527FULL );
+   S->h[5] = m128_const1_64( 0x9B05688C9B05688CULL );
+   S->h[6] = m128_const1_64( 0x1F83D9AB1F83D9ABULL );
+   S->h[7] = m128_const1_64( 0x5BE0CD195BE0CD19ULL );
+   
+//   for( int i = 0; i < 8; ++i )
+//      S->h[i] = _mm_set1_epi32( blake2s_IV[i] );
 
    uint32_t *p = ( uint32_t * )( P );
 
@@ -76,41 +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] ] ); \
-   d = mm_ror_32( _mm_xor_si128( d, a ), 16 ); \
+   uint8_t s0 = sigma0; \
+   uint8_t s1 = sigma1; \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s0 ] ); \
+   d = mm128_swap32_16( _mm_xor_si128( d, a ) ); \
    c = _mm_add_epi32( c, d ); \
-   b = mm_ror_32( _mm_xor_si128( b, c ), 12 ); \
-   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ blake2s_sigma[r][2*i+1] ] ); \
-   d = mm_ror_32( _mm_xor_si128( d, a ),  8 ); \
+   b = mm128_ror_32( _mm_xor_si128( b, c ), 12 ); \
+   a = _mm_add_epi32( _mm_add_epi32( a, b ), m[ s1 ] ); \
+   d = mm128_shuflr32_8( _mm_xor_si128( d, a ) ); \
    c = _mm_add_epi32( c, d ); \
-   b = mm_ror_32( _mm_xor_si128( b, c ),  7 ); \
+   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 );
@@ -132,26 +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
       {
@@ -183,8 +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];
@@ -193,6 +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] );
@@ -206,6 +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 ), \
@@ -219,7 +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_swap32_16( _mm256_xor_si256( d, a ) ); \
+   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_shuflr32_8( _mm256_xor_si256( d, a ) ); \
+   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] ); \
@@ -231,6 +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 );
@@ -244,7 +368,7 @@ do { \
    ROUND8W( 9 );
 
    for( size_t i = 0; i < 8; ++i )
-      S->h[i] = _mm256_xor_si256( _mm256_xor_si256( S->h[i], v[i] ), v[i + 8] );
+      S->h[i] = mm256_xor3( S->h[i], v[i], v[i + 8] );
 
 #undef G8W
 #undef ROUND8W
@@ -267,8 +391,18 @@ int blake2s_8way_init( blake2s_8way_state *S, const uint8_t outlen )
    memset( P->personal, 0, sizeof( P->personal ) );
 
    memset( S, 0, sizeof( blake2s_8way_state ) );
-   for( int i = 0; i < 8; ++i )
-      S->h[i] = _mm256_set1_epi32( blake2s_IV[i] );
+   S->h[0] = m256_const1_64( 0x6A09E6676A09E667ULL );
+   S->h[1] = m256_const1_64( 0xBB67AE85BB67AE85ULL );
+   S->h[2] = m256_const1_64( 0x3C6EF3723C6EF372ULL );
+   S->h[3] = m256_const1_64( 0xA54FF53AA54FF53AULL );
+   S->h[4] = m256_const1_64( 0x510E527F510E527FULL );
+   S->h[5] = m256_const1_64( 0x9B05688C9B05688CULL );
+   S->h[6] = m256_const1_64( 0x1F83D9AB1F83D9ABULL );
+   S->h[7] = m256_const1_64( 0x5BE0CD195BE0CD19ULL );
+
+
+//   for( int i = 0; i < 8; ++i )
+//      S->h[i] = _mm256_set1_epi32( blake2s_IV[i] );
 
    uint32_t *p = ( uint32_t * )( P );
 
@@ -329,9 +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_xor3( 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 )
 {

algo/blake/blake2s-hash-4way.h

@@ -14,9 +14,9 @@
 #ifndef __BLAKE2S_HASH_4WAY_H__
 #define __BLAKE2S_HASH_4WAY_H__ 1
 
-#if defined(__SSE4_2__)
+#if defined(__SSE2__)
 
-#include "avxdefs.h"
+#include "simd-utils.h"
 
 #include <stddef.h>
 #include <stdint.h>
@@ -60,7 +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 ];
@@ -74,10 +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 ];
@@ -91,6 +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
 
@@ -107,6 +131,6 @@ int blake2s_8way_final( blake2s_8way_state *S, void *out, uint8_t outlen );
 }
 #endif
 
-#endif  // __SSE4_2__
+#endif  // __SSE2__
 
 #endif

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>
 
@@ -32,14 +34,15 @@ static void blake2s_hash_end(uint32_t *output, const uint32_t *input)
 	blake2s_final(&s_ctx, (uint8_t*) output, BLAKE2S_OUTBYTES);
 }
 */
-int scanhash_blake2s(int thr_id, struct work *work,
-	uint32_t max_nonce, uint64_t *hashes_done)
+int scanhash_blake2s( 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 _ALIGN(64) hash64[8];
 	uint32_t _ALIGN(64) endiandata[20];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	const uint32_t Htarg = ptarget[7];
 	const uint32_t first_nonce = pdata[19];
@@ -55,7 +58,7 @@ int scanhash_blake2s(int thr_id, 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;
@@ -69,18 +72,4 @@ int scanhash_blake2s(int thr_id, 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

algo/blake/blakecoin-4way.c

@@ -14,14 +14,14 @@ 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 );
 
-     mm_deinterleave_4x32( state, state+32, state+64, state+96, vhash, 256 );
+     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
 }
 
-int scanhash_blakecoin_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done )
+int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*4] __attribute__ ((aligned (64)));
    uint32_t hash[8*4] __attribute__ ((aligned (32)));
@@ -29,41 +29,34 @@ int scanhash_blakecoin_4way( int thr_id, struct work *work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t HTarget = ptarget[7];
-   uint32_t _ALIGN(32) edata[20];
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
+   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
+   int thr_id = mythr->id;  // thr_id arg is deprecated
    if ( opt_benchmark )
       HTarget = 0x7f;
 
-   swab32_array( edata, pdata, 20 );
-   mm_interleave_4x32( vdata, edata, edata, edata, edata, 640 );
+   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 );
 
-   uint32_t *noncep = vdata + 76;   // 19*4
    do {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
+      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
       pdata[19] = n;
       blakecoin_4way_hash( hash, vdata );
 
       for ( int i = 0; i < 4; i++ )
-      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget ) )
+      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget )
+           && !opt_benchmark )
       {
           pdata[19] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+          submit_solution( work, hash+(i<<3), mythr );
       }
       n += 4;
 
-   } while ( (num_found == 0) && (n < max_nonce) 
-             && !work_restart[thr_id].restart );
+   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
 
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #endif
@@ -78,16 +71,15 @@ 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 );
 
-     mm256_deinterleave_8x32( state,     state+ 32, state+ 64, state+ 96,
-                              state+128, state+160, state+192, state+224,
-                              vhash, 256 );
+     dintrlv_8x32( state,     state+ 32, state+ 64, state+ 96, state+128,
+                   state+160, state+192, state+224, vhash, 256 );
 }
 
-int scanhash_blakecoin_8way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done )
+int scanhash_blakecoin_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[20*8] __attribute__ ((aligned (64)));
    uint32_t hash[8*8] __attribute__ ((aligned (32)));
@@ -95,46 +87,34 @@ int scanhash_blakecoin_8way( int thr_id, struct work *work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    const uint32_t first_nonce = pdata[19];
    uint32_t HTarget = ptarget[7];
-   uint32_t _ALIGN(32) edata[20];
    uint32_t n = first_nonce;
-   uint32_t *nonces = work->nonces;
-   uint32_t *noncep = vdata + 152;   // 19*8
-   int num_found = 0;
+   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
+   int thr_id = mythr->id;  // thr_id arg is deprecated
    if ( opt_benchmark )
       HTarget = 0x7f;
 
-   // we need big endian data...
-   swab32_array( edata, pdata, 20 );
-   mm256_interleave_8x32( vdata, edata, edata, edata, edata,
-                                 edata, edata, edata, edata, 640 );
+   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 {
-      be32enc( noncep,    n   );
-      be32enc( noncep +1, n+1 );
-      be32enc( noncep +2, n+2 );
-      be32enc( noncep +3, n+3 );
-      be32enc( noncep +4, n+4 );
-      be32enc( noncep +5, n+5 );
-      be32enc( noncep +6, n+6 );
-      be32enc( noncep +7, n+7 );
+      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
+                                                  n+3, n+2, n+1, n ) );
       pdata[19] = n;
       blakecoin_8way_hash( hash, vdata );
 
       for ( int i = 0; i < 8; i++ )
-      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget ) )
+      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget )
+          && !opt_benchmark )
       {
           pdata[19] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+          submit_solution( work, hash+(i<<3), mythr );
       }
       n += 8;
-   } while ( (num_found == 0) && (n < max_nonce)
-             && !work_restart[thr_id].restart );
+   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
 
    *hashes_done = n - first_nonce + 1;
-   return num_found;
+   return 0;
 }
 
 #endif

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;
 }
 

algo/blake/blakecoin-gate.h

@@ -13,18 +13,18 @@
 
 #if defined (BLAKECOIN_8WAY)
 void blakecoin_8way_hash(void *state, const void *input);
-int scanhash_blakecoin_8way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_blakecoin_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 
 #if defined (BLAKECOIN_4WAY)
 void blakecoin_4way_hash(void *state, const void *input);
-int scanhash_blakecoin_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 
 void blakecoinhash( void *state, const void *input );
-int scanhash_blakecoin( int thr_id, struct work *work, uint32_t max_nonce,
-                      uint64_t *hashes_done );
+int scanhash_blakecoin( struct work *work, uint32_t max_nonce,
+                      uint64_t *hashes_done, struct thr_info *mythr );
 
 #endif

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"
 
@@ -39,13 +42,14 @@ void blakecoinhash( void *state, const void *input )
 	memcpy( state, hash, 32 );
 }
 
-int scanhash_blakecoin( int thr_id, struct work *work, uint32_t max_nonce,
-                        uint64_t *hashes_done )
+int scanhash_blakecoin( 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;
 	const uint32_t first_nonce = pdata[19];
 	uint32_t HTarget = ptarget[7];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	uint32_t _ALIGN(32) hash64[8];
 	uint32_t _ALIGN(32) endiandata[20];
@@ -92,33 +96,4 @@ int scanhash_blakecoin( int thr_id, 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

algo/blake/decred-4way.c

@@ -21,13 +21,13 @@ 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 );
-     mm_deinterleave_4x32( state, state+32, state+64, state+96, vhash, 256 );
+     dintrlv_4x32( state, state+32, state+64, state+96, vhash, 256 );
 }
 
-int scanhash_decred_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                          uint64_t *hashes_done)
+int scanhash_decred_4way( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t vdata[48*4] __attribute__ ((aligned (64)));
    uint32_t hash[8*4] __attribute__ ((aligned (32)));
@@ -37,17 +37,16 @@ int scanhash_decred_4way( int thr_id, struct work *work, uint32_t max_nonce,
    const uint32_t first_nonce = pdata[DECRED_NONCE_INDEX];
    uint32_t n = first_nonce;
    const uint32_t HTarget = opt_benchmark ? 0x7f : ptarget[7];
-   uint32_t *nonces = work->nonces;
-   int num_found = 0;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
    // copy to buffer guaranteed to be aligned.
    memcpy( edata, pdata, 180 );
 
    // use the old way until  new way updated for size.
-   mm_interleave_4x32x( vdata, edata, edata, edata, edata, 180*8 );
+   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 {
@@ -59,18 +58,17 @@ int scanhash_decred_4way( int thr_id, struct work *work, uint32_t max_nonce,
       decred_hash_4way( hash, vdata );
 
       for ( int i = 0; i < 4; i++ )
-      if (  (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget ) )
+      if (  (hash+(i<<3))[7] <= HTarget )
+      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
       {
           pdata[DECRED_NONCE_INDEX] = n+i;
-          nonces[ num_found++ ] = n+i;
-          work_set_target_ratio( work, hash+(i<<3) );
+          submit_solution( work, hash+(i<<3), mythr );
       }
       n += 4;
-  } while ( (num_found == 0) && (n < max_nonce) 
-            && !work_restart[thr_id].restart );
+  } while ( (n < max_nonce) && !work_restart[thr_id].restart );
 
   *hashes_done = n - first_nonce + 1;
-  return num_found;
+  return 0;
 }
 
 #endif

algo/blake/decred-gate.c

@@ -8,7 +8,7 @@ uint32_t *decred_get_nonceptr( uint32_t *work_data )
    return &work_data[ DECRED_NONCE_INDEX ];
 }
 
-double decred_calc_network_diff( struct work* work )
+long double decred_calc_network_diff( struct work* work )
 {
    // sample for diff 43.281 : 1c05ea29
    // todo: endian reversed on longpoll could be zr5 specific...
@@ -16,7 +16,7 @@ double decred_calc_network_diff( struct work* work )
    uint32_t bits = ( nbits & 0xffffff );
    int16_t shift = ( swab32(nbits) & 0xff ); // 0x1c = 28
    int m;
-   double d = (double)0x0000ffff / (double)bits;
+   long double d = (long double)0x0000ffff / (long double)bits;
 
    for ( m = shift; m < 29; m++ )
        d *= 256.0;
@@ -25,7 +25,7 @@ double decred_calc_network_diff( struct work* work )
    if ( shift == 28 )
        d *= 256.0; // testnet
    if ( opt_debug_diff )
-       applog( LOG_DEBUG, "net diff: %f -> shift %u, bits %08x", d,
+       applog( LOG_DEBUG, "net diff: %f -> shift %u, bits %08x", (double)d,
                            shift, bits );
    return net_diff;
 }
@@ -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,
@@ -70,7 +70,10 @@ void decred_be_build_stratum_request( char *req, struct work *work,
          rpc_user, work->job_id, xnonce2str, ntimestr, noncestr );
    free(xnonce2str);
 }
+
+#if !defined(min)
 #define min(a,b) (a>b ? (b) :(a))
+#endif
 
 void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
 {
@@ -78,7 +81,6 @@ 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 +89,6 @@ 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 +115,7 @@ 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);
 }
@@ -140,6 +139,7 @@ bool decred_ready_to_mine( struct work* work, struct stratum_ctx* stratum,
    return true;
 }
 
+int decred_get_work_data_size() { return DECRED_DATA_SIZE; }
 
 bool register_decred_algo( algo_gate_t* gate )
 {
@@ -152,9 +152,8 @@ 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->display_extra_data    = (void*)&decred_decode_extradata;
+//  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;
   gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
@@ -163,7 +162,7 @@ bool register_decred_algo( algo_gate_t* gate )
   gate->nbits_index           = DECRED_NBITS_INDEX;
   gate->ntime_index           = DECRED_NTIME_INDEX;
   gate->nonce_index           = DECRED_NONCE_INDEX;
-  gate->work_data_size        = DECRED_DATA_SIZE;
+  gate->get_work_data_size    = (void*)&decred_get_work_data_size;
   gate->work_cmp_size         = DECRED_WORK_COMPARE_SIZE;
   allow_mininginfo            = false;
   have_gbt                    = false;

algo/blake/decred-gate.h

@@ -14,7 +14,7 @@
 
 #if defined (__AVX2__) 
 //void blakehash_84way(void *state, const void *input);
-//int scanhash_blake_8way( int thr_id, struct work *work, uint32_t max_nonce,
+//int scanhash_blake_8way( struct work *work, uint32_t max_nonce,
 //                         uint64_t *hashes_done );
 #endif
 
@@ -24,13 +24,13 @@
 
 #if defined (DECRED_4WAY)
 void decred_hash_4way(void *state, const void *input);
-int scanhash_decred_4way( int thr_id, struct work *work, uint32_t max_nonce,
-                          uint64_t *hashes_done );
+int scanhash_decred_4way( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 
 void decred_hash( void *state, const void *input );
-int scanhash_decred( int thr_id, struct work *work, uint32_t max_nonce,
-                     uint64_t *hashes_done );
+int scanhash_decred( struct work *work, uint32_t max_nonce,
+                     uint64_t *hashes_done, struct thr_info *mythr );
 
 #endif
 

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>
@@ -52,12 +55,14 @@ void decred_hash_simple(void *state, const void *input)
         sph_blake256_close(&ctx, state);
 }
 
-int scanhash_decred(int thr_id, struct work *work, uint32_t max_nonce, uint64_t *hashes_done)
+int scanhash_decred( struct work *work, uint32_t max_nonce,
+               uint64_t *hashes_done, struct thr_info *mythr )
 {
         uint32_t _ALIGN(64) endiandata[48];
         uint32_t _ALIGN(64) hash32[8];
         uint32_t *pdata = work->data;
         uint32_t *ptarget = work->target;
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 //        #define DCR_NONCE_OFT32 35
 
@@ -75,25 +80,15 @@ int scanhash_decred(int thr_id, struct work *work, uint32_t max_nonce, uint64_t
                 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++;
@@ -141,7 +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,
@@ -267,8 +262,7 @@ 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->display_extra_data    = (void*)&decred_decode_extradata;
+  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;
   gate->submit_getwork_result = (void*)&std_be_submit_getwork_result;
@@ -284,3 +278,5 @@ bool register_decred_algo( algo_gate_t* gate )
   return true;
 }
 */
+
+#endif

algo/blake/pentablake-4way.c

@@ -1,6 +1,6 @@
 #include "pentablake-gate.h"
 
-#if defined (__AVX2__)
+#if defined(PENTABLAKE_4WAY)
 
 #include <stdlib.h>
 #include <stdint.h>
@@ -10,13 +10,8 @@
 #include "blake-hash-4way.h"
 #include "sph_blake.h"
 
-//#define DEBUG_ALGO
-
 extern void pentablakehash_4way( void *output, const void *input )
 {
-	unsigned char _ALIGN(32) hash[128];
-//	// same as uint32_t hashA[16], hashB[16];
-//	#define hashB hash+64
 
      uint64_t hash0[8] __attribute__ ((aligned (64)));
      uint64_t hash1[8] __attribute__ ((aligned (64)));
@@ -27,80 +22,33 @@ extern void pentablakehash_4way( void *output, const void *input )
 
 
      blake512_4way_init( &ctx );
-     blake512_4way( &ctx, input, 80 );
-     blake512_4way_close( &ctx, vhash );
-
-uint64_t sin0[10], sin1[10], sin2[10], sin3[10];
-mm256_deinterleave_4x64( sin0, sin1, sin2, sin3, input, 640 );
-sph_blake512_context ctx2_blake;
-sph_blake512_init(&ctx2_blake);
-sph_blake512(&ctx2_blake, sin0, 80);
-sph_blake512_close(&ctx2_blake, (void*) hash);
-
-mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
-uint64_t* hash64 = (uint64_t*)hash;
-for( int i = 0; i < 8; i++ )
-{
-   if ( hash0[i] != hash64[i] )
-      printf("hash mismatch %u\n",i);
-}
-
-     blake512_4way_init( &ctx );
-     blake512_4way( &ctx, vhash, 64 );
+     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_update( &ctx, vhash, 64 );
      blake512_4way_close( &ctx, vhash );
 
-     mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
      memcpy( output,    hash0, 32 );
      memcpy( output+32, hash1, 32 );
      memcpy( output+64, hash2, 32 );
      memcpy( output+96, hash3, 32 );
-
-/*
-     uint64_t sin0[10] __attribute__ ((aligned (64)));
-     uint64_t sin1[10] __attribute__ ((aligned (64)));
-     uint64_t sin2[10] __attribute__ ((aligned (64)));
-     uint64_t sin3[10] __attribute__ ((aligned (64)));
-
-	sph_blake512_context     ctx_blake;
-
-	sph_blake512_init(&ctx_blake);
-	sph_blake512(&ctx_blake, input, 80);
-	sph_blake512_close(&ctx_blake, hash);
-
-        sph_blake512_init(&ctx_blake);
-	sph_blake512(&ctx_blake, hash, 64);
-	sph_blake512_close(&ctx_blake, hash);
-
-        sph_blake512_init(&ctx_blake);
-	sph_blake512(&ctx_blake, hash, 64);
-	sph_blake512_close(&ctx_blake, hash);
-
-        sph_blake512_init(&ctx_blake);
-	sph_blake512(&ctx_blake, hash, 64);
-	sph_blake512_close(&ctx_blake, hash);
-
-        sph_blake512_init(&ctx_blake);
-	sph_blake512(&ctx_blake, hash, 64);
-	sph_blake512_close(&ctx_blake, hash);
-
-	memcpy(output, hash, 32);
-*/
 }
 
-int scanhash_pentablake_4way( int thr_id, struct work *work,
-                              uint32_t max_nonce, uint64_t *hashes_done )
+int scanhash_pentablake_4way( struct work *work,
+      uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr )
 {
     uint32_t hash[4*8] __attribute__ ((aligned (64)));
     uint32_t vdata[20*4] __attribute__ ((aligned (64)));
@@ -110,9 +58,8 @@ int scanhash_pentablake_4way( int thr_id, struct work *work,
     uint32_t n = pdata[19] - 1;
     const uint32_t first_nonce = pdata[19];
     const uint32_t Htarg = ptarget[7];
-    uint32_t *nonces = work->nonces;
-    int num_found = 0;
     uint32_t *noncep = vdata + 73;   // 9*8 + 1
+    int thr_id = mythr->id;  // thr_id arg is deprecated
 
 //    uint32_t _ALIGN(32) hash64[8];
 //    uint32_t _ALIGN(32) endiandata[32];
@@ -138,7 +85,7 @@ int scanhash_pentablake_4way( int thr_id, struct work *work,
     swab32_array( endiandata, pdata, 20 );
 
     uint64_t *edata = (uint64_t*)endiandata;
-    mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
+    intrlv_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
 
     for ( int m=0; m < 6; m++ )
     {
@@ -155,10 +102,10 @@ int scanhash_pentablake_4way( int thr_id, struct work *work,
 
               for ( int i = 0; i < 4; i++ )
               if ( !( (hash+(i<<3))[7] & mask )
-                  && fulltest( hash+(i<<3), ptarget ) )
+                  && fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
               {
-                 nonces[ num_found++ ] = n+i;
-                 work_set_target_ratio( work, hash+(i<<3) );
+                 pdata[19] = n + i;
+                 submit_solution( work, hash+(i<<3), mythr );
               }
               n += 4;
 

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;
 };
 

algo/blake/pentablake-gate.h

@@ -4,18 +4,19 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX2__)
-  #define PENTABLAKE_4WAY
-#endif
+// 4way is broken
+//#if defined(__AVX2__)
+//  #define PENTABLAKE_4WAY
+//#endif
 
 #if defined(PENTABLAKE_4WAY)
 void pentablakehash_4way( void *state, const void *input );
-int scanhash_pentablake_4way( int thr_id, struct work *work,
-                              uint32_t max_nonce, uint64_t *hashes_done );
+int scanhash_pentablake_4way( struct work *work,
+           uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 
 void pentablakehash( void *state, const void *input );
-int scanhash_pentablake( int thr_id, struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done );
+int scanhash_pentablake( struct work *work, uint32_t max_nonce,
+            uint64_t *hashes_done, struct thr_info *mythr );
 #endif
 

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>
@@ -40,8 +43,8 @@ extern void pentablakehash(void *output, const void *input)
 
 }
 
-int scanhash_pentablake(int thr_id, struct work *work, uint32_t max_nonce,
-      uint64_t *hashes_done)
+int scanhash_pentablake( 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;
@@ -49,6 +52,7 @@ int scanhash_pentablake(int thr_id, struct work *work, uint32_t max_nonce,
 	uint32_t n = pdata[19] - 1;
 	const uint32_t first_nonce = pdata[19];
 	const uint32_t Htarg = ptarget[7];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
 
 	uint32_t _ALIGN(32) hash64[8];
 	uint32_t _ALIGN(32) endiandata[32];
@@ -110,3 +114,4 @@ int scanhash_pentablake(int thr_id, struct work *work, uint32_t max_nonce,
 	return 0;
 } 
 
+#endif

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;
 }
 

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];

algo/blake/sph_blake.c

@@ -630,6 +630,69 @@ static const sph_u64 CB[16] = {
 		H7 ^= S3 ^ V7 ^ VF; \
 	} while (0)
 
+#define COMPRESS32_LE   do { \
+      sph_u32 M0, M1, M2, M3, M4, M5, M6, M7; \
+      sph_u32 M8, M9, MA, MB, MC, MD, ME, MF; \
+      sph_u32 V0, V1, V2, V3, V4, V5, V6, V7; \
+      sph_u32 V8, V9, VA, VB, VC, VD, VE, VF; \
+      V0 = H0; \
+      V1 = H1; \
+      V2 = H2; \
+      V3 = H3; \
+      V4 = H4; \
+      V5 = H5; \
+      V6 = H6; \
+      V7 = H7; \
+      V8 = S0 ^ CS0; \
+      V9 = S1 ^ CS1; \
+      VA = S2 ^ CS2; \
+      VB = S3 ^ CS3; \
+      VC = T0 ^ CS4; \
+      VD = T0 ^ CS5; \
+      VE = T1 ^ CS6; \
+      VF = T1 ^ CS7; \
+      M0 = *((uint32_t*)(buf +  0)); \
+      M1 = *((uint32_t*)(buf +  4)); \
+      M2 = *((uint32_t*)(buf +  8)); \
+      M3 = *((uint32_t*)(buf + 12)); \
+      M4 = *((uint32_t*)(buf + 16)); \
+      M5 = *((uint32_t*)(buf + 20)); \
+      M6 = *((uint32_t*)(buf + 24)); \
+      M7 = *((uint32_t*)(buf + 28)); \
+      M8 = *((uint32_t*)(buf + 32)); \
+      M9 = *((uint32_t*)(buf + 36)); \
+      MA = *((uint32_t*)(buf + 40)); \
+      MB = *((uint32_t*)(buf + 44)); \
+      MC = *((uint32_t*)(buf + 48)); \
+      MD = *((uint32_t*)(buf + 52)); \
+      ME = *((uint32_t*)(buf + 56)); \
+      MF = *((uint32_t*)(buf + 60)); \
+      ROUND_S(0); \
+      ROUND_S(1); \
+      ROUND_S(2); \
+      ROUND_S(3); \
+      ROUND_S(4); \
+      ROUND_S(5); \
+      ROUND_S(6); \
+      ROUND_S(7); \
+      if (BLAKE32_ROUNDS == 14) { \
+      ROUND_S(8); \
+      ROUND_S(9); \
+      ROUND_S(0); \
+      ROUND_S(1); \
+      ROUND_S(2); \
+      ROUND_S(3); \
+      } \
+      H0 ^= S0 ^ V0 ^ V8; \
+      H1 ^= S1 ^ V1 ^ V9; \
+      H2 ^= S2 ^ V2 ^ VA; \
+      H3 ^= S3 ^ V3 ^ VB; \
+      H4 ^= S0 ^ V4 ^ VC; \
+      H5 ^= S1 ^ V5 ^ VD; \
+      H6 ^= S2 ^ V6 ^ VE; \
+      H7 ^= S3 ^ V7 ^ VF; \
+   } while (0)
+
 #endif
 
 #if SPH_64
@@ -843,6 +906,45 @@ blake32(sph_blake_small_context *sc, const void *data, size_t len)
 	sc->ptr = ptr;
 }
 
+static void
+blake32_le(sph_blake_small_context *sc, const void *data, size_t len)
+{
+   unsigned char *buf;
+   size_t ptr;
+   DECL_STATE32
+
+   buf = sc->buf;
+   ptr = sc->ptr;
+
+   if (len < (sizeof sc->buf) - ptr) {
+      memcpy(buf + ptr, data, len);
+      ptr += len;
+      sc->ptr = ptr;
+      return;
+   }
+
+   READ_STATE32(sc);
+   while (len > 0) {
+      size_t clen;
+
+      clen = (sizeof sc->buf) - ptr;
+      if (clen > len)
+         clen = len;
+      memcpy(buf + ptr, data, clen);
+      ptr += clen;
+      data = (const unsigned char *)data + clen;
+      len -= clen;
+      if (ptr == sizeof sc->buf) {
+         if ((T0 = SPH_T32(T0 + 512)) < 512)
+            T1 = SPH_T32(T1 + 1);
+         COMPRESS32_LE;
+         ptr = 0;
+      }
+   }
+   WRITE_STATE32(sc);
+   sc->ptr = ptr;
+}
+
 static void
 blake32_close(sph_blake_small_context *sc,
 	unsigned ub, unsigned n, void *dst, size_t out_size_w32)
@@ -1050,6 +1152,12 @@ sph_blake256(void *cc, const void *data, size_t len)
 	blake32(cc, data, len);
 }
 
+void
+sph_blake256_update_le(void *cc, const void *data, size_t len)
+{
+   blake32_le(cc, data, len);
+}
+
 /* see sph_blake.h */
 void
 sph_blake256_close(void *cc, void *dst)

algo/blake/sph_blake.h

@@ -198,6 +198,7 @@ void sph_blake256_init(void *cc);
  * @param len    the input data length (in bytes)
  */
 void sph_blake256(void *cc, const void *data, size_t len);
+void sph_blake256_update_le(void *cc, const void *data, size_t len);
 
 /**
  * Terminate the current BLAKE-256 computation and output the result into

algo/blake/sph_blake2b.c

@@ -30,18 +30,11 @@
 #include <stdlib.h>
 #include <stdint.h>
 #include <string.h>
-
+#include "simd-utils.h"
 #include "algo/sha/sph_types.h"
 #include "sph_blake2b.h"
 
-// Cyclic right rotation.
-
-#ifndef ROTR64
-#define ROTR64(x, y)  (((x) >> (y)) ^ ((x) << (64 - (y))))
-#endif
-
 // Little-endian byte access.
-
 #define B2B_GET64(p)                            \
 	(((uint64_t) ((uint8_t *) (p))[0]) ^        \
 	(((uint64_t) ((uint8_t *) (p))[1]) << 8) ^  \
@@ -52,47 +45,143 @@
 	(((uint64_t) ((uint8_t *) (p))[6]) << 48) ^ \
 	(((uint64_t) ((uint8_t *) (p))[7]) << 56))
 
-// G Mixing function.
+#if defined(__AVX2__)
 
-#define B2B_G(a, b, c, d, x, y) {   \
-	v[a] = v[a] + v[b] + x;         \
-	v[d] = ROTR64(v[d] ^ v[a], 32); \
-	v[c] = v[c] + v[d];             \
-	v[b] = ROTR64(v[b] ^ v[c], 24); \
-	v[a] = v[a] + v[b] + y;         \
-	v[d] = ROTR64(v[d] ^ v[a], 16); \
-	v[c] = v[c] + v[d];             \
-	v[b] = ROTR64(v[b] ^ v[c], 63); }
+#define BLAKE2B_G( Sa, Sb, Sc, Sd, Se, Sf, Sg, Sh ) \
+{ \
+  V[0] = _mm256_add_epi64( V[0], _mm256_add_epi64( V[1], \
+              _mm256_set_epi64x( m[ sigmaR[ Sg ] ], m[ sigmaR[ Se ] ], \
+                                 m[ sigmaR[ Sc ] ], m[ sigmaR[ Sa ] ] ) ) ); \
+  V[3] = mm256_swap64_32( _mm256_xor_si256( V[3], V[0] ) ); \
+  V[2] = _mm256_add_epi64( V[2], V[3] ); \
+  V[1] = mm256_shuflr64_24( _mm256_xor_si256( V[1], V[2] ) ); \
+\
+  V[0] = _mm256_add_epi64( V[0], _mm256_add_epi64( V[1], \
+              _mm256_set_epi64x( m[ sigmaR[ Sh ] ], m[ sigmaR[ Sf ] ], \
+                                 m[ sigmaR[ Sd ] ], m[ sigmaR[ Sb ] ] ) ) ); \
+  V[3] = mm256_shuflr64_16( _mm256_xor_si256( V[3], V[0] ) ); \
+  V[2] = _mm256_add_epi64( V[2], V[3] ); \
+  V[1] = mm256_ror_64( _mm256_xor_si256( V[1], V[2] ), 63 ); \
+}
+
+#define BLAKE2B_ROUND( R ) \
+{ \
+  __m256i *V = (__m256i*)v; \
+  const uint8_t *sigmaR = sigma[R]; \
+  BLAKE2B_G(  0,  1,  2,  3,  4,  5,  6,  7 ); \
+  V[3] = mm256_shufll_64( V[3] ); \
+  V[2] = mm256_swap_128( V[2] ); \
+  V[1] = mm256_shuflr_64( V[1] ); \
+  BLAKE2B_G(  8,  9, 10, 11, 12, 13, 14, 15 ); \
+  V[3] = mm256_shuflr_64( V[3] ); \
+  V[2] = mm256_swap_128( V[2] ); \
+  V[1] = mm256_shufll_64( V[1] ); \
+}
+
+#elif defined(__SSE2__)
+// always true
+
+#define BLAKE2B_G( Va, Vb, Vc, Vd, Sa, Sb, Sc, Sd ) \
+{ \
+   Va = _mm_add_epi64( Va, _mm_add_epi64( Vb, \
+                 _mm_set_epi64x( m[ sigmaR[ Sc ] ], m[ sigmaR[ Sa ] ] ) ) ); \
+   Vd = mm128_swap64_32( _mm_xor_si128( Vd, Va ) ); \
+   Vc = _mm_add_epi64( Vc, Vd ); \
+   Vb = mm128_shuflr64_24( _mm_xor_si128( Vb, Vc ) ); \
+\
+   Va = _mm_add_epi64( Va, _mm_add_epi64( Vb, \
+                 _mm_set_epi64x( m[ sigmaR[ Sd ] ], m[ sigmaR[ Sb ] ] ) ) ); \
+   Vd = mm128_shuflr64_16( _mm_xor_si128( Vd, Va ) ); \
+   Vc = _mm_add_epi64( Vc, Vd ); \
+   Vb = mm128_ror_64( _mm_xor_si128( Vb, Vc ), 63 ); \
+}
+
+#define BLAKE2B_ROUND( R ) \
+{ \
+   __m128i *V = (__m128i*)v; \
+   __m128i V2, V3, V6, V7; \
+   const uint8_t *sigmaR = sigma[R]; \
+   BLAKE2B_G( V[0], V[2], V[4], V[6], 0, 1, 2, 3 ); \
+   BLAKE2B_G( V[1], V[3], V[5], V[7], 4, 5, 6, 7 ); \
+   V2 = mm128_alignr_64( V[3], V[2] ); \
+   V3 = mm128_alignr_64( V[2], V[3] ); \
+   V6 = mm128_alignr_64( V[6], V[7] ); \
+   V7 = mm128_alignr_64( V[7], V[6] ); \
+   BLAKE2B_G( V[0], V2, V[5], V6,  8,  9, 10, 11 ); \
+   BLAKE2B_G( V[1], V3, V[4], V7, 12, 13, 14, 15 ); \
+   V[2] = mm128_alignr_64( V2, V3 ); \
+   V[3] = mm128_alignr_64( V3, V2 ); \
+   V[6] = mm128_alignr_64( V7, V6 ); \
+   V[7] = mm128_alignr_64( V6, V7 ); \
+}
+
+#else
+// never used, SSE2 is always available
+
+#ifndef ROTR64
+#define ROTR64(x, y)  (((x) >> (y)) ^ ((x) << (64 - (y))))
+#endif
+
+#define BLAKE2B_G( R, Va, Vb, Vc, Vd, Sa, Sb ) \
+{ \
+   Va = Va + Vb + m[ sigma[R][Sa] ]; \
+   Vd = ROTR64( Vd ^ Va, 32 ); \
+   Vc = Vc + Vd; \
+   Vb = ROTR64( Vb ^ Vc, 24 ); \
+\
+   Va = Va + Vb + m[ sigma[R][Sb] ]; \
+   Vd = ROTR64( Vd ^ Va, 16 ); \
+   Vc = Vc + Vd; \
+   Vb = ROTR64( Vb ^ Vc, 63 ); \
+}
+
+#define BLAKE2B_ROUND( R ) \
+{ \
+   BLAKE2B_G( R, v[ 0], v[ 4], v[ 8], v[12],  0,  1 ); \
+   BLAKE2B_G( R, v[ 1], v[ 5], v[ 9], v[13],  2,  3 ); \
+   BLAKE2B_G( R, v[ 2], v[ 6], v[10], v[14],  4,  5 ); \
+   BLAKE2B_G( R, v[ 3], v[ 7], v[11], v[15],  6,  7 ); \
+   BLAKE2B_G( R, v[ 0], v[ 5], v[10], v[15],  8,  9 ); \
+   BLAKE2B_G( R, v[ 1], v[ 6], v[11], v[12], 10, 11 ); \
+   BLAKE2B_G( R, v[ 2], v[ 7], v[ 8], v[13], 12, 13 ); \
+   BLAKE2B_G( R, v[ 3], v[ 4], v[ 9], v[14], 14, 15 ); \
+}
+
+#endif
 
 // Initialization Vector.
 
-static const uint64_t blake2b_iv[8] = {
+static const uint64_t blake2b_iv[8] __attribute__ ((aligned (32))) =
+{
 	0x6A09E667F3BCC908, 0xBB67AE8584CAA73B,
 	0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1,
 	0x510E527FADE682D1, 0x9B05688C2B3E6C1F,
 	0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179
 };
 
+static const uint8_t sigma[12][16] __attribute__ ((aligned (32))) =
+{
+      { 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 }
+};
+
 // Compression function. "last" flag indicates last block.
 
 static void blake2b_compress( sph_blake2b_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;
-	uint64_t v[16], m[16];
+	uint64_t v[16] __attribute__ ((aligned (32)));
+   uint64_t m[16] __attribute__ ((aligned (32)));
+   int i;
 
 	for (i = 0; i < 8; i++) {           // init work variables
 		v[i] = ctx->h[i];
@@ -103,20 +192,11 @@ static void blake2b_compress( sph_blake2b_ctx *ctx, int last )
 	v[13] ^= ctx->t[1];                 // high 64 bits
 	if (last)                           // last block flag set ?
 		v[14] = ~v[14];
-
 	for (i = 0; i < 16; i++)            // get little-endian words
 		m[i] = B2B_GET64(&ctx->b[8 * i]);
 
-	for (i = 0; i < 12; i++) {          // twelve rounds
-		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]]);
-		B2B_G( 2, 6, 10, 14, m[sigma[i][ 4]], m[sigma[i][ 5]]);
-		B2B_G( 3, 7, 11, 15, m[sigma[i][ 6]], m[sigma[i][ 7]]);
-		B2B_G( 0, 5, 10, 15, m[sigma[i][ 8]], m[sigma[i][ 9]]);
-		B2B_G( 1, 6, 11, 12, m[sigma[i][10]], m[sigma[i][11]]);
-		B2B_G( 2, 7,  8, 13, m[sigma[i][12]], m[sigma[i][13]]);
-		B2B_G( 3, 4,  9, 14, m[sigma[i][14]], m[sigma[i][15]]);
-	}
+	for (i = 0; i < 12; i++)
+      BLAKE2B_ROUND( i );   
 
 	for( i = 0; i < 8; ++i )
 		ctx->h[i] ^= v[i] ^ v[i + 8];
@@ -184,7 +264,8 @@ void sph_blake2b_final( sph_blake2b_ctx *ctx, void *out )
 
 	while (ctx->c < 128)                // fill up with zeros
 		ctx->b[ctx->c++] = 0;
-	blake2b_compress(ctx, 1);           // final block flag = 1
+
+   blake2b_compress(ctx, 1);           // final block flag = 1
 
 	// little endian convert and store
 	for (i = 0; i < ctx->outlen; i++) {

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

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

algo/blake/sse2/blake/sse41/api.h

@@ -1,2 +0,0 @@
-#define CRYPTO_BYTES 64
-

algo/blake/sse2/blake/sse41/architectures

@@ -1,2 +0,0 @@
-amd64
-x86

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
-

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
-
-*/
-
-

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 ) ;
-
-
-
-
-
-

algo/blake/sse2/blake/sse41/implementors

@@ -1,2 +0,0 @@
-Jean-Philippe Aumasson
-Samuel Neves

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
-

algo/bmw/bmw-hash-4way.h

@@ -41,15 +41,18 @@ extern "C"{
 #endif
 
 #include <stddef.h>
-#ifdef __AVX2__
 
 #include "algo/sha/sph_types.h"
-#include "avxdefs.h"
+#include "simd-utils.h"
 
 #define SPH_SIZE_bmw256   256
 
 #define SPH_SIZE_bmw512   512
 
+#if defined(__SSE2__)
+
+// BMW-256 4 way 32
+
 typedef struct {
    __m128i buf[64];
    __m128i H[16];
@@ -59,37 +62,126 @@ typedef struct {
 
 typedef bmw_4way_small_context bmw256_4way_context;
 
+void bmw256_4way_init( bmw256_4way_context *ctx );
+
+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);
+
+void bmw256_4way_addbits_and_close(
+        void *cc, unsigned ub, unsigned n, void *dst);
+
+#endif  // __SSE2__
+
+#if defined(__AVX2__)
+
+// BMW-256 8 way 32
+
+typedef struct {
+   __m256i buf[16];
+   __m256i H[16];
+   size_t ptr;
+   uint32_t bit_count;  // assume bit_count fits in 32 bits
+} bmw_8way_small_context __attribute__ ((aligned (64)));
+
+typedef bmw_8way_small_context bmw256_8way_context;
+
+void bmw256_8way_init( bmw256_8way_context *ctx );
+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__)
+
+// BMW-512 2 way 64
+
+typedef struct {
+   __m128i buf[16];
+   __m128i H[16];
+   size_t ptr;
+   uint64_t bit_count; 
+} bmw_2way_big_context __attribute__ ((aligned (64)));
+
+typedef bmw_2way_big_context bmw512_2way_context;
+
+void bmw512_2way_init( bmw512_2way_context *ctx );
+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 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 bmw256_4way_init(void *cc);
-
-void bmw256_4way(void *cc, const void *data, size_t len);
-
-void bmw256_4way_close(void *cc, void *dst);
-
-void bmw256_4way_addbits_and_close(
-	void *cc, unsigned ub, unsigned n, void *dst);
-
 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);
 
 void bmw512_4way_addbits_and_close(
 	void *cc, unsigned ub, unsigned n, void *dst);
 
-#endif
+#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
 
-#endif
+#endif // BMW_HASH_H__

algo/bmw/bmw256.c

@@ -19,14 +19,15 @@ void bmwhash(void *output, const void *input)
 */
 }
 
-int scanhash_bmw(int thr_id, struct work *work,
-	uint32_t max_nonce, uint64_t *hashes_done)
+int scanhash_bmw( 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 _ALIGN(64) hash64[8];
 	uint32_t _ALIGN(64) endiandata[20];
+   int thr_id = mythr->id;
 
 	const uint32_t Htarg = ptarget[7];
 	const uint32_t first_nonce = pdata[19];

algo/bmw/bmw512-4way.c

@@ -0,0 +1,113 @@
+#include "bmw512-gate.h"
+#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_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 (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];
+    int thr_id = mythr->id;  // thr_id arg is deprecated
+
+   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+   do {
+       *noncev = mm256_intrlv_blend_32( mm256_bswap_32(
+                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
+	
+      bmw512hash_4way( hash, vdata );
+
+      for ( int lane = 0; lane < 4; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
+      {
+          extr_lane_4x64( lane_hash, hash, lane, 256 );
+          if ( fulltest( lane_hash, ptarget ) )
+          {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+          }
+      }
+      n += 4;
+
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
+#endif

algo/bmw/bmw512-gate.c

@@ -0,0 +1,20 @@
+#include "bmw512-gate.h"
+
+bool register_bmw512_algo( algo_gate_t* gate )
+{
+  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
+  gate->scanhash        = (void*)&scanhash_bmw512;
+  gate->hash            = (void*)&bmw512hash;
+#endif
+  return true;
+};
+
+

algo/bmw/bmw512-gate.h

@@ -0,0 +1,33 @@
+#ifndef BMW512_GATE_H__
+#define BMW512_GATE_H__ 1
+
+#include "algo-gate-api.h"
+#include <stdint.h>
+
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+  #define BMW512_8WAY 1
+#elif defined(__AVX2__)
+  #define BMW512_4WAY 1
+#endif
+
+#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 );
+
+#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

algo/bmw/bmw512.c

@@ -0,0 +1,55 @@
+#include "algo-gate-api.h"
+
+#if !defined(BMW512_8WAY) && !defined(BMW512_4WAY)
+
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+
+#include "sph_bmw.h"
+
+void bmw512hash(void *state, const void *input)
+{
+    sph_bmw512_context ctx;
+    uint32_t hash[32];	
+   
+    sph_bmw512_init( &ctx );
+    sph_bmw512( &ctx,input, 80 );
+    sph_bmw512_close( &ctx, hash );
+
+    memcpy( state, hash, 32 );
+}
+
+int scanhash_bmw512( 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 n = pdata[19] - 1;
+	const uint32_t first_nonce = pdata[19];
+	//const uint32_t Htarg = ptarget[7];
+   int thr_id = mythr->id;  // thr_id arg is deprecated
+
+	uint32_t _ALIGN(32) hash64[8];
+	uint32_t endiandata[32];
+
+   for (int i=0; i < 19; i++) 
+           be32enc(&endiandata[i], pdata[i]);
+
+	do {
+	
+		pdata[19] = ++n;
+		be32enc(&endiandata[19], n); 
+		bmw512hash(hash64, endiandata);
+        if (((hash64[7]&0xFFFFFF00)==0) && 
+				fulltest(hash64, ptarget)) {
+            *hashes_done = n - first_nonce + 1;
+			return true;
+		}
+	} while (n < max_nonce && !work_restart[thr_id].restart);
+	
+	*hashes_done = n - first_nonce + 1;
+	pdata[19] = n;
+	return 0;
+}
+#endif

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 */

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
 
 /**

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

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

algo/cryptonight/cryptolight.c

@@ -1,364 +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);
-}
-
-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);
-}
-
-int scanhash_cryptolight(int thr_id, struct work *work,
-		uint32_t max_nonce, uint64_t *hashes_done)
-{
-        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];
-
-	struct cryptonight_ctx *ctx = (struct cryptonight_ctx*)malloc(sizeof(struct cryptonight_ctx));
-
-#ifndef NO_AES_NI
-		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;
-};
-

algo/cryptonight/cryptonight-aesni.c

@@ -1,334 +0,0 @@
-#include <x86intrin.h>
-#include <memory.h>
-#include "cryptonight.h"
-#include "miner.h"
-#include "crypto/c_keccak.h"
-#include <immintrin.h>
-//#include "avxdefs.h"
-
-void aesni_parallel_noxor(uint8_t *long_state, uint8_t *text, uint8_t *ExpandedKey);
-void aesni_parallel_xor(uint8_t *text, uint8_t *ExpandedKey, uint8_t *long_state);
-void that_fucking_loop(uint8_t a[16], uint8_t b[16], uint8_t *long_state);
-
-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)
-{
-#ifndef NO_AES_NI
-	__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);
-#endif
-}
-
-// Special thanks to Intel for helping me
-// with ExpandAESKey256() and its subroutines
-static inline void ExpandAESKey256(char *keybuf)
-{
-#ifndef NO_AES_NI
-	__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;
-#endif
-}
-
-// 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 )
-{
-#ifndef NO_AES_NI
-
-    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 );
-    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 );
-	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;
-        nextblock[0] = a[0] + hi;
-        nextblock[1] = a[1] + lo;
-        a[0] = b[0] ^ nextblock[0];
-        a[1] = b[1] ^ nextblock[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 );
-    b[0] = nextblock[0];
-    b[1] = nextblock[1];
-
-    __asm__( "mulq %3\n\t"
-             : "=d" ( hi ),
-               "=a" ( lo )
-             : "%a" ( c[0] ),
-               "rm" ( b[0] )
-             : "cc" );
-
-    nextblock[0] = a[0] + hi;
-    nextblock[1] = a[1] + lo;
-
-    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
-}

algo/cryptonight/cryptonight-common.c

@@ -1,108 +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"
-
-#ifndef NO_AES_NI
-  #include "algo/groestl/aes_ni/hash-groestl256.h"
-#endif
-
-#include "crypto/c_groestl.h"
-#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) {
-#ifdef NO_AES_NI
-    groestl(input, len * 8, (uint8_t*)output);
-#else
-    hashState_groestl256 ctx;
-    init_groestl256( &ctx, 32 );
-    update_and_final_groestl256( &ctx, output, input, len * 8 );
-#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 )
-{
-
-#ifdef NO_AES_NI
-  cryptonight_hash_ctx ( output, input, len );
-#else 
-  cryptonight_hash_aes( output, input, len );
-#endif
-}
-
-void cryptonight_hash_suw( void *restrict output, const void *input )
-{
-#ifdef NO_AES_NI
-  cryptonight_hash_ctx ( output, input, 76 );
-#else
-  cryptonight_hash_aes( output, input, 76 );
-#endif
-}
-
-int scanhash_cryptonight( int thr_id, struct work *work, uint32_t max_nonce,
-                   uint64_t *hashes_done )
- {
-    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 hash[32 / 4] __attribute__((aligned(32)));
-    do
-    {
-       *nonceptr = ++n;
-       cryptonight_hash( hash, pdata, 76 );
-       if (unlikely( hash[7] < Htarg ))
-       {
-           *hashes_done = n - first_nonce + 1;
-	   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 )
-{
-  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;
-};
-

algo/cryptonight/cryptonight.c

@@ -1,274 +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) {
-	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
-}
-
-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);
-	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]);
-		/* 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)]);
-	}
-
-    __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);
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx.state.hs.b[0] & 3](&ctx.state, 200, output);
-	oaes_free((OAES_CTX **) &ctx.aes_ctx);
-}
-

algo/cryptonight/cryptonight.h

@@ -1,49 +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( int thr_id, struct work *work, uint32_t max_nonce,
-                           uint64_t *hashes_done );
-
-void cryptonight_hash_aes( void *restrict output, const void *input, int len );
-
-#endif
-

algo/cubehash/cube-hash-2way.c

@@ -7,12 +7,444 @@
 
 // 2x128
 
+
+// The result of hashing 10 rounds of initial data which consists of params
+// zero padded.
+static const uint64_t IV256[] =
+{
+0xCCD6F29FEA2BD4B4, 0x35481EAE63117E71, 0xE5D94E6322512D5B, 0xF4CC12BE7E624131,
+0x42AF2070C2D0B696, 0x3361DA8CD0720C35, 0x8EF8AD8328CCECA4, 0x40E5FBAB4680AC00,
+0x6107FBD5D89041C3, 0xF0B266796C859D41, 0x5FA2560309392549, 0x93CB628565C892FD,
+0x9E4B4E602AF2B5AE, 0x85254725774ABFDD, 0x4AB6AAD615815AEB, 0xD6032C0A9CDAF8AF
+};
+
+static const uint64_t IV512[] =
+{
+0x50F494D42AEA2A61, 0x4167D83E2D538B8B, 0xC701CF8C3FEE2313, 0x50AC5695CC39968E,
+0xA647A8B34D42C787, 0x825B453797CF0BEF, 0xF22090C4EEF864D2, 0xA23911AED0E5CD33,
+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 = mm512_rol_32( x2, 7 );
+        y1 = mm512_rol_32( x3, 7 );
+        x2 = mm512_rol_32( x0, 7 );
+        x3 = mm512_rol_32( x1, 7 );
+        x0 = _mm512_xor_si512( y0, x4 );
+        x1 = _mm512_xor_si512( y1, 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 = mm512_rol_32( x1, 11 );
+        x1 = mm512_rol_32( x0, 11 );
+        y1 = mm512_rol_32( x3, 11 );
+        x3 = mm512_rol_32( x2, 11 );
+        x0 = _mm512_xor_si512( y0, x4 );
+        x1 = _mm512_xor_si512( x1, x5 );
+        x2 = _mm512_xor_si512( y1, 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 );
+}
+
+// 8 ways, 4 way parallel double buffered
+static void transform_4way_2buf( cube_4way_2buf_context *sp )
+{
+    int r;
+    const int rounds = sp->rounds;
+
+    __m512i x0, x1, x2, x3, x4, x5, x6, x7;
+    __m512i y0, y1, y2, y3, y4, y5, y6, y7;
+    __m512i tx0, tx1, ty0, ty1;
+
+    x0 = _mm512_load_si512( (__m512i*)sp->h0     );
+    x1 = _mm512_load_si512( (__m512i*)sp->h0 + 1 );
+    x2 = _mm512_load_si512( (__m512i*)sp->h0 + 2 );
+    x3 = _mm512_load_si512( (__m512i*)sp->h0 + 3 );
+    x4 = _mm512_load_si512( (__m512i*)sp->h0 + 4 );
+    x5 = _mm512_load_si512( (__m512i*)sp->h0 + 5 );
+    x6 = _mm512_load_si512( (__m512i*)sp->h0 + 6 );
+    x7 = _mm512_load_si512( (__m512i*)sp->h0 + 7 );
+
+    y0 = _mm512_load_si512( (__m512i*)sp->h1     );
+    y1 = _mm512_load_si512( (__m512i*)sp->h1 + 1 );
+    y2 = _mm512_load_si512( (__m512i*)sp->h1 + 2 );
+    y3 = _mm512_load_si512( (__m512i*)sp->h1 + 3 );
+    y4 = _mm512_load_si512( (__m512i*)sp->h1 + 4 );
+    y5 = _mm512_load_si512( (__m512i*)sp->h1 + 5 );
+    y6 = _mm512_load_si512( (__m512i*)sp->h1 + 6 );
+    y7 = _mm512_load_si512( (__m512i*)sp->h1 + 7 );
+
+
+    for ( r = 0; r < rounds; ++r )
+    {
+        x4 = _mm512_add_epi32( x0, x4 );
+        y4 = _mm512_add_epi32( y0, y4 );
+        x5 = _mm512_add_epi32( x1, x5 );
+        y5 = _mm512_add_epi32( y1, y5 );
+        tx0 = mm512_rol_32( x2, 7 );
+        ty0 = mm512_rol_32( y2, 7 );
+        tx1 = mm512_rol_32( x3, 7 );
+        ty1 = mm512_rol_32( y3, 7 );
+        x6 = _mm512_add_epi32( x2, x6 );
+        y6 = _mm512_add_epi32( y2, y6 ); 
+        x7 = _mm512_add_epi32( x3, x7 );
+        y7 = _mm512_add_epi32( y3, y7 );
+        x2 = mm512_rol_32( x0, 7 );
+        y2 = mm512_rol_32( y0, 7 );
+        x3 = mm512_rol_32( x1, 7 );
+        y3 = mm512_rol_32( y1, 7 );
+        x0 = _mm512_xor_si512( tx0, x4 );
+        y0 = _mm512_xor_si512( ty0, y4 );
+        x1 = _mm512_xor_si512( tx1, x5 );
+        y1 = _mm512_xor_si512( ty1, y5 );
+        x4 = mm512_swap128_64( x4 );
+        y4 = mm512_swap128_64( y4 );
+        x5 = mm512_swap128_64( x5 );
+        y5 = mm512_swap128_64( y5 );
+        x2 = _mm512_xor_si512( x2, x6 );
+        y2 = _mm512_xor_si512( y2, y6 );
+        x3 = _mm512_xor_si512( x3, x7 );
+        y3 = _mm512_xor_si512( y3, y7 );
+        x6 = mm512_swap128_64( x6 );
+        y6 = mm512_swap128_64( y6 );
+        x7 = mm512_swap128_64( x7 );
+        y7 = mm512_swap128_64( y7 );
+        x4 = _mm512_add_epi32( x0, x4 );
+        y4 = _mm512_add_epi32( y0, y4 );
+        x5 = _mm512_add_epi32( x1, x5 );
+        y5 = _mm512_add_epi32( y1, y5 );
+        tx0 = mm512_rol_32( x1, 11 );
+        ty0 = mm512_rol_32( y1, 11 );
+        tx1 = mm512_rol_32( x3, 11 );
+        ty1 = mm512_rol_32( y3, 11 );
+        x6 = _mm512_add_epi32( x2, x6 );
+        y6 = _mm512_add_epi32( y2, y6 );
+        x7 = _mm512_add_epi32( x3, x7 );
+        y7 = _mm512_add_epi32( y3, y7 );
+        x1 = mm512_rol_32( x0, 11 );
+        y1 = mm512_rol_32( y0, 11 );
+        x3 = mm512_rol_32( x2, 11 );
+        y3 = mm512_rol_32( y2, 11 );
+        x0 = _mm512_xor_si512( tx0, x4 );
+        y0 = _mm512_xor_si512( ty0, y4 );
+        x1 = _mm512_xor_si512( x1, x5 );
+        y1 = _mm512_xor_si512( y1, y5 );
+        x4 = mm512_swap64_32( x4 );
+        y4 = mm512_swap64_32( y4 );
+        x5 = mm512_swap64_32( x5 );
+        y5 = mm512_swap64_32( y5 );
+        x2 = _mm512_xor_si512( tx1, x6 );
+        y2 = _mm512_xor_si512( ty1, y6 );
+        x3 = _mm512_xor_si512( x3, x7 );
+        y3 = _mm512_xor_si512( y3, y7 );
+        x6 = mm512_swap64_32( x6 );
+        y6 = mm512_swap64_32( y6 );
+        x7 = mm512_swap64_32( x7 );
+        y7 = mm512_swap64_32( y7 );
+    }
+
+    _mm512_store_si512( (__m512i*)sp->h0,     x0 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 1, x1 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 2, x2 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 3, x3 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 4, x4 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 5, x5 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 6, x6 );
+    _mm512_store_si512( (__m512i*)sp->h0 + 7, x7 );
+
+    _mm512_store_si512( (__m512i*)sp->h1,     y0 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 1, y1 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 2, y2 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 3, y3 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 4, y4 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 5, y5 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 6, y6 );
+    _mm512_store_si512( (__m512i*)sp->h1 + 7, y7 );
+}
+
+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] );
+
+    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_2buf_full( cube_4way_2buf_context *sp,
+                         void *output0, void *output1, int hashbitlen,
+                         const void *data0, const void *data1, size_t size )
+{
+    __m512i *h0 = (__m512i*)sp->h0;
+    __m512i *h1 = (__m512i*)sp->h1;
+    __m128i *iv = (__m128i*)( hashbitlen == 512 ? (__m128i*)IV512
+                                                : (__m128i*)IV256 );
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = 32/16;
+    sp->rounds    = 16;
+    sp->pos       = 0;
+
+    h1[0] = h0[0] = m512_const1_128( iv[0] );
+    h1[1] = h0[1] = m512_const1_128( iv[1] );
+    h1[2] = h0[2] = m512_const1_128( iv[2] );
+    h1[3] = h0[3] = m512_const1_128( iv[3] );
+    h1[4] = h0[4] = m512_const1_128( iv[4] );
+    h1[5] = h0[5] = m512_const1_128( iv[5] );
+    h1[6] = h0[6] = m512_const1_128( iv[6] );
+    h1[7] = h0[7] = m512_const1_128( iv[7] );
+
+    const int len = size >> 4;
+    const __m512i *in0 = (__m512i*)data0;
+    const __m512i *in1 = (__m512i*)data1;
+    __m512i *hash0 = (__m512i*)output0;
+    __m512i *hash1 = (__m512i*)output1;
+    int i;
+
+    for ( i = 0; i < len; i++ )
+    {
+        sp->h0[ sp->pos ] = _mm512_xor_si512( sp->h0[ sp->pos ], in0[i] );
+        sp->h1[ sp->pos ] = _mm512_xor_si512( sp->h1[ sp->pos ], in1[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform_4way_2buf( sp );
+           sp->pos = 0;
+        }
+    }
+
+    // pos is zero for 64 byte data, 1 for 80 byte data.
+    __m512i tmp = m512_const2_64( 0, 0x0000000000000080 );
+    sp->h0[ sp->pos ] = _mm512_xor_si512( sp->h0[ sp->pos ], tmp );
+    sp->h1[ sp->pos ] = _mm512_xor_si512( sp->h1[ sp->pos ], tmp );
+
+    transform_4way_2buf( sp );
+
+    tmp = m512_const2_64( 0x0000000100000000, 0 );
+    sp->h0[7] = _mm512_xor_si512( sp->h0[7], tmp );
+    sp->h1[7] = _mm512_xor_si512( sp->h1[7], tmp );
+
+    for ( i = 0; i < 10; ++i )
+       transform_4way_2buf( sp );
+
+    memcpy( hash0, sp->h0, sp->hashlen<<6);
+    memcpy( hash1, sp->h1, 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 
+
+// This isn't expected to be used with AVX512 so HW rotate intruction
+// is assumed not avaiable.
+// Use double buffering to optimize serial bit rotations. Full double
+// buffering isn't practical because it needs twice as many registers
+// with AVX2 having only half as many as AVX512.
+#define ROL2( out0, out1, in0, in1, c ) \
+{ \
+ __m256i t0 = _mm256_slli_epi32( in0, c ); \
+ __m256i t1 = _mm256_slli_epi32( in1, c ); \
+ out0 = _mm256_srli_epi32( in0, 32-(c) ); \
+ out1 = _mm256_srli_epi32( in1, 32-(c) ); \
+ out0 = _mm256_or_si256( out0, t0 ); \
+ out1 = _mm256_or_si256( out1, t1 ); \
+}
+
 static void transform_2way( cube_2way_context *sp )
 {
     int r;
     const int rounds = sp->rounds;
 
-    __m256i x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3;
+    __m256i x0, x1, x2, x3, x4, x5, x6, x7, y0, y1;
 
     x0 = _mm256_load_si256( (__m256i*)sp->h     );
     x1 = _mm256_load_si256( (__m256i*)sp->h + 1 );
@@ -29,20 +461,10 @@ static void transform_2way( cube_2way_context *sp )
         x5 = _mm256_add_epi32( x1, x5 );
         x6 = _mm256_add_epi32( x2, x6 );
         x7 = _mm256_add_epi32( x3, x7 );
-        y0 = x2;
-        y1 = x3;
-        y2 = x0;
-        y3 = x1;
-        x0 = _mm256_xor_si256( _mm256_slli_epi32( y0,  7 ),
-                               _mm256_srli_epi32( y0, 25 ) );
-        x1 = _mm256_xor_si256( _mm256_slli_epi32( y1,  7 ),
-                               _mm256_srli_epi32( y1, 25 ) );
-        x2 = _mm256_xor_si256( _mm256_slli_epi32( y2,  7 ),
-                               _mm256_srli_epi32( y2, 25 ) );
-        x3 = _mm256_xor_si256( _mm256_slli_epi32( y3,  7 ),
-                               _mm256_srli_epi32( y3, 25 ) );
-        x0 = _mm256_xor_si256( x0, x4 );
-        x1 = _mm256_xor_si256( x1, x5 );
+        ROL2( y0, y1, x2, x3, 7 );
+        ROL2( x2, x3, x0, x1, 7 );
+        x0 = _mm256_xor_si256( y0, x4 );
+        x1 = _mm256_xor_si256( y1, x5 );
         x2 = _mm256_xor_si256( x2, x6 );
         x3 = _mm256_xor_si256( x3, x7 );
         x4 = mm256_swap128_64( x4 );
@@ -53,21 +475,11 @@ static void transform_2way( cube_2way_context *sp )
         x5 = _mm256_add_epi32( x1, x5 );
         x6 = _mm256_add_epi32( x2, x6 );
         x7 = _mm256_add_epi32( x3, x7 );
-        y0 = x1;
-        y1 = x0;
-        y2 = x3;
-        y3 = x2;
-        x0 = _mm256_xor_si256( _mm256_slli_epi32( y0, 11 ),
-                               _mm256_srli_epi32( y0, 21 ) );
-        x1 = _mm256_xor_si256( _mm256_slli_epi32( y1, 11 ),
-                               _mm256_srli_epi32( y1, 21 ) );
-        x2 = _mm256_xor_si256( _mm256_slli_epi32( y2, 11 ),
-                               _mm256_srli_epi32( y2, 21 ) );
-        x3 = _mm256_xor_si256( _mm256_slli_epi32( y3, 11 ),
-                               _mm256_srli_epi32( y3, 21 ) );
-        x0 = _mm256_xor_si256( x0, x4 );
+        ROL2( y0, x1, x1, x0, 11 );
+        ROL2( y1, x3, x3, x2, 11 );
+        x0 = _mm256_xor_si256( y0, x4 );
         x1 = _mm256_xor_si256( x1, x5 );
-        x2 = _mm256_xor_si256( x2, x6 );
+        x2 = _mm256_xor_si256( y1, x6 );
         x3 = _mm256_xor_si256( x3, x7 );
         x4 = mm256_swap64_32( x4 );
         x5 = mm256_swap64_32( x5 );
@@ -83,53 +495,37 @@ 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 );
-
-}
-
-cube_2way_context cube_2way_ctx_cache __attribute__ ((aligned (64)));
-
-int cube_2way_reinit( cube_2way_context *sp )
-{
-   memcpy( sp, &cube_2way_ctx_cache, sizeof(cube_2way_context) );
-   return 0;
-
 }
 
 int cube_2way_init( cube_2way_context *sp, int hashbitlen, int rounds,
-                       int blockbytes )
+                    int blockbytes )
 {
-    int i;
+    __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;
 
-    // all sizes of __m128i
-    cube_2way_ctx_cache.hashlen   = hashbitlen/128;
-    cube_2way_ctx_cache.blocksize = blockbytes/16;
-    cube_2way_ctx_cache.rounds    = rounds;
-    cube_2way_ctx_cache.pos       = 0;
-
-    for ( i = 0; i < 8; ++i )
-       cube_2way_ctx_cache.h[i] = m256_zero;
-
-    cube_2way_ctx_cache.h[0] = _mm256_set_epi32(
-                                   0, rounds, blockbytes, hashbitlen / 8,
-                                   0, rounds, blockbytes, hashbitlen / 8 );
-
-    for ( i = 0; i < 10; ++i )
-       transform_2way( &cube_2way_ctx_cache );
-
-    memcpy( sp, &cube_2way_ctx_cache, sizeof(cube_2way_context) );
+    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;
 }
 
-
 int cube_2way_update( cube_2way_context *sp, const void *data, size_t size )
 {
-    const int len = size / 16;
+    const int len = size >> 4;
     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] );
@@ -140,7 +536,6 @@ int cube_2way_update( cube_2way_context *sp, const void *data, size_t size )
            sp->pos = 0;
         }
     }
-
     return 0;
 }
 
@@ -151,25 +546,22 @@ 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_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0x80,
-                                     0,0,0,0, 0,0,0,0, 0,0,0,0, 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 ) );
-    for ( i = 0; i < 10; ++i )
-       transform_2way( &cube_2way_ctx_cache );
+    sp->h[7] = _mm256_xor_si256( sp->h[7],
+                                   m256_const2_64( 0x0000000100000000, 0 ) );
 
-    for ( i = 0; i < sp->hashlen; i++ )
-       hash[i] = sp->h[i];
+    for ( i = 0; i < 10; ++i )           transform_2way( sp );
 
+    memcpy( hash, sp->h, sp->hashlen<<5 );
     return 0;
 }
 
 int cube_2way_update_close( cube_2way_context *sp, void *output,
                                const void *data, size_t size )
 {
-    const int len = size / 16;
+    const int len = size >> 4;
     const __m256i *in = (__m256i*)data;
     __m256i *hash = (__m256i*)output;
     int i;
@@ -187,18 +579,65 @@ 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_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0x80,
-                                     0,0,0,0, 0,0,0,0, 0,0,0,0, 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 ) );
-    for ( i = 0; i < 10; ++i )
-       transform_2way( &cube_2way_ctx_cache );
+    sp->h[7] = _mm256_xor_si256( sp->h[7],
+                                    m256_const2_64( 0x0000000100000000, 0 ) );
 
-    for ( i = 0; i < sp->hashlen; i++ )
-       hash[i] = sp->h[i];
+    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;
 }
 

algo/cubehash/cube-hash-2way.h

@@ -1,35 +1,77 @@
 #ifndef CUBE_HASH_2WAY_H__
-#define CUBE_HASH_2WAY_H__
+#define CUBE_HASH_2WAY_H__ 1
+
+#include <stdint.h>
+#include "simd-utils.h"
 
 #if defined(__AVX2__)
 
-#include <stdint.h>
-#include "avxdefs.h"
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
 
-// 2x128, 2 way parallel SSE2
+struct _cube_4way_context
+{
+    __m512i h[8];
+    int hashlen;
+    int rounds;
+    int blocksize;
+    int pos; 
+} __attribute__ ((aligned (128)));
+
+struct _cube_4way_2buf_context
+{
+    __m512i h0[8];
+    __m512i h1[8];
+    int hashlen;
+    int rounds;
+    int blocksize;
+    int pos;
+} __attribute__ ((aligned (128)));
+
+
+typedef struct _cube_4way_context cube_4way_context;
+
+typedef struct _cube_4way_2buf_context cube_4way_2buf_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_4way_2buf_full( cube_4way_2buf_context *sp,
+                         void *output0, void *output1, int hashbitlen,
+                         const void *data0, const void *data1, size_t size );
+
+#endif
+
+// 2x128, 2 way parallel AVX2
 
 struct _cube_2way_context
 {
+    __m256i h[8];
     int hashlen;           // __m128i
     int rounds;
     int blocksize;         // __m128i
     int pos;               // number of __m128i read into x from current block
-    __m256i h[8] __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

algo/cubehash/cubehash_sse2.c

@@ -0,0 +1,382 @@
+/* CubeHash 16/32 is recommended for SHA-3 "normal", 16/1 for "formal" */
+#define CUBEHASH_ROUNDS	16
+#define CUBEHASH_BLOCKBYTES 32
+#define OPTIMIZE_SSE2
+#if defined(OPTIMIZE_SSE2)
+#include <emmintrin.h>
+#endif
+#ifdef __AVX2__
+#include <immintrin.h>
+#endif
+#include "cubehash_sse2.h"
+#include "algo/sha/sha3-defs.h"
+#include <stdbool.h>
+#include <unistd.h>
+#include <memory.h>
+#include "simd-utils.h"
+#include <stdio.h>
+
+static void transform( cubehashParam *sp )
+{
+    int r;
+    const int rounds = sp->rounds;
+
+#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_swap_256( x0 );
+        x0 = mm512_rol_32(  x0, 7 );
+        x0 = _mm512_xor_si512( x0, x1 );
+        x1 = mm512_swap128_64( x1 );
+        x1 = _mm512_add_epi32( x0, x1 );
+        x0 = mm512_swap256_128( x0 );
+        x0 = mm512_rol_32( x0, 11 );
+        x0 = _mm512_xor_si512( x0, 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;
+
+    x0 = _mm256_load_si256( (__m256i*)sp->x     );
+    x1 = _mm256_load_si256( (__m256i*)sp->x + 1 );   
+    x2 = _mm256_load_si256( (__m256i*)sp->x + 2 );
+    x3 = _mm256_load_si256( (__m256i*)sp->x + 3 );
+
+    for ( r = 0; r < rounds; ++r )
+    { 
+        x2 = _mm256_add_epi32( x0, x2 );
+        x3 = _mm256_add_epi32( x1, x3 );
+        y0 = x0;
+        x0 = mm256_rol_32( x1, 7 );
+        x1 = mm256_rol_32( y0, 7 );
+        x0 = _mm256_xor_si256( x0, x2 );
+        x1 = _mm256_xor_si256( x1, 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 );
+        y1 = mm256_swap_128( x1 );
+        x0 = mm256_rol_32( y0, 11 );
+        x1 = mm256_rol_32( y1, 11 );
+        x0 = _mm256_xor_si256( x0, x2 );
+        x1 = _mm256_xor_si256( x1, x3 );
+        x2 = mm256_swap64_32( x2 );
+        x3 = mm256_swap64_32( x3 );
+    }
+
+    _mm256_store_si256( (__m256i*)sp->x,     x0 );
+    _mm256_store_si256( (__m256i*)sp->x + 1, x1 );
+    _mm256_store_si256( (__m256i*)sp->x + 2, x2 );
+    _mm256_store_si256( (__m256i*)sp->x + 3, x3 );
+
+#else
+    __m128i x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3;
+
+    x0 = _mm_load_si128( (__m128i*)sp->x     );
+    x1 = _mm_load_si128( (__m128i*)sp->x + 1 );
+    x2 = _mm_load_si128( (__m128i*)sp->x + 2 );
+    x3 = _mm_load_si128( (__m128i*)sp->x + 3 );
+    x4 = _mm_load_si128( (__m128i*)sp->x + 4 );
+    x5 = _mm_load_si128( (__m128i*)sp->x + 5 );
+    x6 = _mm_load_si128( (__m128i*)sp->x + 6 );
+    x7 = _mm_load_si128( (__m128i*)sp->x + 7 );
+
+    for (r = 0; r < rounds; ++r) {
+	x4 = _mm_add_epi32(x0, x4);
+	x5 = _mm_add_epi32(x1, x5);
+	x6 = _mm_add_epi32(x2, x6);
+	x7 = _mm_add_epi32(x3, x7);
+	y0 = x2;
+	y1 = x3;
+	y2 = x0;
+	y3 = x1;
+	x0 = _mm_xor_si128(_mm_slli_epi32(y0, 7), _mm_srli_epi32(y0, 25));
+	x1 = _mm_xor_si128(_mm_slli_epi32(y1, 7), _mm_srli_epi32(y1, 25));
+	x2 = _mm_xor_si128(_mm_slli_epi32(y2, 7), _mm_srli_epi32(y2, 25));
+	x3 = _mm_xor_si128(_mm_slli_epi32(y3, 7), _mm_srli_epi32(y3, 25));
+	x0 = _mm_xor_si128(x0, x4);
+	x1 = _mm_xor_si128(x1, x5);
+	x2 = _mm_xor_si128(x2, x6);
+	x3 = _mm_xor_si128(x3, x7);
+	x4 = _mm_shuffle_epi32(x4, 0x4e);
+	x5 = _mm_shuffle_epi32(x5, 0x4e);
+	x6 = _mm_shuffle_epi32(x6, 0x4e);
+	x7 = _mm_shuffle_epi32(x7, 0x4e);
+	x4 = _mm_add_epi32(x0, x4);
+	x5 = _mm_add_epi32(x1, x5);
+	x6 = _mm_add_epi32(x2, x6);
+	x7 = _mm_add_epi32(x3, x7);
+	y0 = x1;
+	y1 = x0;
+	y2 = x3;
+	y3 = x2;
+	x0 = _mm_xor_si128(_mm_slli_epi32(y0, 11), _mm_srli_epi32(y0, 21));
+	x1 = _mm_xor_si128(_mm_slli_epi32(y1, 11), _mm_srli_epi32(y1, 21));
+	x2 = _mm_xor_si128(_mm_slli_epi32(y2, 11), _mm_srli_epi32(y2, 21));
+	x3 = _mm_xor_si128(_mm_slli_epi32(y3, 11), _mm_srli_epi32(y3, 21));
+	x0 = _mm_xor_si128(x0, x4);
+	x1 = _mm_xor_si128(x1, x5);
+	x2 = _mm_xor_si128(x2, x6);
+	x3 = _mm_xor_si128(x3, x7);
+	x4 = _mm_shuffle_epi32(x4, 0xb1);
+	x5 = _mm_shuffle_epi32(x5, 0xb1);
+	x6 = _mm_shuffle_epi32(x6, 0xb1);
+	x7 = _mm_shuffle_epi32(x7, 0xb1);
+    }
+
+    _mm_store_si128( (__m128i*)sp->x,     x0 );
+    _mm_store_si128( (__m128i*)sp->x + 1, x1 );
+    _mm_store_si128( (__m128i*)sp->x + 2, x2 );
+    _mm_store_si128( (__m128i*)sp->x + 3, x3 );
+    _mm_store_si128( (__m128i*)sp->x + 4, x4 );
+    _mm_store_si128( (__m128i*)sp->x + 5, x5 );
+    _mm_store_si128( (__m128i*)sp->x + 6, x6 );
+    _mm_store_si128( (__m128i*)sp->x + 7, x7 );
+
+#endif
+}  // transform
+
+/*
+// The result of hashing 10 rounds of initial data which is params and
+// mostly zeros.
+static const uint64_t IV256[] =
+{
+0xCCD6F29FEA2BD4B4, 0x35481EAE63117E71, 0xE5D94E6322512D5B, 0xF4CC12BE7E624131,
+0x42AF2070C2D0B696, 0x3361DA8CD0720C35, 0x8EF8AD8328CCECA4, 0x40E5FBAB4680AC00,
+0x6107FBD5D89041C3, 0xF0B266796C859D41, 0x5FA2560309392549, 0x93CB628565C892FD,
+0x9E4B4E602AF2B5AE, 0x85254725774ABFDD, 0x4AB6AAD615815AEB, 0xD6032C0A9CDAF8AF
+};
+
+static const uint64_t IV512[] =
+{
+0x50F494D42AEA2A61, 0x4167D83E2D538B8B, 0xC701CF8C3FEE2313, 0x50AC5695CC39968E,
+0xA647A8B34D42C787, 0x825B453797CF0BEF, 0xF22090C4EEF864D2, 0xA23911AED0E5CD33,
+0x148FE485FCD398D9, 0xB64445321B017BEF, 0x2FF5781C6A536159, 0x0DBADEA991FA7934,
+0xA5A70E75D65C8A2B, 0xBC796576B1C62456, 0xE7989AF11921C8F7, 0xD43E3B447795D246
+};
+*/
+
+int cubehashInit(cubehashParam *sp, int hashbitlen, int rounds, int blockbytes)
+{
+    __m128i *x = (__m128i*)sp->x;
+    sp->hashlen   = hashbitlen/128;
+    sp->blocksize = blockbytes/16;
+    sp->rounds    = rounds;
+    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 );
+    }   
+
+    return SUCCESS;
+}
+
+int cubehashUpdate( cubehashParam *sp, const byte *data, size_t size )
+{
+    const int len = size / 16;
+    const __m128i* in = (__m128i*)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->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ], in[i] );
+        sp->pos++;
+        if ( sp->pos == sp->blocksize )
+        {
+           transform( sp );
+           sp->pos = 0;
+        }
+    }
+
+    return SUCCESS;
+}
+
+int cubehashDigest( cubehashParam *sp, byte *digest )
+{
+    __m128i* hash = (__m128i*)digest;
+    int i;
+
+    // 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 );
+    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 cubehashUpdateDigest( cubehashParam *sp, byte *digest,
+                          const byte *data, size_t size )
+{
+    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 );
+    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 );
+    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;
+}
+

algo/cubehash/cubehash_sse2.h

@@ -15,11 +15,11 @@
 
 struct _cubehashParam
 {
+    __m128i _ALIGN(64) x[8];  // aligned for __m512i
     int hashlen;           // __m128i
     int rounds;
     int blocksize;         // __m128i
     int pos;	           // number of __m128i read into x from current block
-    __m128i _ALIGN(256) 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

algo/cubehash/sse2/cubehash_sse2.c

@@ -1,273 +0,0 @@
-/* CubeHash 16/32 is recommended for SHA-3 "normal", 16/1 for "formal" */
-#define CUBEHASH_ROUNDS	16
-#define CUBEHASH_BLOCKBYTES 32
-#define OPTIMIZE_SSE2
-#if defined(OPTIMIZE_SSE2)
-#include <emmintrin.h>
-#endif
-#ifdef __AVX2__
-#include <immintrin.h>
-#endif
-#include "cubehash_sse2.h"
-#include "algo/sha/sha3-defs.h"
-#include <stdbool.h>
-#include <unistd.h>
-#include <memory.h>
-#include "avxdefs.h"
-
-static void transform( cubehashParam *sp )
-{
-    int r;
-    const int rounds = sp->rounds;
-
-#ifdef __AVX2__
-
-    register __m256i x0, x1, x2, x3, y0, y1;
-
-    x0 = _mm256_load_si256( (__m256i*)sp->x     );
-    x1 = _mm256_load_si256( (__m256i*)sp->x + 1 );   
-    x2 = _mm256_load_si256( (__m256i*)sp->x + 2 );
-    x3 = _mm256_load_si256( (__m256i*)sp->x + 3 );
-
-    for ( r = 0; r < rounds; ++r )
-    { 
-        x2 = _mm256_add_epi32( x0, x2 );
-        x3 = _mm256_add_epi32( x1, x3 );
-        y0 = x0;
-        x0 = _mm256_xor_si256( _mm256_slli_epi32( x1, 7 ),
-                               _mm256_srli_epi32( x1, 25 ) );
-        x1 = _mm256_xor_si256( _mm256_slli_epi32( y0, 7 ),
-                               _mm256_srli_epi32( y0, 25 ) );
-        x0 = _mm256_xor_si256( x0, x2 );
-        x1 = _mm256_xor_si256( x1, x3 );
-        x2 = _mm256_shuffle_epi32( x2, 0x4e );
-        x3 = _mm256_shuffle_epi32( x3, 0x4e );
-        x2 = _mm256_add_epi32( x0, x2 );
-        x3 = _mm256_add_epi32( x1, x3 );
-        y0 = _mm256_permute4x64_epi64( x0, 0x4e );
-        y1 = _mm256_permute4x64_epi64( x1, 0x4e );
-        x0 = _mm256_xor_si256( _mm256_slli_epi32( y0, 11 ),
-                               _mm256_srli_epi32( y0, 21 ) );
-        x1 = _mm256_xor_si256( _mm256_slli_epi32( y1, 11 ), 
-                               _mm256_srli_epi32( y1, 21 ) );
-        x0 = _mm256_xor_si256( x0, x2 );
-        x1 = _mm256_xor_si256( x1, x3 );
-        x2 = _mm256_shuffle_epi32( x2, 0xb1 );
-        x3 = _mm256_shuffle_epi32( x3, 0xb1 );
-    }
-
-    _mm256_store_si256( (__m256i*)sp->x,     x0 );
-    _mm256_store_si256( (__m256i*)sp->x + 1, x1 );
-    _mm256_store_si256( (__m256i*)sp->x + 2, x2 );
-    _mm256_store_si256( (__m256i*)sp->x + 3, x3 );
-
-#else
-    __m128i x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3;
-
-    x0 = _mm_load_si128( (__m128i*)sp->x     );
-    x1 = _mm_load_si128( (__m128i*)sp->x + 1 );
-    x2 = _mm_load_si128( (__m128i*)sp->x + 2 );
-    x3 = _mm_load_si128( (__m128i*)sp->x + 3 );
-    x4 = _mm_load_si128( (__m128i*)sp->x + 4 );
-    x5 = _mm_load_si128( (__m128i*)sp->x + 5 );
-    x6 = _mm_load_si128( (__m128i*)sp->x + 6 );
-    x7 = _mm_load_si128( (__m128i*)sp->x + 7 );
-
-    for (r = 0; r < rounds; ++r) {
-	x4 = _mm_add_epi32(x0, x4);
-	x5 = _mm_add_epi32(x1, x5);
-	x6 = _mm_add_epi32(x2, x6);
-	x7 = _mm_add_epi32(x3, x7);
-	y0 = x2;
-	y1 = x3;
-	y2 = x0;
-	y3 = x1;
-	x0 = _mm_xor_si128(_mm_slli_epi32(y0, 7), _mm_srli_epi32(y0, 25));
-	x1 = _mm_xor_si128(_mm_slli_epi32(y1, 7), _mm_srli_epi32(y1, 25));
-	x2 = _mm_xor_si128(_mm_slli_epi32(y2, 7), _mm_srli_epi32(y2, 25));
-	x3 = _mm_xor_si128(_mm_slli_epi32(y3, 7), _mm_srli_epi32(y3, 25));
-	x0 = _mm_xor_si128(x0, x4);
-	x1 = _mm_xor_si128(x1, x5);
-	x2 = _mm_xor_si128(x2, x6);
-	x3 = _mm_xor_si128(x3, x7);
-	x4 = _mm_shuffle_epi32(x4, 0x4e);
-	x5 = _mm_shuffle_epi32(x5, 0x4e);
-	x6 = _mm_shuffle_epi32(x6, 0x4e);
-	x7 = _mm_shuffle_epi32(x7, 0x4e);
-	x4 = _mm_add_epi32(x0, x4);
-	x5 = _mm_add_epi32(x1, x5);
-	x6 = _mm_add_epi32(x2, x6);
-	x7 = _mm_add_epi32(x3, x7);
-	y0 = x1;
-	y1 = x0;
-	y2 = x3;
-	y3 = x2;
-	x0 = _mm_xor_si128(_mm_slli_epi32(y0, 11), _mm_srli_epi32(y0, 21));
-	x1 = _mm_xor_si128(_mm_slli_epi32(y1, 11), _mm_srli_epi32(y1, 21));
-	x2 = _mm_xor_si128(_mm_slli_epi32(y2, 11), _mm_srli_epi32(y2, 21));
-	x3 = _mm_xor_si128(_mm_slli_epi32(y3, 11), _mm_srli_epi32(y3, 21));
-	x0 = _mm_xor_si128(x0, x4);
-	x1 = _mm_xor_si128(x1, x5);
-	x2 = _mm_xor_si128(x2, x6);
-	x3 = _mm_xor_si128(x3, x7);
-	x4 = _mm_shuffle_epi32(x4, 0xb1);
-	x5 = _mm_shuffle_epi32(x5, 0xb1);
-	x6 = _mm_shuffle_epi32(x6, 0xb1);
-	x7 = _mm_shuffle_epi32(x7, 0xb1);
-    }
-
-    _mm_store_si128( (__m128i*)sp->x,     x0 );
-    _mm_store_si128( (__m128i*)sp->x + 1, x1 );
-    _mm_store_si128( (__m128i*)sp->x + 2, x2 );
-    _mm_store_si128( (__m128i*)sp->x + 3, x3 );
-    _mm_store_si128( (__m128i*)sp->x + 4, x4 );
-    _mm_store_si128( (__m128i*)sp->x + 5, x5 );
-    _mm_store_si128( (__m128i*)sp->x + 6, x6 );
-    _mm_store_si128( (__m128i*)sp->x + 7, x7 );
-
-#endif
-}  // transform
-
-// Cubehash context initializing is very expensive.
-// Cache the intial value for faster reinitializing.
-cubehashParam cube_ctx_cache __attribute__ ((aligned (64)));
-
-int cubehashReinit( cubehashParam *sp )
-{
-   memcpy( sp, &cube_ctx_cache, sizeof(cubehashParam) );
-   return SUCCESS;
-
-}
-
-// Initialize the cache then copy to sp.
-int cubehashInit(cubehashParam *sp, int hashbitlen, int rounds, int blockbytes)
-{
-    int i;
-
-    if ( hashbitlen < 8 ) return BAD_HASHBITLEN;
-    if ( hashbitlen > 512 ) return BAD_HASHBITLEN;
-    if ( hashbitlen != 8 * (hashbitlen / 8) ) return BAD_HASHBITLEN;
-
-    /* Sanity checks */
-    if ( rounds <= 0 || rounds > 32 )
-       rounds = CUBEHASH_ROUNDS;
-    if ( blockbytes <= 0 || blockbytes >= 256)
-       blockbytes = CUBEHASH_BLOCKBYTES;
-
-    // all sizes of __m128i
-    cube_ctx_cache.hashlen   = hashbitlen/128;
-    cube_ctx_cache.blocksize = blockbytes/16;
-    cube_ctx_cache.rounds    = rounds;
-    cube_ctx_cache.pos       = 0;
-
-    for ( i = 0; i < 8; ++i )
-       cube_ctx_cache.x[i] = _mm_setzero_si128();;
-
-    cube_ctx_cache.x[0] = _mm_set_epi32( 0, rounds, blockbytes,
-                                         hashbitlen / 8 );
-
-    for ( i = 0; i < 10; ++i )
-       transform( &cube_ctx_cache );
-
-    memcpy( sp, &cube_ctx_cache, sizeof(cubehashParam) );
-    return SUCCESS;
-}
-
-int cubehashUpdate( cubehashParam *sp, const byte *data, size_t size )
-{
-    const int len = size / 16;
-    const __m128i* in = (__m128i*)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->x[ sp->pos ] = _mm_xor_si128( sp->x[ sp->pos ], in[i] );
-        sp->pos++;
-        if ( sp->pos == sp->blocksize )
-        {
-           transform( sp );
-           sp->pos = 0;
-        }
-    }
-
-    return SUCCESS;
-}
-
-int cubehashDigest( cubehashParam *sp, byte *digest )
-{
-    __m128i* hash = (__m128i*)digest;
-    int i;
-
-    // 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 ) );
-    transform( sp );
-
-    sp->x[7] = _mm_xor_si128( sp->x[7], _mm_set_epi32( 1,0,0,0 ) );
-    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 cubehashUpdateDigest( cubehashParam *sp, byte *digest,
-                          const byte *data, size_t size )
-{
-    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 ],
-                                      _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                                    0,0,0,0, 0,0,0,0x80 ) );
-    transform( sp );
-
-    sp->x[7] = _mm_xor_si128( sp->x[7], _mm_set_epi32( 1,0,0,0 ) );
-    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;
-}
-

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};
@@ -60,336 +53,207 @@ MYALIGN const unsigned int	zero[]			= {0x00000000, 0x00000000, 0x00000000, 0x000
 MYALIGN const unsigned int	mul2ipt[]		= {0x728efc00, 0x6894e61a, 0x3fc3b14d, 0x25d9ab57, 0xfd5ba600, 0x2a8c71d7, 0x1eb845e3, 0xc96f9234};
 
 
-//#include "crypto_hash.h"
+#define ECHO_SUBBYTES4(state, j) \
+   state[0][j] = _mm_aesenc_si128(state[0][j], k1);\
+   k1 = _mm_add_epi32(k1, M128(const1));\
+   state[1][j] = _mm_aesenc_si128(state[1][j], k1);\
+   k1 = _mm_add_epi32(k1, M128(const1));\
+   state[2][j] = _mm_aesenc_si128(state[2][j], k1);\
+   k1 = _mm_add_epi32(k1, M128(const1));\
+   state[3][j] = _mm_aesenc_si128(state[3][j], k1);\
+   k1 = _mm_add_epi32(k1, M128(const1));\
+   state[0][j] = _mm_aesenc_si128(state[0][j], m128_zero ); \
+   state[1][j] = _mm_aesenc_si128(state[1][j], m128_zero ); \
+   state[2][j] = _mm_aesenc_si128(state[2][j], m128_zero ); \
+   state[3][j] = _mm_aesenc_si128(state[3][j], m128_zero )
 
- int crypto_hash(
-   unsigned char *out,
-   const unsigned char *in,
-   unsigned long long inlen
- )
- {
-
-	 if(hash_echo(512, in, inlen * 8, out) == SUCCESS) 
-		 return 0;
-	 
-	 return -1;
- }
-
-/*
-int main()
-{
-	return 0;
-}
-*/
-
-#if 0
-void DumpState(__m128i *ps)
-{
-	int i, j, k;
-	unsigned int ucol;
-
-	for(j = 0; j < 4; j++)
-	{
-		for(i = 0; i < 4; i++)
-		{
-			printf("row %d,col %d : ", i, j);
-			for(k = 0; k < 4; k++)
-			{
-				ucol = *((int*)ps + 16 * i + 4 * j + k);
-				printf("%02x%02x%02x%02x ", (ucol >> 0) & 0xff, (ucol >> 8) & 0xff, (ucol >> 16) & 0xff, (ucol >> 24) & 0xff);
-			}
-
-			printf("\n");
-		}
-	}
-
-	printf("\n");
-}
-#endif
-
-
-
-
-#ifndef NO_AES_NI
 #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))
-#else
-#define ECHO_SUBBYTES(state, i, j) \
-				AES_ROUND_VPERM(state[i][j], t1, t2, t3, t4, s1, s2, s3);\
-				state[i][j] = _mm_xor_si128(state[i][j], k1);\
-				AES_ROUND_VPERM(state[i][j], t1, t2, t3, t4, s1, s2, s3);\
-				k1 = _mm_add_epi32(k1, M128(const1))
-
-#define ECHO_SUB_AND_MIX(state, i, j, state2, c, r1, r2, r3, r4) \
-				AES_ROUND_VPERM_CORE(state[i][j], t1, t2, t3, t4, s1, s2, s3);\
-				ktemp = k1;\
-				TRANSFORM(ktemp, _k_ipt, t1, t4);\
-				state[i][j] = _mm_xor_si128(state[i][j], ktemp);\
-				AES_ROUND_VPERM_CORE(state[i][j], t1, t2, t3, t4, s1, s2, s3);\
-				k1 = _mm_add_epi32(k1, M128(const1));\
-				s1 = state[i][j];\
-				s2 = s1;\
-				TRANSFORM(s2, mul2ipt, t1, t2);\
-				s3 = _mm_xor_si128(s1, s2);\
-				state2[r1][c] = _mm_xor_si128(state2[r1][c], s2);\
-				state2[r2][c] = _mm_xor_si128(state2[r2][c], s1);\
-				state2[r3][c] = _mm_xor_si128(state2[r3][c], s1);\
-				state2[r4][c] = _mm_xor_si128(state2[r4][c], s3)
-
-
-
-#endif
-
+	state[i][j] = _mm_aesenc_si128(state[i][j], k1);\
+   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]);\
-				t1 = _mm_srli_epi16(state1[0][j], 7);\
-				t1 = _mm_and_si128(t1, M128(lsbmask));\
-				t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
-				s2 = _mm_xor_si128(s2, t2);\
-				state2[0][j] = s2;\
-				state2[1][j] = state1[0][j];\
-				state2[2][j] = state1[0][j];\
-				state2[3][j] = _mm_xor_si128(s2, state1[0][j]);\
-				s2 = _mm_add_epi8(state1[1][(j + 1) & 3], state1[1][(j + 1) & 3]);\
-				t1 = _mm_srli_epi16(state1[1][(j + 1) & 3], 7);\
-				t1 = _mm_and_si128(t1, M128(lsbmask));\
-				t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
-				s2 = _mm_xor_si128(s2, t2);\
-				state2[0][j] = _mm_xor_si128(state2[0][j], _mm_xor_si128(s2, state1[1][(j + 1) & 3]));\
-				state2[1][j] = _mm_xor_si128(state2[1][j], s2);\
-				state2[2][j] = _mm_xor_si128(state2[2][j], state1[1][(j + 1) & 3]);\
-				state2[3][j] = _mm_xor_si128(state2[3][j], state1[1][(j + 1) & 3]);\
-				s2 = _mm_add_epi8(state1[2][(j + 2) & 3], state1[2][(j + 2) & 3]);\
-				t1 = _mm_srli_epi16(state1[2][(j + 2) & 3], 7);\
-				t1 = _mm_and_si128(t1, M128(lsbmask));\
-				t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
-				s2 = _mm_xor_si128(s2, t2);\
-				state2[0][j] = _mm_xor_si128(state2[0][j], state1[2][(j + 2) & 3]);\
-				state2[1][j] = _mm_xor_si128(state2[1][j], _mm_xor_si128(s2, state1[2][(j + 2) & 3]));\
-				state2[2][j] = _mm_xor_si128(state2[2][j], s2);\
-				state2[3][j] = _mm_xor_si128(state2[3][j], state1[2][(j + 2) & 3]);\
-				s2 = _mm_add_epi8(state1[3][(j + 3) & 3], state1[3][(j + 3) & 3]);\
-				t1 = _mm_srli_epi16(state1[3][(j + 3) & 3], 7);\
-				t1 = _mm_and_si128(t1, M128(lsbmask));\
-				t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
-				s2 = _mm_xor_si128(s2, t2);\
-				state2[0][j] = _mm_xor_si128(state2[0][j], state1[3][(j + 3) & 3]);\
-				state2[1][j] = _mm_xor_si128(state2[1][j], state1[3][(j + 3) & 3]);\
-				state2[2][j] = _mm_xor_si128(state2[2][j], _mm_xor_si128(s2, state1[3][(j + 3) & 3]));\
-				state2[3][j] = _mm_xor_si128(state2[3][j], s2)
+	s2 = _mm_add_epi8(state1[0][j], state1[0][j]);\
+	t1 = _mm_srli_epi16(state1[0][j], 7);\
+	t1 = _mm_and_si128(t1, M128(lsbmask));\
+	t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
+	s2 = _mm_xor_si128(s2, t2);\
+	state2[0][j] = s2;\
+	state2[1][j] = state1[0][j];\
+	state2[2][j] = state1[0][j];\
+	state2[3][j] = _mm_xor_si128(s2, state1[0][j]);\
+	s2 = _mm_add_epi8(state1[1][(j + 1) & 3], state1[1][(j + 1) & 3]);\
+	t1 = _mm_srli_epi16(state1[1][(j + 1) & 3], 7);\
+	t1 = _mm_and_si128(t1, M128(lsbmask));\
+	t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
+	s2 = _mm_xor_si128(s2, t2);\
+	state2[0][j] = mm128_xor3(state2[0][j], s2, state1[1][(j + 1) & 3] );\
+	state2[1][j] = _mm_xor_si128(state2[1][j], s2);\
+	state2[2][j] = _mm_xor_si128(state2[2][j], state1[1][(j + 1) & 3]);\
+	state2[3][j] = _mm_xor_si128(state2[3][j], state1[1][(j + 1) & 3]);\
+	s2 = _mm_add_epi8(state1[2][(j + 2) & 3], state1[2][(j + 2) & 3]);\
+	t1 = _mm_srli_epi16(state1[2][(j + 2) & 3], 7);\
+	t1 = _mm_and_si128(t1, M128(lsbmask));\
+	t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
+	s2 = _mm_xor_si128(s2, t2);\
+	state2[0][j] = _mm_xor_si128(state2[0][j], state1[2][(j + 2) & 3]);\
+	state2[1][j] = mm128_xor3(state2[1][j], s2, state1[2][(j + 2) & 3] );\
+	state2[2][j] = _mm_xor_si128(state2[2][j], s2);\
+	state2[3][j] = _mm_xor_si128(state2[3][j], state1[2][(j + 2) & 3]);\
+	s2 = _mm_add_epi8(state1[3][(j + 3) & 3], state1[3][(j + 3) & 3]);\
+	t1 = _mm_srli_epi16(state1[3][(j + 3) & 3], 7);\
+	t1 = _mm_and_si128(t1, M128(lsbmask));\
+	t2 = _mm_shuffle_epi8(M128(mul2mask), t1);\
+	s2 = _mm_xor_si128(s2, t2);\
+	state2[0][j] = _mm_xor_si128(state2[0][j], state1[3][(j + 3) & 3]);\
+	state2[1][j] = _mm_xor_si128(state2[1][j], state1[3][(j + 3) & 3]);\
+	state2[2][j] = mm128_xor3(state2[2][j], s2, state1[3][(j + 3) & 3] );\
+	state2[3][j] = _mm_xor_si128(state2[3][j], s2)
 
 
 #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)
+   ECHO_SUBBYTES4(_state, 0);\
+   ECHO_SUBBYTES4(_state, 1);\
+   ECHO_SUBBYTES4(_state, 2);\
+   ECHO_SUBBYTES4(_state, 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_SUBBYTES4(_state2, 0);\
+   ECHO_SUBBYTES4(_state2, 1);\
+   ECHO_SUBBYTES4(_state2, 2);\
+   ECHO_SUBBYTES4(_state2, 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 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]
+	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]
 
 
 void Compress(hashState_echo *ctx, const unsigned char *pmsg, unsigned int uBlockCount)
 {
-	unsigned int r, b, i, j;
-//      __m128i t1, t2, t3, t4, s1, s2, s3, k1, ktemp;
-	__m128i t1, t2, s2, k1;
-	__m128i _state[4][4], _state2[4][4], _statebackup[4][4]; 
+   unsigned int r, b, i, j;
+   __m128i t1, t2, s2, k1;
+   __m128i _state[4][4], _state2[4][4], _statebackup[4][4]; 
 
+   for(i = 0; i < 4; i++)
+	for(j = 0; j < ctx->uHashSize / 256; j++)
+		_state[i][j] = ctx->state[i][j];
 
-	for(i = 0; i < 4; i++)
-		for(j = 0; j < ctx->uHashSize / 256; j++)
-			_state[i][j] = ctx->state[i][j];
+   for(b = 0; b < uBlockCount; b++)
+   {
+   	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++)
+	      {
+		     _state[i][j] = _mm_load_si128((__m128i*)pmsg + 4 * (j - (ctx->uHashSize / 256)) + i);
+	      }
+	   }
 
-#ifdef NO_AES_NI
-	// transform cv
-	for(i = 0; i < 4; i++)
-		for(j = 0; j < ctx->uHashSize / 256; j++)
-		{
-			TRANSFORM(_state[i][j], _k_ipt, t1, t2);
-		}
-#endif
+	   // save state
+	   SAVESTATE(_statebackup, _state);
 
-	for(b = 0; b < uBlockCount; b++)
-	{
-		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++)
-			{
-				_state[i][j] = _mm_loadu_si128((__m128i*)pmsg + 4 * (j - (ctx->uHashSize / 256)) + i);
-
-#ifdef NO_AES_NI
-				// transform message
-				TRANSFORM(_state[i][j], _k_ipt, t1, t2);
-#endif
-			}
-		}
-
-		// save state
-		SAVESTATE(_statebackup, _state);
-
-
-		k1 = ctx->k;
-
-#ifndef NO_AES_NI
-		for(r = 0; r < ctx->uRounds / 2; r++)
-		{
-			ECHO_ROUND_UNROLL2;
-		}
-
-#else
-		for(r = 0; r < ctx->uRounds / 2; r++)
-		{
-			_state2[0][0] = M128(zero); _state2[1][0] = M128(zero); _state2[2][0] = M128(zero); _state2[3][0] = M128(zero);
-			_state2[0][1] = M128(zero); _state2[1][1] = M128(zero); _state2[2][1] = M128(zero); _state2[3][1] = M128(zero);
-			_state2[0][2] = M128(zero); _state2[1][2] = M128(zero); _state2[2][2] = M128(zero); _state2[3][2] = M128(zero);
-			_state2[0][3] = M128(zero); _state2[1][3] = M128(zero); _state2[2][3] = M128(zero); _state2[3][3] = M128(zero);																			
-
-			ECHO_SUB_AND_MIX(_state, 0, 0, _state2, 0, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state, 1, 0, _state2, 3, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state, 2, 0, _state2, 2, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state, 3, 0, _state2, 1, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state, 0, 1, _state2, 1, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state, 1, 1, _state2, 0, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state, 2, 1, _state2, 3, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state, 3, 1, _state2, 2, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state, 0, 2, _state2, 2, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state, 1, 2, _state2, 1, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state, 2, 2, _state2, 0, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state, 3, 2, _state2, 3, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state, 0, 3, _state2, 3, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state, 1, 3, _state2, 2, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state, 2, 3, _state2, 1, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state, 3, 3, _state2, 0, 3, 0, 1, 2);
-
-			_state[0][0] = M128(zero); _state[1][0] = M128(zero); _state[2][0] = M128(zero); _state[3][0] = M128(zero);
-			_state[0][1] = M128(zero); _state[1][1] = M128(zero); _state[2][1] = M128(zero); _state[3][1] = M128(zero);
-			_state[0][2] = M128(zero); _state[1][2] = M128(zero); _state[2][2] = M128(zero); _state[3][2] = M128(zero);
-			_state[0][3] = M128(zero); _state[1][3] = M128(zero); _state[2][3] = M128(zero); _state[3][3] = M128(zero);																			
-
-			ECHO_SUB_AND_MIX(_state2, 0, 0, _state, 0, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state2, 1, 0, _state, 3, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state2, 2, 0, _state, 2, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state2, 3, 0, _state, 1, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state2, 0, 1, _state, 1, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state2, 1, 1, _state, 0, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state2, 2, 1, _state, 3, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state2, 3, 1, _state, 2, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state2, 0, 2, _state, 2, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state2, 1, 2, _state, 1, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state2, 2, 2, _state, 0, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state2, 3, 2, _state, 3, 3, 0, 1, 2);
-			ECHO_SUB_AND_MIX(_state2, 0, 3, _state, 3, 0, 1, 2, 3);
-			ECHO_SUB_AND_MIX(_state2, 1, 3, _state, 2, 1, 2, 3, 0);
-			ECHO_SUB_AND_MIX(_state2, 2, 3, _state, 1, 2, 3, 0, 1);
-			ECHO_SUB_AND_MIX(_state2, 3, 3, _state, 0, 3, 0, 1, 2);
-
-		}
-#endif
+	   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] = _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;
-	}
-
-#ifdef NO_AES_NI
-	// transform state
-	for(i = 0; i < 4; i++)
-		for(j = 0; j < 4; j++)
-		{
-			TRANSFORM(_state[i][j], _k_opt, t1, t2);
-		}
-#endif
-
-		SAVESTATE(ctx->state, _state);
+	   if(ctx->uHashSize == 256)
+	   {
+	      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;
+   }
+	SAVESTATE(ctx->state, _state);
 
 }
 
@@ -552,13 +416,13 @@ 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;
@@ -675,18 +539,177 @@ 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)
 {

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"
@@ -30,6 +30,7 @@
 typedef struct
 {
 	__m128i			state[4][4];
+        BitSequence             buffer[192];
 	__m128i			k;
 	__m128i			hashsize;
 	__m128i			const1536;
@@ -39,9 +40,8 @@ typedef struct
 	unsigned int	uBlockLength;
 	unsigned int	uBufferBytes;
 	DataLength		processed_bits;
-	BitSequence		buffer[192];
 
-} hashState_echo;
+} hashState_echo __attribute__ ((aligned (64)));
 
 HashReturn init_echo(hashState_echo *state, int hashbitlen);
 
@@ -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