v3.14.3

v3.14.2
v3.14.1
2025-09-17 23:44:27 +00:00 · 2020-06-18 17:30:26 -04:00 · 2020-05-30 21:20:44 -04:00 · 2020-05-21 13:00:29 -04:00 · 2020-05-20 13:56:35 -04:00 · 2020-05-18 18:17:27 -04:00
270 changed files with 12243 additions and 29032 deletions
--- a/3
+++ b/3
@@ -33,3 +33,6 @@ Jay D Dee
 xcouiz@gmail.com

 Cryply
+
+Colin Percival
+Alexander Peslyak
--- a/Makefile.am
+++ b/Makefile.am
@@ -21,15 +21,6 @@ cpuminer_SOURCES = \
  api.c \
  sysinfos.c \
  algo-gate-api.c\
-  crypto/oaes_lib.c \
-  crypto/c_keccak.c \
-  crypto/c_groestl.c \
-  crypto/c_blake256.c \
-  crypto/c_jh.c \
-  crypto/c_skein.c \
-  crypto/hash.c \
-  crypto/aesb.c \
-  crypto/magimath.cpp \
  algo/argon2/argon2a/argon2a.c \
  algo/argon2/argon2a/ar2/argon2.c \
  algo/argon2/argon2a/ar2/opt.c \
@@ -76,11 +67,6 @@ cpuminer_SOURCES = \
  algo/bmw/bmw512-gate.c \
  algo/bmw/bmw512.c \
  algo/bmw/bmw512-4way.c \
-  algo/cryptonight/cryptolight.c \
-  algo/cryptonight/cryptonight-common.c\
-  algo/cryptonight/cryptonight-aesni.c\
-  algo/cryptonight/cryptonight.c\
-  algo/cubehash/sph_cubehash.c \
  algo/cubehash/cubehash_sse2.c\
  algo/cubehash/cube-hash-2way.c \
  algo/echo/sph_echo.c \
@@ -103,9 +89,6 @@ cpuminer_SOURCES = \
  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 \
@@ -121,9 +104,9 @@ cpuminer_SOURCES = \
  algo/keccak/keccak-hash-4way.c \
  algo/keccak/keccak-4way.c\
  algo/keccak/keccak-gate.c \
+  algo/keccak/sha3d-4way.c \
+  algo/keccak/sha3d.c \
  algo/lanehash/lane.c \
-  algo/luffa/sph_luffa.c \
-  algo/luffa/luffa.c \
  algo/luffa/luffa_for_sse2.c \
  algo/luffa/luffa-hash-2way.c \
  algo/lyra2/lyra2.c \
@@ -145,14 +128,14 @@ cpuminer_SOURCES = \
  algo/lyra2/allium.c \
  algo/lyra2/phi2-4way.c \
  algo/lyra2/phi2.c \
-  algo/m7m.c \
+  algo//m7m/m7m.c \
+  algo/m7m/magimath.cpp \
  algo/nist5/nist5-gate.c \
  algo/nist5/nist5-4way.c \
  algo/nist5/nist5.c \
  algo/nist5/zr5.c \
  algo/panama/panama-hash-4way.c \
  algo/panama/sph_panama.c \
-  algo/radiogatun/sph_radiogatun.c \
  algo/quark/quark-gate.c \
  algo/quark/quark.c \
  algo/quark/quark-4way.c \
@@ -175,11 +158,12 @@ cpuminer_SOURCES = \
  algo/ripemd/lbry-4way.c \
  algo/scrypt/scrypt.c \
  algo/scrypt/neoscrypt.c \
-  algo/scrypt/pluck.c \
  algo/sha/sph_sha2.c \
  algo/sha/sph_sha2big.c \
  algo/sha/sha256-hash-4way.c \
  algo/sha/sha512-hash-4way.c \
+  algo/sha/hmac-sha256-hash.c \
+  algo/sha/hmac-sha256-hash-4way.c \
  algo/sha/sha2.c \
  algo/sha/sha256t-gate.c \
  algo/sha/sha256t-4way.c \
@@ -193,7 +177,6 @@ cpuminer_SOURCES = \
  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 \
@@ -231,7 +214,6 @@ 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 \
@@ -250,7 +232,6 @@ cpuminer_SOURCES = \
  algo/x13/skunk-gate.c \
  algo/x13/skunk-4way.c \
  algo/x13/skunk.c \
-  algo/x13/drop.c \
  algo/x13/x13bcd-4way.c \
  algo/x13/x13bcd.c \
  algo/x14/x14-gate.c \
@@ -276,6 +257,7 @@ cpuminer_SOURCES = \
  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 \
@@ -285,19 +267,17 @@ cpuminer_SOURCES = \
  algo/x17/sonoa-gate.c \
  algo/x17/sonoa-4way.c \
  algo/x17/sonoa.c \
-  algo/x20/x20r.c \
  algo/x22/x22i-4way.c \
  algo/x22/x22i.c \
  algo/x22/x22i-gate.c \
  algo/x22/x25x.c \
  algo/x22/x25x-4way.c \
  algo/yescrypt/yescrypt.c \
-  algo/yescrypt/sha256_Y.c \
  algo/yescrypt/yescrypt-best.c \
  algo/yespower/yespower-gate.c \
  algo/yespower/yespower-blake2b.c \
  algo/yespower/crypto/blake2b-yp.c \
-  algo/yespower/sha256_p.c \
+  algo/yespower/yescrypt-r8g.c \
  algo/yespower/yespower-opt.c

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

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

 mailto://jayddee246@gmail.com

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

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

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

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

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

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

 MacOS, OSx and Android are not supported.

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

 Supported Algorithms
 --------------------
@@ -53,7 +67,6 @@ Supported Algorithms
                          argon2d500    argon2d-dyn,  Dynamic (DYN)
                          argon2d4096   argon2d-uis, Unitus, (UIS)
                          axiom         Shabal-256 MemoHash
-                          bastion
                          blake         Blake-256 (SFR)
                          blake2b       Blake2b 256
                          blake2s       Blake-2 S
@@ -64,10 +77,7 @@ Supported Algorithms
                          decred
                          deep          Deepcoin (DCN)
                          dmd-gr        Diamond-Groestl
-                          drop          Dropcoin
-                          fresh         Fresh
                          groestl       Groestl coin
-                          heavy         Heavy
                          hex           x16r-hex
                          hmq1725       Espers
                          hodl          Hodlcoin
@@ -83,6 +93,7 @@ Supported Algorithms
                          lyra2z        
                          lyra2z330     Lyra2 330 rows, Zoin (ZOI)
                          m7m           Magi (XMG)
+                          minotaur      Ringcoin (RNG)
                          myr-gr        Myriad-Groestl
                          neoscrypt     NeoScrypt(128, 2, 1)
                          nist5         Nist5
@@ -97,10 +108,10 @@ Supported Algorithms
                          qubit         Qubit
                          scrypt        scrypt(1024, 1, 1) (default)
                          scrypt:N      scrypt(N, 1, 1)
-                          scryptjane:nf
                          sha256d       Double SHA-256
                          sha256q       Quad SHA-256, Pyrite (PYE)
                          sha256t       Triple SHA-256, Onecoin (OC)
+                          sha3d         Double keccak256 (BSHA3)
                          shavite3      Shavite3
                          skein         Skein+Sha (Skeincoin)
                          skein2        Double Skein (Woodcoin)
@@ -134,6 +145,7 @@ Supported Algorithms
                          xevan         Bitsend (BSD)
                          yescrypt      Globalboost-Y (BSTY)
                          yescryptr8    BitZeny (ZNY)
+                          yescryptr8g   Koto (KOTO)
                          yescryptr16   Eli
                          yescryptr32   WAVI
                          yespower      Cryply
@@ -141,6 +153,27 @@ Supported Algorithms
                          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
 ------

--- a/README.txt
+++ b/README.txt
@@ -1,8 +1,8 @@
 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 is a console program that is executed from a DOS or Powershell
+prompt. There is no GUI and no mouse support.

 Miner programs are often flagged as malware by antivirus programs. This is
 a false positive, they are flagged simply because they are cryptocurrency 
@@ -15,8 +15,8 @@ the features listed at cpuminer startup to ensure you are mining at
 optimum speed using the best available features.

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

 AMD CPUs older than Piledriver, including Athlon x2 and Phenom II x4, are not
 supported by cpuminer-opt due to an incompatible implementation of SSE2 on
@@ -31,14 +31,29 @@ https://en.wikipedia.org/wiki/List_of_Intel_CPU_microarchitectures
 https://en.wikipedia.org/wiki/List_of_AMD_CPU_microarchitectures


-Exe name                Compile flags            Arch name
+Exe file name              Compile flags              Arch name

-cpuminer-sse2.exe      "-msse2"                  Core2, Nehalem   
-cpuminer-aes-sse42.exe "-march=westmere"         Westmere
-cpuminer-avx.exe       "-march=corei7-avx"       Sandybridge
-cpuminer-avx2.exe      "-march=core-avx2 -maes"  Haswell, Skylake, Coffeelake
-cpuminer-avx512.exe    "-march=skylake-avx512"   Skylake-X, Cascadelake-X
-cpuminer-zen           "-march=znver1"           AMD Ryzen, Threadripper
+cpuminer-sse2.exe            "-msse2"                 Core2, Nehalem   
+cpuminer-aes-sse42.exe       "-march=westmere"        Westmere
+cpuminer-avx.exe             "-march=corei7-avx"      Sandybridge, Ivybridge
+cpuminer-avx2.exe            "-march=core-avx2 -maes" Haswell*
+cpuminer-avx512.exe          "-march=skylake-avx512"  Skylake-X, Cascadelake-X
+cpuminer-zen.exe             "-march=znver1"          AMD Ryzen, Threadripper
+cpuminer-avx512-sha-vaes.exe "-march=icelake-client"  Icelake*
+
+* Haswell includes Broadwell, Skylake, Kabylake, Coffeelake & Cometlake. 
+Icelake is only available on some laptops. Mining with a laptop is not
+recommended. The icelake build is included in anticipation of Intel eventually
+releasing a desktop CPU with a microarchitecture newer than Skylake.
+
+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 instalation or compiled by me from
+source code obtained from the author's official repository. The exact
+procedure is documented in the build instructions for Windows:
+https://github.com/JayDDee/cpuminer-opt/wiki/Compiling-from-source

 If you like this software feel free to donate:

--- a/289
+++ b/289
@@ -65,6 +65,295 @@ If not what makes it happen or not happen?
 Change Log
 ----------

+v3.14.3
+
+#265: more mutex changes to reduce blocking with high thread count.
+
+#267: fixed hodl algo potential memory alignment issue,
+      add warning when thread count is not valid for mining hodl algo.
+
+v3.14.2
+
+The second line of the Share Accepted log is no longer displayed,
+new Xnonce log is added and other small log tweaks.
+
+#265: Cleanup use of mutex.
+
+v3.14.1
+
+GBT and getwork log changes:
+ fixed missing TTF in New Block log,
+ ntime no longer byte-swapped for display in New Work log,
+ fixed zero effective hash rate in Periodic Report log,
+ deleted "Current block is..." log.
+
+Renamed stratum "New Job" log to "New Work" to be consistent with the solo
+version of the log. Added more data to both versions.
+
+v3.14.0
+
+Changes to solo mining:
+  - segwit is supported by getblocktemplate,
+  - longpolling is not working and is disabled,
+  - Periodic Report log is output,
+  - New Block log includes TTF estimates,
+  - Stratum thread no longer created when using getwork or GBT.
+
+Fixed BUG log mining sha256d.
+
+v3.13.1.1
+
+Fixed Windows crash mining minotaur algo.
+
+Fixed GCC 10 compile again.
+Added -fno-common to testing to be consistent with GCC 10 default.
+
+v3.13.1
+
+Added minotaur algo for Ringcoin.
+
+v3.13.0.1
+
+Issue #262: Fixed xevan AVX2 invalid shares.
+
+v3.13.0
+
+Updated Windows binaries compiled with GCC 9. Included DLLs also updated.
+Icelake build (cpuminer-avx512-sha-vaes.exe) now included in Windows
+binaries package.
+
+No source code changes.
+
+v3.12.8.2
+
+Fixed x12 AVX2 rejects.
+Fixed phi AVX2 crash.
+
+v3.12.8.1
+
+Issue #261: Fixed yescryptr8g invalid shares.
+
+v3.12.8
+
+Yespower sha256 prehash made thread safe.
+
+Rewrote diff conversion functions from scratch to be simpler and use 
+long double (float80) and int128 arithmetic for improved accuracy and
+precision.
+
+Some code cleanup and assorted small changes.
+
+v3.12.7
+
+Issue #257: fixed a file descriptor leak which caused the CPU temperature
+and frequency query to report zeros after mining for a couple of hours.
+
+Issue #253: stale share reduction for yescrypt, sonoa.
+
+v3.12.6.1
+
+Issue #252: Fixed SSL mining (stratum+tcps://)
+
+Issue #254 Fixed benchmark.
+
+Issue #253: Implemented stale share reduction for yespower, x25x, x22i, x21s,
+x16*, scryptn2, more to come.
+
+v3.12.6
+
+Issue #246: improved stale share detection for getwork.
+
+Improved precision of target_to_diff conversion from 4 digits to 20+.
+
+Display hash and target debug data for all rejected shares.
+
+A graphical representation of CPU affinity is displayed when using --threads.
+
+Added highest and lowest accepted share to summary log.
+
+Other small changes to logs to improve consistency and clarity.
+
+v3.12.5
+
+Issues #246 & #251: fixed incorrect share diff for stratum and getwork,
+fixed incorrect target diff for getwork. Stats should now be correct for
+getwork as well as stratum.
+
+Issue #252: Fixed stratum+tcps not using curl ssl.
+
+Getwork: reduce stale blocks, faster response to new work.
+
+Added ntime to new job/work logs.
+
+README.md now lists the parameters for yespower variations that don't have
+a specific algo name.
+
+v3.12.4.6
+
+Issue #246: fixed getwork repeated new block logs with same height. New work
+for the same block is now reported as "New work" instead of "New block".
+Also added a check that work is new before generating "New work" log.
+
+Added target diff to getwork new block log.
+
+Changed share ratio in share result log to simple fraction, no longer %.
+
+Added debug log to display mininginfo, use -D.
+
+v3.12.4.5
+
+Issue #246: better stale share detection for getwork, and enhanced logging
+of stale shares for stratum & getwork.
+
+Issue #251: fixed incorrect share difficulty and share ratio in share
+result log.
+
+Changed submit log to include share diff and block height.
+
+Small cosmetic changes to logs. 
+
+v3.12.4.4
+
+Issue #246: Fixed net hashrate in getwork block log,
+            removed duplicate getwork block log, 
+            other small tweaks to stats logs for getwork.
+
+Issue #248: Fixed chronic stale shares with scrypt:1048576 (scryptn2). 
+
+v3.12.4.3
+
+Fixed segfault in new block log for getwork.
+
+Disabled silent discarding of stale work after the submit is logged.
+
+v3.12.4.2
+
+Issue #245: fixed getwork stale shares, solo mining with getwork now works.
+
+Issue #246: implemented block and summary logs for getwork.
+
+v3.12.4.1
+
+Issue #245: fix scantime when mining solo with getwork.
+
+Added debug logs for creation of stratum and longpoll threads, use -D to
+enable.
+
+v3.12.4
+
+Issue #244: Change longpoll to ignore job id.
+
+Lyra2rev2 AVX2 +3%, AVX512 +6%.
+
+v3.12.3.1
+
+Issue #241: Fixed regression that broke coinbase address in v3.11.7.
+
+v3.12.3
+
+Issue #238: Fixed skunk AVX2.
+
+Issue #239: Faster AVX2 & AVX512 for skein +44%, skein2 +30%, plus marginal
+increases for skunk, x16r, x16rv2, x16rt, x16rt-veil, x16s, x21s.
+
+Faster anime VAES +57%, AVX512 +21%, AVX2 +3%.
+
+Redesigned code reponsible for #236.
+
+v3.12.2
+
+Fixed xevan, skein, skein2 AVX2, #238.
+
+Reversed polarity of AVX2 vector bit test utilities, and all users, to be
+logically and semantically correct. Follow up to issue #236. 
+
+v3.12.1
+
+Fixed anime AVX2 low difficulty shares, git issue #236.
+
+Periodic summary now reports lost hash rate due to rejected and stale shares,
+displayed only when non-zero.
+
+v3.12.0.1
+
+Fixed hodl rejects, git issue #237.
+
+Fixed debug code added in v3.12.0 to work with AVX2 to be enabled only
+after low difficulty share have been seen to avoid unnecessarily excessive
+log outout.
+
+Added more digits of precision to diff in log output to help diagnose
+low difficulty shares.
+
+v3.12.0
+
+Faster phi2 AVX2 +62%, AVX512 +150% on Intel CPUs. AMD Ryzen AVX2 is
+YMMV due to its inferiour AVX2 implementation.
+
+Fixed Hodl stats, rejects are still an issue since v3.9.5, git issue #237.
+
+API can now be enabled with "-b port" or "--api-bind port".
+It will use the default address 127.0.0.1.
+
+Editorial: Short form options should only be used on the command line to save
+typing. Configuration files and scripts should always use the long form
+"--api-bind addr:port" without relying on any defaults. This is a general
+recommendation that applies to all options for any application.
+
+Removed obsolete cryptonight, all variants, and supporting code for more
+size reduction and faster compiling.
+
+Tweaked the timing of the CPU temperature and frequency log (Linux only).
+
+Added some debug code to collect more info aboout low difficulty rejects,
+git issue #236.
+
+v3.11.9
+
+Fixed x16r invalid shares when Luffa was first in hash order.
+
+API is disabled by default.
+
+New startup message for status of stratum connection, API & extranonce.
+
+New log report for CPU temperature, frequency of fastest and slowest cores.
+
+Compile time is a little shorter and binary file size a little smaller
+using conditional compilation..
+
+Removed code for Bastion, Drop, Heavy, Luffa an Pluck algos and other unused
+code.
+
+v3.11.8
+
+Fixed network hashrate showing incorrect data, should be close now.
+
+Fixed compile errors when using GCC 10 with default flag -fno-common.
+
+Faster x16r, x16rv2, x16rt, x16s, x21s, veil, hex with midstate prehash.
+
+Decoupled sapling usage from block version 5 in yescryptr8g.
+
+More detailed data reporting for low difficulty rejected shares.
+
+v3.11.7
+
+Added yescryptr8g algo for KOTO, including support for block version 5.
+
+Added sha3d algo for BSHA3.
+
+Removed memcmp and clean_job checks from get_new_work, now only check job_id.
+
+Small improvement to sha512 and sha256 parallel implementations that don't
+use SHA.
+
+v3.11.6
+
+Fixed CPU temperature regression from v3.11.5.
+
+More improvements to share log. More compact, highlight incremented counter,
+block height when solved, job id when stale.
+
 v3.11.5

 Fixed AVX512 detection that could cause compilation errors on CPUs
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.15.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.16.1 -*- Autoconf -*-

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

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

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

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

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

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

 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-

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

 # AM_CONDITIONAL                                            -*- Autoconf -*-

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

-# Copyright (C) 2004-2017 Free Software Foundation, Inc.
+# Copyright (C) 2004-2018 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
--- a/algo-gate-api.c
+++ b/algo-gate-api.c
@@ -90,34 +90,172 @@ void algo_not_implemented()
 }

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

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

 void init_algo_gate( algo_gate_t* gate )
 {
   gate->miner_thread_init       = (void*)&return_true;
-   gate->scanhash                = (void*)&null_scanhash;
+   gate->scanhash                = (void*)&scanhash_generic;
   gate->hash                    = (void*)&null_hash;
-   gate->hash_suw                = (void*)&null_hash_suw;
   gate->get_new_work            = (void*)&std_get_new_work;
-   gate->get_nonceptr            = (void*)&std_get_nonceptr;
   gate->work_decode             = (void*)&std_le_work_decode;
   gate->decode_extra_data       = (void*)&do_nothing;
   gate->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->submit_getwork_result   = (void*)&std_le_submit_getwork_result;
@@ -129,7 +267,6 @@ void init_algo_gate( algo_gate_t* gate )
   gate->resync_threads          = (void*)&do_nothing;
   gate->do_this_thread          = (void*)&return_true;
   gate->longpoll_rpc_call       = (void*)&std_longpoll_rpc_call;
-   gate->stratum_handle_response = (void*)&std_stratum_handle_response;
   gate->get_work_data_size      = (void*)&std_get_work_data_size;
   gate->optimizations           = EMPTY_SET;
   gate->ntime_index             = STD_NTIME_INDEX;
@@ -162,23 +299,16 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
    case ALGO_ARGON2D500:    register_argon2d_dyn_algo   ( gate ); break;
    case ALGO_ARGON2D4096:   register_argon2d4096_algo   ( gate ); break;
    case ALGO_AXIOM:         register_axiom_algo         ( gate ); break;
-    case ALGO_BASTION:       register_bastion_algo       ( gate ); break;
    case ALGO_BLAKE:         register_blake_algo         ( gate ); break;
    case ALGO_BLAKE2B:       register_blake2b_algo       ( gate ); break;
    case ALGO_BLAKE2S:       register_blake2s_algo       ( gate ); break;
    case ALGO_BLAKECOIN:     register_blakecoin_algo     ( gate ); break;
    case ALGO_BMW512:        register_bmw512_algo        ( gate ); break;
    case ALGO_C11:           register_c11_algo           ( gate ); break;
-    case ALGO_CRYPTOLIGHT:   register_cryptolight_algo   ( gate ); break;
-    case ALGO_CRYPTONIGHT:   register_cryptonight_algo   ( gate ); break;
-    case ALGO_CRYPTONIGHTV7: register_cryptonightv7_algo ( gate ); break;
    case ALGO_DECRED:        register_decred_algo        ( gate ); break;
    case ALGO_DEEP:          register_deep_algo          ( gate ); break;
    case ALGO_DMD_GR:        register_dmd_gr_algo        ( gate ); break;
-    case ALGO_DROP:          register_drop_algo          ( gate ); break;
-    case ALGO_FRESH:         register_fresh_algo         ( gate ); break;
    case ALGO_GROESTL:       register_groestl_algo       ( gate ); break;
-    case ALGO_HEAVY:         register_heavy_algo         ( gate ); break;
    case ALGO_HEX:           register_hex_algo           ( gate ); break;
    case ALGO_HMQ1725:       register_hmq1725_algo       ( gate ); break;
    case ALGO_HODL:          register_hodl_algo          ( gate ); break;
@@ -186,7 +316,6 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
    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;
@@ -194,13 +323,13 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
    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_MINOTAUR:      register_minotaur_algo      ( gate ); break;
    case ALGO_MYR_GR:        register_myriad_algo        ( gate ); break;
    case ALGO_NEOSCRYPT:     register_neoscrypt_algo     ( gate ); break;
    case ALGO_NIST5:         register_nist5_algo         ( gate ); break;
    case ALGO_PENTABLAKE:    register_pentablake_algo    ( gate ); break;
    case ALGO_PHI1612:       register_phi1612_algo       ( gate ); break;
    case ALGO_PHI2:          register_phi2_algo          ( gate ); break;
-    case ALGO_PLUCK:         register_pluck_algo         ( gate ); break;
    case ALGO_POLYTIMOS:     register_polytimos_algo     ( gate ); break;
    case ALGO_POWER2B:       register_power2b_algo       ( gate ); break;
    case ALGO_QUARK:         register_quark_algo         ( gate ); break;
@@ -209,6 +338,7 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
    case ALGO_SHA256D:       register_sha256d_algo       ( gate ); break;
    case ALGO_SHA256Q:       register_sha256q_algo       ( gate ); break;
    case ALGO_SHA256T:       register_sha256t_algo       ( gate ); break;
+    case ALGO_SHA3D:         register_sha3d_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;
@@ -240,13 +370,9 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
    case ALGO_X22I:          register_x22i_algo          ( gate ); break;
    case ALGO_X25X:          register_x25x_algo          ( gate ); break;
    case ALGO_XEVAN:         register_xevan_algo         ( gate ); break;
-/*    case ALGO_YESCRYPT:     register_yescrypt_05_algo     ( gate ); break;
-     case ALGO_YESCRYPTR8:   register_yescryptr8_05_algo   ( gate ); break;
-     case ALGO_YESCRYPTR16:  register_yescryptr16_05_algo  ( gate ); break;
-     case ALGO_YESCRYPTR32:  register_yescryptr32_05_algo  ( gate ); break;
-*/
    case ALGO_YESCRYPT:      register_yescrypt_algo      ( gate ); break;
    case ALGO_YESCRYPTR8:    register_yescryptr8_algo    ( gate ); break;
+    case ALGO_YESCRYPTR8G:   register_yescryptr8g_algo   ( gate ); break;
    case ALGO_YESCRYPTR16:   register_yescryptr16_algo   ( gate ); break;
    case ALGO_YESCRYPTR32:   register_yescryptr32_algo   ( gate ); break;
    case ALGO_YESPOWER:      register_yespower_algo      ( gate ); break;
@@ -270,30 +396,6 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
 // restore warnings
 #pragma GCC diagnostic pop

-// override std defaults with jr2 defaults
-bool register_json_rpc2( algo_gate_t *gate )
-{
-  applog(LOG_WARNING,"\nCryptonight algorithm and variants are no longer");
-  applog(LOG_WARNING,"supported by cpuminer-opt. Shares submitted will");
-  applog(LOG_WARNING,"likely be rejected. Proceed at your own risk.\n");
-
-//  gate->wait_for_diff           = (void*)&do_nothing;
-  gate->get_new_work            = (void*)&jr2_get_new_work;
-  gate->get_nonceptr            = (void*)&jr2_get_nonceptr;
-  gate->stratum_gen_work        = (void*)&jr2_stratum_gen_work;
-  gate->build_stratum_request   = (void*)&jr2_build_stratum_request;
-  gate->submit_getwork_result   = (void*)&jr2_submit_getwork_result;
-  gate->longpoll_rpc_call       = (void*)&jr2_longpoll_rpc_call;
-  gate->work_decode             = (void*)&jr2_work_decode;
-  gate->stratum_handle_response = (void*)&jr2_stratum_handle_response;
-  gate->nonce_index             = JR2_NONCE_INDEX;
-  jsonrpc_2 = true;   // still needed
-  opt_extranonce = false;
-//  have_gbt = false;
-  return true;
- }
-
-// run the alternate hash function for a specific algo
 void exec_hash_function( int algo, void *output, const void *pdata )
 {
  algo_gate_t gate;   
@@ -313,39 +415,37 @@ void exec_hash_function( int algo, void *output, const void *pdata )
 const char* const algo_alias_map[][2] =
 {
 //   alias                proper
-  { "argon2d-crds",      "argon2d250"   },
-  { "argon2d-dyn",       "argon2d500"   },
-  { "argon2d-uis",       "argon2d4096"  },
-  { "bcd",               "x13bcd"       },
-  { "bitcore",           "timetravel10" },
-  { "bitzeny",           "yescryptr8"   },
-  { "blake256r8",        "blakecoin"    },
-  { "blake256r8vnl",     "vanilla"      },
-  { "blake256r14",       "blake"        },
-  { "blake256r14dcr",    "decred"       },
-  { "cryptonote",        "cryptonight"  },
-  { "cryptonight-light", "cryptolight"  },
-  { "diamond",           "dmd-gr"       },
-  { "droplp",            "drop"         },
-  { "espers",            "hmq1725"      },
-  { "flax",              "c11"          },
-  { "hsr",               "x13sm3"       },
-  { "jackpot",           "jha"          },
-  { "jane",              "scryptjane"   }, 
-  { "lyra2",             "lyra2re"      },
-  { "lyra2v2",           "lyra2rev2"    },
-  { "lyra2v3",           "lyra2rev3"    },
-  { "myrgr",             "myr-gr"       },
-  { "myriad",            "myr-gr"       },
-  { "neo",               "neoscrypt"    },
-  { "phi",               "phi1612"      },
-  { "sib",               "x11gost"      },
-  { "timetravel8",       "timetravel"   },
-  { "veil",              "x16rt-veil"   },
-  { "x16r-hex",          "hex"          },
-  { "yenten",            "yescryptr16"  },
-  { "ziftr",             "zr5"          },
-  { NULL,                NULL           }   
+  { "argon2d-crds",      "argon2d250"     },
+  { "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"            },
+  { "jane",              "scryptjane"     }, 
+  { "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
@@ -358,7 +458,7 @@ void get_algo_alias( char** algo_or_alias )
    if ( !strcasecmp( *algo_or_alias, algo_alias_map[i][ ALIAS ] ) )
    {
      // found valid alias, return proper name
-      *algo_or_alias = (char* const)( algo_alias_map[i][ PROPER ] );
+      *algo_or_alias = (char*)( algo_alias_map[i][ PROPER ] );
      return;
    }
 }
--- a/algo-gate-api.h
+++ b/algo-gate-api.h
@@ -75,7 +75,7 @@

 // my hack at creating a set data type using bit masks. Set inclusion,
 // exclusion union and intersection operations are provided for convenience. In // some cases it may be desireable to use boolean algebra directly on the
-// data to perfomr set operations. Sets can be represented as single
+// data to perform set operations. Sets can be represented as single
 // elements, a bitwise OR of multiple elements, a bitwise OR of multiple
 // set variables or constants, or combinations of the above.
 // Examples:
@@ -110,65 +110,62 @@ inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }

 typedef struct
 {
-// mandatory functions, must be overwritten
+// Mandatory functions, one of these is mandatory. If a generic scanhash
+// is used a custom hash function must be registered, with a custom scanhash
+// the custom hash function can be called directly and doesn't need to be
+// registered in 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

 // Allocate thread local buffers and other initialization specific to miner
 // threads.
-bool ( *miner_thread_init )      ( int );
-
-// Generate global blockheader from stratum data.
-void ( *stratum_gen_work )       ( struct stratum_ctx*, struct work* );
+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*,
-                                   bool );
-
-// Return pointer to nonce in blockheader.
-uint32_t *( *get_nonceptr )      ( uint32_t* );
+void ( *get_new_work )          ( struct work*, struct work*, int, uint32_t* );

 // Decode getwork blockheader
-bool ( *work_decode )            ( const json_t*, struct work* );
+bool ( *work_decode )           ( struct work* );

 // Extra getwork data
-void ( *decode_extra_data )      ( struct work*, uint64_t* );
+void ( *decode_extra_data )     ( struct work*, uint64_t* );

-bool ( *submit_getwork_result )  ( CURL*, struct work* );
+bool ( *submit_getwork_result ) ( CURL*, struct work* );

-void ( *gen_merkle_root )        ( char*, struct stratum_ctx* );
+void ( *gen_merkle_root )       ( char*, struct stratum_ctx* );

 // Increment extranonce
-void ( *build_extraheader )      ( struct work*, 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,
+                                   unsigned char* );

-void ( *build_block_header )     ( struct work*, uint32_t, uint32_t*,
-	                                uint32_t*, uint32_t, uint32_t );
 // Build mining.submit message
-void ( *build_stratum_request )  ( char*, struct work*, struct stratum_ctx* );
+void ( *build_stratum_request ) ( char*, struct work*, struct stratum_ctx* );

-char* ( *malloc_txs_request )    ( struct work* );
+char* ( *malloc_txs_request )   ( struct work* );

 // Big or little
-void ( *set_work_data_endian )   ( struct work* );
+void ( *set_work_data_endian )  ( struct work* );

-double ( *calc_network_diff )    ( struct work* );
+double ( *calc_network_diff )   ( struct work* );

 // Wait for first work
-bool ( *ready_to_mine )          ( struct work*, struct stratum_ctx*, int );
+bool ( *ready_to_mine )         ( struct work*, struct stratum_ctx*, int );

 // Diverge mining threads
-bool ( *do_this_thread )         ( int );
+bool ( *do_this_thread )        ( int );

 // After do_this_thread
-void ( *resync_threads )         ( struct work* );
+void ( *resync_threads )        ( struct work* );
+
+// No longer needed
+json_t* (*longpoll_rpc_call)      ( CURL*, int*, char* );

-json_t* (*longpoll_rpc_call)     ( CURL*, int*, char* );
-bool ( *stratum_handle_response )( json_t* );
 set_t optimizations;
 int  ( *get_work_data_size )     ();
 int  ntime_index;
@@ -206,46 +203,80 @@ 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.

-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 );

-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 );

@@ -256,15 +287,12 @@ 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 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 );
@@ -277,17 +305,12 @@ int std_get_work_data_size();
 // by calling the algo's register function.
 bool register_algo_gate( int algo, algo_gate_t *gate );

-// Called by algos toverride any default gate functions that are applicable
+// 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 );

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

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

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

   do {
-      be32enc(&endiandata[19], nonce);
-      argon2d_crds_hash( hash, endiandata );
+      be32enc(&edata[19], nonce);
+      argon2d_crds_hash( hash, edata );
      if ( hash[7] <= Htarg && fulltest( hash, ptarget ) && !opt_benchmark )
      {
          pdata[19] = nonce;
@@ -103,31 +104,32 @@ void argon2d_dyn_hash( void *output, const void *input )
 int scanhash_argon2d_dyn( struct work *work, uint32_t max_nonce,
                      uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t _ALIGN(64) endiandata[20];
+   uint32_t _ALIGN(64) edata[20];
   uint32_t _ALIGN(64) hash[8];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
-   const uint32_t first_nonce = pdata[19];
-   const uint32_t Htarg = ptarget[7];
+   const int thr_id = mythr->id; 
+   const uint32_t first_nonce = (const uint32_t)pdata[19];
+   const uint32_t last_nonce = (const uint32_t)max_nonce;
   uint32_t nonce = first_nonce;
+   const bool bench = opt_benchmark;

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

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

@@ -146,36 +148,34 @@ int scanhash_argon2d4096( struct work *work, uint32_t max_nonce,
                           uint64_t *hashes_done, struct thr_info *mythr )
 {
   uint32_t _ALIGN(64) vhash[8];
-   uint32_t _ALIGN(64) endiandata[20];
+   uint32_t _ALIGN(64) edata[20];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = (const uint32_t)max_nonce;
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;  // thr_id arg is deprecated
   uint32_t t_cost = 1; // 1 iteration
   uint32_t m_cost = 4096; // use 4MB
   uint32_t parallelism = 1; // 1 thread, 2 lanes
+   const bool bench = opt_benchmark;

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

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

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

--- a/algo/blake/blake-4way.c
+++ b/algo/blake/blake-4way.c
@@ -48,7 +48,7 @@ int scanhash_blake_4way( struct work *work, uint32_t max_nonce,
      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;

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

--- a/algo/blake/blake2b-4way.c
+++ b/algo/blake/blake2b-4way.c
@@ -45,7 +45,7 @@ int scanhash_blake2b_8way( struct work *work, uint32_t max_nonce,
          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 8;
@@ -100,7 +100,7 @@ int scanhash_blake2b_4way( struct work *work, uint32_t max_nonce,
          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 4;
--- a/algo/blake/blake2b-hash-4way.c
+++ b/algo/blake/blake2b-hash-4way.c
@@ -33,6 +33,8 @@

 #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 },
@@ -203,9 +205,9 @@ void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out )
   casti_m512i( out, 3 ) = ctx->h[3];
 }

-#endif
+#endif   // AVX512

-#if defined(__AVX2__)
+// AVX2

 // G Mixing function.

@@ -369,4 +371,4 @@ void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out )
   casti_m256i( out, 3 ) = ctx->h[3];
 }

-#endif
+#endif  // AVX2
--- a/algo/blake/blake2b.c
+++ b/algo/blake/blake2b.c
@@ -4,6 +4,9 @@
 */

 #include "blake2b-gate.h"
+
+#if !defined(BLAKE2B_8WAY) && !defined(BLAKE2B_4WAY)
+
 #include <string.h>
 #include <stdint.h>
 #include "algo/blake/sph_blake2b.h"
@@ -58,3 +61,4 @@ int scanhash_blake2b( struct work *work, uint32_t max_nonce,
 	return 0;
 }

+#endif
--- a/algo/blake/blake2s-4way.c
+++ b/algo/blake/blake2s-4way.c
@@ -49,7 +49,7 @@ int scanhash_blake2s_16way( struct work *work, uint32_t max_nonce,
         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
         {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
         }
      }
      n += 16;
@@ -104,7 +104,7 @@ int scanhash_blake2s_8way( struct work *work, uint32_t max_nonce,
         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
         {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
         }
      }
      n += 8;
@@ -157,7 +157,7 @@ int scanhash_blake2s_4way( struct work *work, uint32_t max_nonce,
         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
         {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
              }
      }
      n += 4;
--- a/algo/blake/blake2s.c
+++ b/algo/blake/blake2s.c
@@ -1,5 +1,7 @@
 #include "blake2s-gate.h"

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

@@ -70,3 +72,4 @@ int scanhash_blake2s( struct work *work,

 	return 0;
 }
+#endif
--- a/algo/blake/blakecoin-4way.c
+++ b/algo/blake/blakecoin-4way.c
@@ -49,7 +49,7 @@ int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,
           && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;

@@ -108,7 +108,7 @@ int scanhash_blakecoin_8way( struct work *work, uint32_t max_nonce,
          && !opt_benchmark )
      {
          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 8;
   } while ( (n < max_nonce) && !work_restart[thr_id].restart );
--- a/algo/blake/blakecoin.c
+++ b/algo/blake/blakecoin.c
@@ -1,4 +1,7 @@
 #include "blakecoin-gate.h"
+
+#if !defined(BLAKECOIN_8WAY) && !defined(BLAKECOIN_4WAY)
+
 #define BLAKE32_ROUNDS 8
 #include "sph_blake.h"

@@ -93,3 +96,4 @@ int scanhash_blakecoin( struct work *work, uint32_t max_nonce,
 	return 0;
 }

+#endif
--- a/algo/blake/decred-4way.c
+++ b/algo/blake/decred-4way.c
@@ -62,7 +62,7 @@ int scanhash_decred_4way( struct work *work, uint32_t max_nonce,
      if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
      {
          pdata[DECRED_NONCE_INDEX] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;
  } while ( (n < max_nonce) && !work_restart[thr_id].restart );
--- a/algo/blake/decred-gate.c
+++ b/algo/blake/decred-gate.c
@@ -78,7 +78,6 @@ void decred_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
   uint32_t extraheader[32] = { 0 };
   int headersize = 0;
   uint32_t* extradata = (uint32_t*) sctx->xnonce1;
-   size_t t;
   int i;

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

-   // Increment extranonce2 
-   for ( t = 0; t < sctx->xnonce2_size && !( ++sctx->job.xnonce2[t] ); t++ );
-
   // Assemble block header 
   memset( g_work->data, 0, sizeof(g_work->data) );
   g_work->data[0] = le32dec( sctx->job.version );
@@ -153,7 +149,7 @@ bool register_decred_algo( algo_gate_t* gate )
  gate->hash      = (void*)&decred_hash;
 #endif
  gate->optimizations = AVX2_OPT;
-  gate->get_nonceptr          = (void*)&decred_get_nonceptr;
+//  gate->get_nonceptr          = (void*)&decred_get_nonceptr;
  gate->decode_extra_data     = (void*)&decred_decode_extradata;
  gate->build_stratum_request = (void*)&decred_be_build_stratum_request;
  gate->work_decode           = (void*)&std_be_work_decode;
--- a/algo/blake/decred.c
+++ b/algo/blake/decred.c
@@ -1,4 +1,7 @@
 #include "decred-gate.h"
+
+#if !defined(DECRED_8WAY) && !defined(DECRED_4WAY)
+
 #include "sph_blake.h"

 #include <string.h>
@@ -275,3 +278,5 @@ bool register_decred_algo( algo_gate_t* gate )
  return true;
 }
 */
+
+#endif
--- a/algo/blake/pentablake-4way.c
+++ b/algo/blake/pentablake-4way.c
@@ -105,7 +105,7 @@ int scanhash_pentablake_4way( struct work *work,
                  && fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
              {
                 pdata[19] = n + i;
-                 submit_lane_solution( work, hash+(i<<3), mythr, i );
+                 submit_solution( work, hash+(i<<3), mythr );
              }
              n += 4;

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

+#endif
--- a/algo/bmw/bmw-hash-4way.h
+++ b/algo/bmw/bmw-hash-4way.h
@@ -138,7 +138,7 @@ void bmw512_2way_close( bmw512_2way_context *ctx, void *dst );

 #if defined(__AVX2__)

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

 typedef struct {
   __m256i buf[16];
@@ -149,7 +149,6 @@ typedef struct {

 typedef bmw_4way_big_context bmw512_4way_context;

-
 void bmw512_4way_init(void *cc);

 void bmw512_4way_update(void *cc, const void *data, size_t len);
@@ -164,6 +163,7 @@ void bmw512_4way_addbits_and_close(

 #if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)

+// BMW-512 64 bit 8 way
 typedef struct {
   __m512i buf[16];
   __m512i H[16];
@@ -171,6 +171,8 @@ typedef struct {
   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 );
--- a/algo/bmw/bmw512-4way.c
+++ b/algo/bmw/bmw512-4way.c
@@ -46,7 +46,7 @@ int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
          if ( fulltest( lane_hash, ptarget ) )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 8;
@@ -99,7 +99,7 @@ int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
          if ( fulltest( lane_hash, ptarget ) )
          {
              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
      n += 4;
--- a/algo/bmw/bmw512-hash-4way.c
+++ b/algo/bmw/bmw512-hash-4way.c
@@ -1507,6 +1507,93 @@ void bmw512_8way_close( bmw512_8way_context *ctx, void *dst )
      casti_m512i( dst, u ) = h1[ v ];
 }

+void bmw512_8way_full( bmw512_8way_context *ctx, void *out, const void *data,
+                                size_t len )
+{
+   __m512i *vdata = (__m512i*)data;
+   __m512i *buf = ctx->buf;
+   __m512i htmp[16];
+   __m512i *H = ctx->H;
+   __m512i *h2 = htmp;
+   uint64_t bit_count = len * 8;
+   size_t ptr = 0;
+   const int buf_size = 128;  // bytes of one lane, compatible with len
+
+// Init
+
+   H[ 0] = m512_const1_64( 0x8081828384858687 );
+   H[ 1] = m512_const1_64( 0x88898A8B8C8D8E8F );
+   H[ 2] = m512_const1_64( 0x9091929394959697 );
+   H[ 3] = m512_const1_64( 0x98999A9B9C9D9E9F );
+   H[ 4] = m512_const1_64( 0xA0A1A2A3A4A5A6A7 );
+   H[ 5] = m512_const1_64( 0xA8A9AAABACADAEAF );
+   H[ 6] = m512_const1_64( 0xB0B1B2B3B4B5B6B7 );
+   H[ 7] = m512_const1_64( 0xB8B9BABBBCBDBEBF );
+   H[ 8] = m512_const1_64( 0xC0C1C2C3C4C5C6C7 );
+   H[ 9] = m512_const1_64( 0xC8C9CACBCCCDCECF );
+   H[10] = m512_const1_64( 0xD0D1D2D3D4D5D6D7 );
+   H[11] = m512_const1_64( 0xD8D9DADBDCDDDEDF );
+   H[12] = m512_const1_64( 0xE0E1E2E3E4E5E6E7 );
+   H[13] = m512_const1_64( 0xE8E9EAEBECEDEEEF );
+   H[14] = m512_const1_64( 0xF0F1F2F3F4F5F6F7 );
+   H[15] = m512_const1_64( 0xF8F9FAFBFCFDFEFF );
+
+// Update
+
+   while ( len > 0 )
+   {
+      size_t clen;
+      clen = buf_size - ptr;
+      if ( clen > len )
+         clen = len;
+      memcpy_512( buf + (ptr>>3), vdata, clen >> 3 );
+      vdata = vdata + (clen>>3);
+      len -= clen;
+      ptr += clen;
+      if ( ptr == buf_size )
+      {
+         __m512i *ht;
+         compress_big_8way( buf, H, h2 );
+         ht = H;
+         H = h2;
+         h2 = ht;
+         ptr = 0;
+      }
+   }
+   if ( H != ctx->H )
+      memcpy_512( ctx->H, H, 16 );
+
+// Close   
+{
+   __m512i h1[16], h2[16];
+   size_t u, v;
+
+   buf[ ptr>>3 ] = m512_const1_64( 0x80 );
+   ptr += 8;
+
+   if (  ptr > (buf_size - 8) )
+   {
+      memset_zero_512( buf + (ptr>>3), (buf_size - ptr) >> 3 );
+      compress_big_8way( buf, H, h1 );
+      ptr = 0;
+      H = h1;
+   }
+   memset_zero_512( buf + (ptr>>3), (buf_size - 8 - ptr) >> 3 );
+   buf[ (buf_size - 8) >> 3 ] = _mm512_set1_epi64( bit_count );
+   compress_big_8way( buf, H, h2 );
+   for ( u = 0; u < 16; u ++ )
+      buf[ u ] = h2[ u ];
+   compress_big_8way( buf, final_b8, h1 );
+   for (u = 0, v = 8; u < 8; u ++, v ++)
+      casti_m512i( out, u ) = h1[ v ];
+}
+
+
+
+}   
+
+
+
 #endif // AVX512

 #ifdef __cplusplus
--- a/algo/bmw/bmw512.c
+++ b/algo/bmw/bmw512.c
@@ -1,5 +1,7 @@
 #include "algo-gate-api.h"

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

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

+#endif // !AVX2
+
 #if SPH_64

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

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

+#endif // !AVX2
+
 #if SPH_64

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

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

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

+#endif  // !AVX2
+
 #if SPH_64

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

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

 #if SPH_64

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

 #endif

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

+#endif // !AVX2
+
 #if SPH_64

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

 #endif

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

+#endif // !AVX2
+
 #if SPH_64

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

+#endif // !AVX2
+
 #if SPH_64

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

 #endif

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

+#endif // !AVX2
+
 #if SPH_64

 /**
--- a/algo/cryptonight/cryptolight.c
+++ b/algo/cryptonight/cryptolight.c
@@ -1,368 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-#include "algo-gate-api.h"
-
-#if defined(__arm__) || defined(_MSC_VER)
-#ifndef NOASM
-#define NOASM
-#endif
-#endif
-
-#include "crypto/oaes_lib.h"
-#include "crypto/c_keccak.h"
-#include "crypto/c_groestl.h"
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "crypto/int-util.h"
-#include "crypto/hash-ops.h"
-
-#if USE_INT128
-
-#if __GNUC__ == 4 && __GNUC_MINOR__ >= 4 && __GNUC_MINOR__ < 6
-typedef unsigned int uint128_t __attribute__ ((__mode__ (TI)));
-#elif defined (_MSC_VER)
-/* only for mingw64 on windows */
-#undef  USE_INT128
-#define USE_INT128 (0)
-#else
-typedef __uint128_t uint128_t;
-#endif
-
-#endif
-
-#define LITE 1
-#if LITE /* cryptonight-light */
-#define MEMORY (1 << 20)
-#define ITER   (1 << 19)
-#else
-#define MEMORY (1 << 21) /* 2 MiB */
-#define ITER   (1 << 20)
-#endif
-
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)
-
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-	union hash_state hs;
-	struct {
-		uint8_t k[64];
-		uint8_t init[INIT_SIZE_BYTE];
-	};
-};
-#pragma pack(pop)
-
-static void do_blake_hash(const void* input, size_t len, char* output) {
-	blake256_hash((uint8_t*)output, input, len);
-}
-
-static void do_groestl_hash(const void* input, size_t len, char* output) {
-	groestl(input, len * 8, (uint8_t*)output);
-}
-
-static void do_jh_hash(const void* input, size_t len, char* output) {
-	int r = jh_hash(HASH_SIZE * 8, input, 8 * len, (uint8_t*)output);
-	assert(likely(SUCCESS == r));
-}
-
-static void do_skein_hash(const void* input, size_t len, char* output) {
-	int r = skein_hash(8 * HASH_SIZE, input, 8 * len, (uint8_t*)output);
-	assert(likely(SKEIN_SUCCESS == r));
-}
-
-extern int aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#if !defined(_MSC_VER) && !defined(NOASM)
-extern int fast_aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int fast_aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#else
-#define fast_aesb_single_round     aesb_single_round
-#define fast_aesb_pseudo_round_mut aesb_pseudo_round_mut
-#endif
-
-#if defined(NOASM) || !defined(__x86_64__)
-static uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi) {
-	// multiplier   = ab = a * 2^32 + b
-	// multiplicand = cd = c * 2^32 + d
-	// ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
-	uint64_t a = hi_dword(multiplier);
-	uint64_t b = lo_dword(multiplier);
-	uint64_t c = hi_dword(multiplicand);
-	uint64_t d = lo_dword(multiplicand);
-
-	uint64_t ac = a * c;
-	uint64_t ad = a * d;
-	uint64_t bc = b * c;
-	uint64_t bd = b * d;
-
-	uint64_t adbc = ad + bc;
-	uint64_t adbc_carry = adbc < ad ? 1 : 0;
-
-	// multiplier * multiplicand = product_hi * 2^64 + product_lo
-	uint64_t product_lo = bd + (adbc << 32);
-	uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
-	*product_hi = ac + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
-	assert(ac <= *product_hi);
-
-	return product_lo;
-}
-#else
-extern uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi);
-#endif
-
-static void (* const extra_hashes[4])(const void *, size_t, char *) = {
-		do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash
-};
-
-
-static inline size_t e2i(const uint8_t* a) {
-#if !LITE
-	return ((uint32_t *)a)[0] & 0x1FFFF0;
-#else
-	return ((uint32_t *)a)[0] & 0xFFFF0;
-#endif
-}
-
-static inline void mul_sum_xor_dst(const uint8_t* a, uint8_t* c, uint8_t* dst) {
-	uint64_t hi, lo = mul128(((uint64_t*) a)[0], ((uint64_t*) dst)[0], &hi) + ((uint64_t*) c)[1];
-	hi += ((uint64_t*) c)[0];
-
-	((uint64_t*) c)[0] = ((uint64_t*) dst)[0] ^ hi;
-	((uint64_t*) c)[1] = ((uint64_t*) dst)[1] ^ lo;
-	((uint64_t*) dst)[0] = hi;
-	((uint64_t*) dst)[1] = lo;
-}
-
-static inline void xor_blocks(uint8_t* a, const uint8_t* b) {
-#if USE_INT128
-	*((uint128_t*) a) ^= *((uint128_t*) b);
-#else
-	((uint64_t*) a)[0] ^= ((uint64_t*) b)[0];
-	((uint64_t*) a)[1] ^= ((uint64_t*) b)[1];
-#endif
-}
-
-static inline void xor_blocks_dst(const uint8_t* a, const uint8_t* b, uint8_t* dst) {
-#if USE_INT128
-	*((uint128_t*) dst) = *((uint128_t*) a) ^ *((uint128_t*) b);
-#else
-	((uint64_t*) dst)[0] = ((uint64_t*) a)[0] ^ ((uint64_t*) b)[0];
-	((uint64_t*) dst)[1] = ((uint64_t*) a)[1] ^ ((uint64_t*) b)[1];
-#endif
-}
-
-struct cryptonight_ctx {
-	uint8_t _ALIGN(16) long_state[MEMORY];
-	union cn_slow_hash_state state;
-	uint8_t _ALIGN(16) text[INIT_SIZE_BYTE];
-	uint8_t _ALIGN(16) a[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) b[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) c[AES_BLOCK_SIZE];
-	oaes_ctx* aes_ctx;
-};
-
-static void cryptolight_hash_ctx(void* output, const void* input, int len, struct cryptonight_ctx* ctx)
-{
-        len = 76;
-	hash_process(&ctx->state.hs, (const uint8_t*) input, len);
-	ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
-	size_t i, j;
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx->aes_ctx, ctx->state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 0], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 1], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 2], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 3], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 4], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 5], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 6], ctx->aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 7], ctx->aes_ctx->key->exp_data);
-		memcpy(&ctx->long_state[i], ctx->text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx->state.k[0], &ctx->state.k[32], ctx->a);
-	xor_blocks_dst(&ctx->state.k[16], &ctx->state.k[48], ctx->b);
-
-	for (i = 0; likely(i < ITER / 4); ++i) {
-		/* Dependency chain: address -> read value ------+
-		 * written value <-+ hard function (AES or MUL) <+
-		 * next address  <-+
-		 */
-		/* Iteration 1 */
-		j = e2i(ctx->a);
-		aesb_single_round(&ctx->long_state[j], ctx->c, ctx->a);
-		xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j]);
-		/* Iteration 2 */
-		mul_sum_xor_dst(ctx->c, ctx->a, &ctx->long_state[e2i(ctx->c)]);
-		/* Iteration 3 */
-		j = e2i(ctx->a);
-		aesb_single_round(&ctx->long_state[j], ctx->b, ctx->a);
-		xor_blocks_dst(ctx->b, ctx->c, &ctx->long_state[j]);
-		/* Iteration 4 */
-		mul_sum_xor_dst(ctx->b, ctx->a, &ctx->long_state[e2i(ctx->b)]);
-	}
-
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		xor_blocks(&ctx->text[0 * AES_BLOCK_SIZE], &ctx->long_state[i + 0 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[0 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[1 * AES_BLOCK_SIZE], &ctx->long_state[i + 1 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[1 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[2 * AES_BLOCK_SIZE], &ctx->long_state[i + 2 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[2 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[3 * AES_BLOCK_SIZE], &ctx->long_state[i + 3 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[3 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[4 * AES_BLOCK_SIZE], &ctx->long_state[i + 4 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[4 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[5 * AES_BLOCK_SIZE], &ctx->long_state[i + 5 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[5 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[6 * AES_BLOCK_SIZE], &ctx->long_state[i + 6 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[6 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[7 * AES_BLOCK_SIZE], &ctx->long_state[i + 7 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx->text[7 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-	}
-	memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
-	hash_permutation(&ctx->state.hs);
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
-	oaes_free((OAES_CTX **) &ctx->aes_ctx);
-}
-
-void cryptolight_hash(void* output, const void* input, int len) {
-	struct cryptonight_ctx *ctx = (struct cryptonight_ctx*)malloc(sizeof(struct cryptonight_ctx));
-	cryptolight_hash_ctx(output, input, len, ctx);
-	free(ctx);
-}
-
-#if defined(__AES__)
-
-static void cryptolight_hash_ctx_aes_ni(void* output, const void* input,
-                       int len, struct cryptonight_ctx* ctx)
-{
-	hash_process(&ctx->state.hs, (const uint8_t*)input, len);
-	ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
-	size_t i, j;
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx->aes_ctx, ctx->state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 0], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 1], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 2], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 3], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 4], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 5], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 6], ctx->aes_ctx->key->exp_data);
-		fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 7], ctx->aes_ctx->key->exp_data);
-		memcpy(&ctx->long_state[i], ctx->text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx->state.k[0], &ctx->state.k[32], ctx->a);
-	xor_blocks_dst(&ctx->state.k[16], &ctx->state.k[48], ctx->b);
-
-	for (i = 0; likely(i < ITER / 4); ++i) {
-		/* Dependency chain: address -> read value ------+
-		 * written value <-+ hard function (AES or MUL) <+
-		 * next address  <-+
-		 */
-		/* Iteration 1 */
-		j = e2i(ctx->a);
-		fast_aesb_single_round(&ctx->long_state[j], ctx->c, ctx->a);
-		xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j]);
-		/* Iteration 2 */
-		mul_sum_xor_dst(ctx->c, ctx->a, &ctx->long_state[e2i(ctx->c)]);
-		/* Iteration 3 */
-		j = e2i(ctx->a);
-		fast_aesb_single_round(&ctx->long_state[j], ctx->b, ctx->a);
-		xor_blocks_dst(ctx->b, ctx->c, &ctx->long_state[j]);
-		/* Iteration 4 */
-		mul_sum_xor_dst(ctx->b, ctx->a, &ctx->long_state[e2i(ctx->b)]);
-	}
-
-	memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-		xor_blocks(&ctx->text[0 * AES_BLOCK_SIZE], &ctx->long_state[i + 0 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[0 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[1 * AES_BLOCK_SIZE], &ctx->long_state[i + 1 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[1 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[2 * AES_BLOCK_SIZE], &ctx->long_state[i + 2 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[2 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[3 * AES_BLOCK_SIZE], &ctx->long_state[i + 3 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[3 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[4 * AES_BLOCK_SIZE], &ctx->long_state[i + 4 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[4 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[5 * AES_BLOCK_SIZE], &ctx->long_state[i + 5 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[5 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[6 * AES_BLOCK_SIZE], &ctx->long_state[i + 6 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[6 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-		xor_blocks(&ctx->text[7 * AES_BLOCK_SIZE], &ctx->long_state[i + 7 * AES_BLOCK_SIZE]);
-		fast_aesb_pseudo_round_mut(&ctx->text[7 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
-	}
-	memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
-	hash_permutation(&ctx->state.hs);
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
-	oaes_free((OAES_CTX **) &ctx->aes_ctx);
-}
-
-#endif
-
-int scanhash_cryptolight( struct work *work,
-		uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr)
-{
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
-	uint32_t *nonceptr = (uint32_t*) (((char*)pdata) + 39);
-	uint32_t n = *nonceptr - 1;
-	const uint32_t first_nonce = n + 1;
-	//const uint32_t Htarg = ptarget[7];
-	uint32_t _ALIGN(32) hash[HASH_SIZE / 4];
-   int thr_id = mythr->id;
-
-	struct cryptonight_ctx *ctx = (struct cryptonight_ctx*)malloc(sizeof(struct cryptonight_ctx));
-
-#if defined(__AES__)
-		do {
-			*nonceptr = ++n;
-			cryptolight_hash_ctx_aes_ni(hash, pdata, 76, ctx);
-			if (unlikely(hash[7] < ptarget[7])) {
-				*hashes_done = n - first_nonce + 1;
-				free(ctx);
-				return true;
-			}
-		} while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-#else
-		do {
-			*nonceptr = ++n;
-			cryptolight_hash_ctx(hash, pdata, 76, ctx);
-			if (unlikely(hash[7] < ptarget[7])) {
-				*hashes_done = n - first_nonce + 1;
-				free(ctx);
-				return true;
-			}
-		} while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-#endif
-	free(ctx);
-	*hashes_done = n - first_nonce + 1;
-	return 0;
-}
-
-bool register_cryptolight_algo( algo_gate_t* gate )
-{
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash  = (void*)&scanhash_cryptolight;
-  gate->hash      = (void*)&cryptolight_hash;
-  gate->hash_suw  = (void*)&cryptolight_hash; 
-  return true;
-};
-
--- a/algo/cryptonight/cryptonight-aesni.c
+++ b/algo/cryptonight/cryptonight-aesni.c
@@ -1,357 +0,0 @@
-#if defined(__AES__)
-
-#include <x86intrin.h>
-#include <memory.h>
-#include "cryptonight.h"
-#include "miner.h"
-#include "crypto/c_keccak.h"
-#include <immintrin.h>
-
-static inline void ExpandAESKey256_sub1(__m128i *tmp1, __m128i *tmp2)
-{
-	__m128i tmp4;
-	*tmp2 = _mm_shuffle_epi32(*tmp2, 0xFF);
-	tmp4 = _mm_slli_si128(*tmp1, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp1 = _mm_xor_si128(*tmp1, tmp4);
-	*tmp1 = _mm_xor_si128(*tmp1, *tmp2);
-}
-
-static inline void ExpandAESKey256_sub2(__m128i *tmp1, __m128i *tmp3)
-{
-	__m128i tmp2, tmp4;
-	
-	tmp4 = _mm_aeskeygenassist_si128(*tmp1, 0x00);
-	tmp2 = _mm_shuffle_epi32(tmp4, 0xAA);
-	tmp4 = _mm_slli_si128(*tmp3, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	tmp4 = _mm_slli_si128(tmp4, 0x04);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp4);
-	*tmp3 = _mm_xor_si128(*tmp3, tmp2);
-}
-
-// Special thanks to Intel for helping me
-// with ExpandAESKey256() and its subroutines
-static inline void ExpandAESKey256(char *keybuf)
-{
-	__m128i tmp1, tmp2, tmp3, *keys;
-	
-	keys = (__m128i *)keybuf;
-	
-	tmp1 = _mm_load_si128((__m128i *)keybuf);
-	tmp3 = _mm_load_si128((__m128i *)(keybuf+0x10));
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x01);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[2] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[3] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x02);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[4] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[5] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x04);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[6] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[7] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x08);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[8] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[9] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x10);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[10] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[11] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x20);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[12] = tmp1;
-	ExpandAESKey256_sub2(&tmp1, &tmp3);
-	keys[13] = tmp3;
-	
-	tmp2 = _mm_aeskeygenassist_si128(tmp3, 0x40);
-	ExpandAESKey256_sub1(&tmp1, &tmp2);
-	keys[14] = tmp1;
-}
-
-// align to 64 byte cache line
-typedef struct 
-{
-    uint8_t long_state[MEMORY] __attribute((aligned(64)));
-    union cn_slow_hash_state state;
-    uint8_t text[INIT_SIZE_BYTE] __attribute((aligned(64)));
-    uint64_t a[AES_BLOCK_SIZE >> 3] __attribute__((aligned(64)));
-    uint64_t b[AES_BLOCK_SIZE >> 3] __attribute__((aligned(64)));
-    uint8_t c[AES_BLOCK_SIZE] __attribute__((aligned(64)));
-} cryptonight_ctx;
-
-static __thread cryptonight_ctx ctx;
-
-void cryptonight_hash_aes( void *restrict output, const void *input, int len )
-{
-    uint8_t ExpandedKey[256] __attribute__((aligned(64)));
-    __m128i *longoutput, *expkey, *xmminput;
-    size_t i, j;
-    
-    keccak( (const uint8_t*)input, 76, (char*)&ctx.state.hs.b, 200 );
-
-    if ( cryptonightV7 && len < 43 )
-      return;
-
-    const uint64_t tweak = cryptonightV7 
-                         ? *((const uint64_t*) (((const uint8_t*)input) + 35))
-                           ^ ctx.state.hs.w[24] : 0; 
-
-    memcpy( ExpandedKey, ctx.state.hs.b, AES_KEY_SIZE );
-    ExpandAESKey256( ExpandedKey );
-    memcpy( ctx.text, ctx.state.init, INIT_SIZE_BYTE );
-    
-    longoutput = (__m128i*)ctx.long_state;
-    xmminput   = (__m128i*)ctx.text;
-    expkey     = (__m128i*)ExpandedKey;
-    
-    // prefetch expkey, xmminput and enough longoutput for 4 iterations
-    _mm_prefetch( xmminput,     _MM_HINT_T0 );
-    _mm_prefetch( xmminput + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey,     _MM_HINT_T0 );
-    _mm_prefetch( expkey + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey + 8, _MM_HINT_T0 );
-    for ( i = 0; i < 64; i += 16 )
-    {
-        __builtin_prefetch( longoutput + i,      1, 0 );
-        __builtin_prefetch( longoutput + i +  4, 1, 0 );
-        __builtin_prefetch( longoutput + i +  8, 1, 0 );
-        __builtin_prefetch( longoutput + i + 12, 1, 0 );
-    }
-
-    // n-4 iterations
-    for ( i = 0; likely( i < MEMORY_M128I - 4*INIT_SIZE_M128I );
-                         i += INIT_SIZE_M128I )
-    {
-        // prefetch 4 iterations ahead.
-        __builtin_prefetch( longoutput + i + 64, 1, 0 );
-        __builtin_prefetch( longoutput + i + 68, 1, 0 );
-
-	for ( j = 0; j < 10; j++ )
-	{
-		xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-		xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-		xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-		xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-		xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-		xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-		xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-		xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-	}
-	_mm_store_si128( &( longoutput[i  ] ), xmminput[0] );
-	_mm_store_si128( &( longoutput[i+1] ), xmminput[1] );
-	_mm_store_si128( &( longoutput[i+2] ), xmminput[2] );
-	_mm_store_si128( &( longoutput[i+3] ), xmminput[3] );
-	_mm_store_si128( &( longoutput[i+4] ), xmminput[4] );
-	_mm_store_si128( &( longoutput[i+5] ), xmminput[5] );
-	_mm_store_si128( &( longoutput[i+6] ), xmminput[6] );
-	_mm_store_si128( &( longoutput[i+7] ), xmminput[7] );
-    }
-    // last 4 iterations
-    for ( ; likely( i < MEMORY_M128I ); i += INIT_SIZE_M128I )
-    {
-        for ( j = 0; j < 10; j++ )
-        {
-                xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-                xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-                xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-                xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-                xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-                xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-                xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-                xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-        _mm_store_si128( &( longoutput[i  ] ), xmminput[0] );
-        _mm_store_si128( &( longoutput[i+1] ), xmminput[1] );
-        _mm_store_si128( &( longoutput[i+2] ), xmminput[2] );
-        _mm_store_si128( &( longoutput[i+3] ), xmminput[3] );
-        _mm_store_si128( &( longoutput[i+4] ), xmminput[4] );
-        _mm_store_si128( &( longoutput[i+5] ), xmminput[5] );
-        _mm_store_si128( &( longoutput[i+6] ), xmminput[6] );
-        _mm_store_si128( &( longoutput[i+7] ), xmminput[7] );
-    }
-
-    ctx.a[0] = ((uint64_t *)ctx.state.k)[0] ^ ((uint64_t *)ctx.state.k)[4];
-    ctx.b[0] = ((uint64_t *)ctx.state.k)[2] ^ ((uint64_t *)ctx.state.k)[6];
-    ctx.a[1] = ((uint64_t *)ctx.state.k)[1] ^ ((uint64_t *)ctx.state.k)[5];
-    ctx.b[1] = ((uint64_t *)ctx.state.k)[3] ^ ((uint64_t *)ctx.state.k)[7];
-
-    uint64_t a[2] __attribute((aligned(16))),
-             b[2] __attribute((aligned(16))),
-             c[2] __attribute((aligned(16)));
-    a[0] = ctx.a[0];
-    a[1] = ctx.a[1];
-    __m128i b_x = _mm_load_si128( (__m128i*)ctx.b );
-    __m128i a_x = _mm_load_si128( (__m128i*)a );
-    __m128i* lsa = (__m128i*)&ctx.long_state[ a[0] & 0x1FFFF0 ];
-    __m128i c_x = _mm_load_si128( lsa );
-    uint64_t *nextblock;
-    uint64_t hi, lo;
-
-    // n-1 iterations
-    for( i = 0; __builtin_expect( i < 0x7ffff, 1 ); i++ )
-    {	  
-	c_x = _mm_aesenc_si128( c_x, a_x );
-	_mm_store_si128( (__m128i*)c, c_x );
-        b_x = _mm_xor_si128( b_x, c_x );
-        nextblock = (uint64_t *)&ctx.long_state[c[0] & 0x1FFFF0];
-        _mm_store_si128( lsa, b_x );
-
-        if ( cryptonightV7 )
-        {
-           const uint8_t tmp = ( (const uint8_t*)(lsa) )[11];
-           const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-           ((uint8_t*)(lsa))[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-        } 
-
-	b[0] = nextblock[0];
-	b[1] = nextblock[1];
-
-        // hi,lo = 64bit x 64bit multiply of c[0] and b[0]
-	__asm__( "mulq %3\n\t"
-	         : "=d" ( hi ),
-	           "=a" ( lo )
-	         : "%a" ( c[0] ),
-	           "rm" ( b[0] )
-		 : "cc" );
-
-        b_x = c_x;
-
-        a[0] += hi;
-        a[1] += lo;
-        nextblock[0] = a[0];
-        nextblock[1] = cryptonightV7 ? a[1] ^ tweak : a[1];
-        a[0] ^= b[0];
-        a[1] ^= b[1];
-
-        lsa = (__m128i*)&ctx.long_state[ a[0] & 0x1FFFF0 ];
-        a_x = _mm_load_si128( (__m128i*)a );
-        c_x = _mm_load_si128( lsa );
-    }
-    // abreviated nth iteration
-    c_x = _mm_aesenc_si128( c_x, a_x );
-    _mm_store_si128( (__m128i*)c, c_x );
-    b_x = _mm_xor_si128( b_x, c_x );
-    nextblock = (uint64_t *)&ctx.long_state[c[0] & 0x1FFFF0];
-    _mm_store_si128( lsa, b_x );
-
-    if ( cryptonightV7 )
-    {
-       const uint8_t tmp = ( (const uint8_t*)(lsa) )[11];
-       const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-       ((uint8_t*)(lsa))[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-    }
-
-    b[0] = nextblock[0];
-    b[1] = nextblock[1];
-
-    __asm__( "mulq %3\n\t"
-             : "=d" ( hi ),
-               "=a" ( lo )
-             : "%a" ( c[0] ),
-               "rm" ( b[0] )
-             : "cc" );
-
-    a[0] += hi;
-    a[1] += lo;
-    nextblock[0] = a[0];
-    nextblock[1] = cryptonightV7 ? a[1] ^ tweak : a[1];
-    a[0] ^= b[0];
-    a[1] ^= b[1];
-
-    memcpy( ExpandedKey, &ctx.state.hs.b[32], AES_KEY_SIZE );
-    ExpandAESKey256( ExpandedKey );
-    memcpy( ctx.text, ctx.state.init, INIT_SIZE_BYTE );
-    
-    // prefetch expkey, all of xmminput and enough longoutput for 4 loops
-    _mm_prefetch( xmminput,     _MM_HINT_T0 );
-    _mm_prefetch( xmminput + 4, _MM_HINT_T0 );
-    for ( i = 0; i < 64; i += 16 )
-    {
-       _mm_prefetch( longoutput + i,      _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i +  4, _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i +  8, _MM_HINT_T0 );
-       _mm_prefetch( longoutput + i + 12, _MM_HINT_T0 );
-    }
-    _mm_prefetch( expkey,     _MM_HINT_T0 );
-    _mm_prefetch( expkey + 4, _MM_HINT_T0 );
-    _mm_prefetch( expkey + 8, _MM_HINT_T0 );
-
-    // n-4 iterations
-    for ( i = 0; likely( i < MEMORY_M128I - 4*INIT_SIZE_M128I );
-                         i += INIT_SIZE_M128I )
-    {
-        // stay 4 iterations ahead.
-        _mm_prefetch( longoutput + i + 64, _MM_HINT_T0 );
-        _mm_prefetch( longoutput + i + 68, _MM_HINT_T0 );
-
-        xmminput[0] = _mm_xor_si128( longoutput[i  ], xmminput[0] );
-        xmminput[1] = _mm_xor_si128( longoutput[i+1], xmminput[1] );
-        xmminput[2] = _mm_xor_si128( longoutput[i+2], xmminput[2] );
-        xmminput[3] = _mm_xor_si128( longoutput[i+3], xmminput[3] );
-        xmminput[4] = _mm_xor_si128( longoutput[i+4], xmminput[4] );
-        xmminput[5] = _mm_xor_si128( longoutput[i+5], xmminput[5] );
-        xmminput[6] = _mm_xor_si128( longoutput[i+6], xmminput[6] );
-        xmminput[7] = _mm_xor_si128( longoutput[i+7], xmminput[7] );
-		
-        for( j = 0; j < 10; j++ )
-        {
-            xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-	    xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-	    xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-	    xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-	    xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-	    xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-	    xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-	    xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-    }
-    // last 4 iterations 
-    for ( ; likely( i < MEMORY_M128I ); i += INIT_SIZE_M128I )
-    {
-        xmminput[0] = _mm_xor_si128( longoutput[i  ], xmminput[0] );
-        xmminput[1] = _mm_xor_si128( longoutput[i+1], xmminput[1] );
-        xmminput[2] = _mm_xor_si128( longoutput[i+2], xmminput[2] );
-        xmminput[3] = _mm_xor_si128( longoutput[i+3], xmminput[3] );
-        xmminput[4] = _mm_xor_si128( longoutput[i+4], xmminput[4] );
-        xmminput[5] = _mm_xor_si128( longoutput[i+5], xmminput[5] );
-        xmminput[6] = _mm_xor_si128( longoutput[i+6], xmminput[6] );
-        xmminput[7] = _mm_xor_si128( longoutput[i+7], xmminput[7] );
-
-        for( j = 0; j < 10; j++ )
-        {
-            xmminput[0] = _mm_aesenc_si128( xmminput[0], expkey[j] );
-            xmminput[1] = _mm_aesenc_si128( xmminput[1], expkey[j] );
-            xmminput[2] = _mm_aesenc_si128( xmminput[2], expkey[j] );
-            xmminput[3] = _mm_aesenc_si128( xmminput[3], expkey[j] );
-            xmminput[4] = _mm_aesenc_si128( xmminput[4], expkey[j] );
-            xmminput[5] = _mm_aesenc_si128( xmminput[5], expkey[j] );
-            xmminput[6] = _mm_aesenc_si128( xmminput[6], expkey[j] );
-            xmminput[7] = _mm_aesenc_si128( xmminput[7], expkey[j] );
-        }
-    }
-
-    memcpy( ctx.state.init, ctx.text, INIT_SIZE_BYTE);
-    keccakf( (uint64_t*)&ctx.state.hs.w, 24 );
-    extra_hashes[ctx.state.hs.b[0] & 3](&ctx.state, 200, output);
-
-}
-#endif
--- a/algo/cryptonight/cryptonight-common.c
+++ b/algo/cryptonight/cryptonight-common.c
@@ -1,127 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-// Modified for CPUminer by Lucas Jones
-
-#include "cpuminer-config.h"
-#include "algo-gate-api.h"
-
-#if defined(__AES__)
-  #include "algo/groestl/aes_ni/hash-groestl256.h"
-#else
-#include "crypto/c_groestl.h"
-#endif
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "cryptonight.h"
-
-/*
-#if defined __unix__ && (!defined __APPLE__)
-#include <sys/mman.h>
-#elif defined _WIN32
-#include <windows.h>
-#endif
-*/
-
-void do_blake_hash(const void* input, size_t len, char* output) {
-    blake256_hash((uint8_t*)output, input, len);
-}
-
-void do_groestl_hash(const void* input, size_t len, char* output) {
-#if defined(__AES__)
-    hashState_groestl256 ctx;
-    init_groestl256( &ctx, 32 );
-    update_and_final_groestl256( &ctx, output, input, len * 8 );
-#else
-    groestl(input, len * 8, (uint8_t*)output);
-#endif
-}
-
-void do_jh_hash(const void* input, size_t len, char* output) {
-    jh_hash(32 * 8, input, 8 * len, (uint8_t*)output);
-}
-
-void do_skein_hash(const void* input, size_t len, char* output) {
-    skein_hash(8 * 32, input, 8 * len, (uint8_t*)output);
-}
-
-void (* const extra_hashes[4])( const void *, size_t, char *) =
-    { do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash };
-
-void cryptonight_hash( void *restrict output, const void *input, int len )
-{
-#if defined(__AES__)
-  cryptonight_hash_aes( output, input, len );
-#else
-  cryptonight_hash_ctx ( output, input, len );
-#endif
-}
-
-void cryptonight_hash_suw( void *restrict output, const void *input )
-{
-#if defined(__AES__)
-  cryptonight_hash_aes( output, input, 76 );
-#else
-  cryptonight_hash_ctx ( output, input, 76 );
-#endif
-}
-
-bool cryptonightV7 = false;
-
-int scanhash_cryptonight( struct work *work, uint32_t max_nonce,
-                   uint64_t *hashes_done, struct thr_info *mythr )
- {
-    uint32_t *pdata = work->data;
-    uint32_t *ptarget = work->target;
-    int thr_id = mythr->id;
-
-    uint32_t *nonceptr = (uint32_t*) (((char*)pdata) + 39);
-    uint32_t n = *nonceptr - 1;
-    const uint32_t first_nonce = n + 1;
-    const uint32_t Htarg = ptarget[7];
-    uint32_t hash[32 / 4] __attribute__((aligned(32)));
-
-//    if (  (  cryptonightV7 && ( *(uint8_t*)pdata <  7 ) )
-//       || ( !cryptonightV7 && ( *(uint8_t*)pdata == 7 ) ) )
-//          applog(LOG_WARNING,"Cryptonight variant mismatch, shares may be rejected.");
-
-    do
-    {
-       *nonceptr = ++n;
-       cryptonight_hash( hash, pdata, 76 );
-       if (unlikely( hash[7] < Htarg ))
-       {
-           *hashes_done = n - first_nonce + 1;
-//           work_set_target_ratio( work, hash );
-	   return true;
-       }
-    } while (likely((n <= max_nonce && !work_restart[thr_id].restart)));
-    
-    *hashes_done = n - first_nonce + 1;
-    return 0;
-}
-
-bool register_cryptonight_algo( algo_gate_t* gate )
-{
-  cryptonightV7 = false;
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash         = (void*)&scanhash_cryptonight;
-  gate->hash             = (void*)&cryptonight_hash;
-  gate->hash_suw         = (void*)&cryptonight_hash_suw;  
-  return true;
-};
-
-bool register_cryptonightv7_algo( algo_gate_t* gate )
-{
-  cryptonightV7 = true;
-  register_json_rpc2( gate );
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash      = (void*)&scanhash_cryptonight;
-  gate->hash          = (void*)&cryptonight_hash;
-  gate->hash_suw      = (void*)&cryptonight_hash_suw;
-  return true;
-};
-
--- a/algo/cryptonight/cryptonight.c
+++ b/algo/cryptonight/cryptonight.c
@@ -1,310 +0,0 @@
-// Copyright (c) 2012-2013 The Cryptonote developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-// Modified for CPUminer by Lucas Jones
-
-#include "miner.h"
-#include <memory.h>
-
-#if defined(__arm__) || defined(_MSC_VER)
-#ifndef NOASM
-#define NOASM
-#endif
-#endif
-
-#include "crypto/oaes_lib.h"
-#include "crypto/c_keccak.h"
-#include "crypto/c_groestl.h"
-#include "crypto/c_blake256.h"
-#include "crypto/c_jh.h"
-#include "crypto/c_skein.h"
-#include "crypto/int-util.h"
-//#include "crypto/hash-ops.h"
-#include "cryptonight.h"
-
-#if USE_INT128
-
-#if __GNUC__ == 4 && __GNUC_MINOR__ >= 4 && __GNUC_MINOR__ < 6
-typedef unsigned int uint128_t __attribute__ ((__mode__ (TI)));
-#elif defined (_MSC_VER)
-/* only for mingw64 on windows */
-#undef  USE_INT128
-#define USE_INT128 (0)
-#else
-typedef __uint128_t uint128_t;
-#endif
-
-#endif
-
-#define LITE 0
-#if LITE /* cryptonight-light */
-#define MEMORY (1 << 20)
-#define ITER   (1 << 19)
-#else
-#define MEMORY (1 << 21) /* 2 MiB */
-#define ITER   (1 << 20)
-#endif
-
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)
-
-/*
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-	union hash_state hs;
-	struct {
-		uint8_t k[64];
-		uint8_t init[INIT_SIZE_BYTE];
-	};
-};
-#pragma pack(pop)
-
-static void do_blake_hash(const void* input, size_t len, char* output) {
-	blake256_hash((uint8_t*)output, input, len);
-}
-
-static void do_groestl_hash(const void* input, size_t len, char* output) {
-	groestl(input, len * 8, (uint8_t*)output);
-}
-
-static void do_jh_hash(const void* input, size_t len, char* output) {
-	int r = jh_hash(HASH_SIZE * 8, input, 8 * len, (uint8_t*)output);
-	assert(likely(SUCCESS == r));
-}
-
-static void do_skein_hash(const void* input, size_t len, char* output) {
-	int r = skein_hash(8 * HASH_SIZE, input, 8 * len, (uint8_t*)output);
-	assert(likely(SKEIN_SUCCESS == r));
-}
-*/
-
-extern int aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#if !defined(_MSC_VER) && !defined(NOASM)
-extern int fast_aesb_single_round(const uint8_t *in, uint8_t*out, const uint8_t *expandedKey);
-extern int fast_aesb_pseudo_round_mut(uint8_t *val, uint8_t *expandedKey);
-#else
-#define fast_aesb_single_round     aesb_single_round
-#define fast_aesb_pseudo_round_mut aesb_pseudo_round_mut
-#endif
-
-
-#if defined(NOASM) || !defined(__x86_64__)
-static uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi) {
-	// multiplier   = ab = a * 2^32 + b
-	// multiplicand = cd = c * 2^32 + d
-	// ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
-	uint64_t a = hi_dword(multiplier);
-	uint64_t b = lo_dword(multiplier);
-	uint64_t c = hi_dword(multiplicand);
-	uint64_t d = lo_dword(multiplicand);
-
-	uint64_t ac = a * c;
-	uint64_t ad = a * d;
-	uint64_t bc = b * c;
-	uint64_t bd = b * d;
-
-	uint64_t adbc = ad + bc;
-	uint64_t adbc_carry = adbc < ad ? 1 : 0;
-
-	// multiplier * multiplicand = product_hi * 2^64 + product_lo
-	uint64_t product_lo = bd + (adbc << 32);
-	uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
-	*product_hi = ac + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
-	assert(ac <= *product_hi);
-
-	return product_lo;
-}
-#else
-extern uint64_t mul128(uint64_t multiplier, uint64_t multiplicand, uint64_t* product_hi);
-#endif
-
-/*
-static void (* const extra_hashes[4])(const void *, size_t, char *) = {
-		do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash
-};
-*/
-
-static inline size_t e2i(const uint8_t* a) {
-#if !LITE
-	return ((uint32_t *)a)[0] & 0x1FFFF0;
-#else
-	return ((uint32_t *)a)[0] & 0xFFFF0;
-#endif
-}
-
-static inline void mul_sum_xor_dst( const uint8_t* a, uint8_t* c, uint8_t* dst, 
-         const uint64_t tweak )
-{
-	uint64_t hi, lo = mul128(((uint64_t*) a)[0], ((uint64_t*) dst)[0], &hi) + ((uint64_t*) c)[1];
-	hi += ((uint64_t*) c)[0];
-
-	((uint64_t*) c)[0] = ((uint64_t*) dst)[0] ^ hi;
-	((uint64_t*) c)[1] = ((uint64_t*) dst)[1] ^ lo;
-	((uint64_t*) dst)[0] = hi;
-	((uint64_t*) dst)[1] = cryptonightV7 ? lo ^ tweak : lo;
-}
-
-static inline void xor_blocks(uint8_t* a, const uint8_t* b) {
-#if USE_INT128
-	*((uint128_t*) a) ^= *((uint128_t*) b);
-#else
-	((uint64_t*) a)[0] ^= ((uint64_t*) b)[0];
-	((uint64_t*) a)[1] ^= ((uint64_t*) b)[1];
-#endif
-}
-
-static inline void xor_blocks_dst(const uint8_t* a, const uint8_t* b, uint8_t* dst) {
-#if USE_INT128
-	*((uint128_t*) dst) = *((uint128_t*) a) ^ *((uint128_t*) b);
-#else
-	((uint64_t*) dst)[0] = ((uint64_t*) a)[0] ^ ((uint64_t*) b)[0];
-	((uint64_t*) dst)[1] = ((uint64_t*) a)[1] ^ ((uint64_t*) b)[1];
-#endif
-}
-
-typedef struct {
-	uint8_t _ALIGN(16) long_state[MEMORY];
-	union cn_slow_hash_state state;
-	uint8_t _ALIGN(16) text[INIT_SIZE_BYTE];
-	uint8_t _ALIGN(16) a[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) b[AES_BLOCK_SIZE];
-	uint8_t _ALIGN(16) c[AES_BLOCK_SIZE];
-	oaes_ctx* aes_ctx;
-} cryptonight_ctx;
-
-static __thread cryptonight_ctx ctx;
-
-void cryptonight_hash_ctx(void* output, const void* input, int len)
-{
-//    hash_process(&ctx.state.hs, (const uint8_t*) input, len);
-    keccak( (const uint8_t*)input, 76, (char*)&ctx.state.hs.b, 200 );
-
-    if ( cryptonightV7 && len < 43 )
-      return;
-    const uint64_t tweak = cryptonightV7
-                         ? *((const uint64_t*) (((const uint8_t*)input) + 35))
-                           ^ ctx.state.hs.w[24] : 0;
-
-    ctx.aes_ctx = (oaes_ctx*) oaes_alloc();
-
-    __builtin_prefetch( ctx.text,             0, 3 );
-    __builtin_prefetch( ctx.text       +  64, 0, 3 );
-    __builtin_prefetch( ctx.long_state,       1, 0 );
-    __builtin_prefetch( ctx.long_state +  64, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 128, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 192, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 256, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 320, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 384, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 448, 1, 0 );
-
-	size_t i, j;
-	memcpy(ctx.text, ctx.state.init, INIT_SIZE_BYTE);
-
-	oaes_key_import_data(ctx.aes_ctx, ctx.state.hs.b, AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-
-    __builtin_prefetch( ctx.long_state + i + 512, 1, 0 );
-    __builtin_prefetch( ctx.long_state + i + 576, 1, 0 );
-
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 0], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 1], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 2], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 3], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 4], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 5], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 6], ctx.aes_ctx->key->exp_data);
-		aesb_pseudo_round_mut(&ctx.text[AES_BLOCK_SIZE * 7], ctx.aes_ctx->key->exp_data);
-		memcpy(&ctx.long_state[i], ctx.text, INIT_SIZE_BYTE);
-	}
-
-	xor_blocks_dst(&ctx.state.k[0], &ctx.state.k[32], ctx.a);
-	xor_blocks_dst(&ctx.state.k[16], &ctx.state.k[48], ctx.b);
-
-	for (i = 0; likely(i < ITER / 4); ++i)
-        {
-           /* Dependency chain: address -> read value ------+
-            * written value <-+ hard function (AES or MUL) <+
-            * next address  <-+
-            */
-           /* Iteration 1 */
-           j = e2i(ctx.a);
-           aesb_single_round(&ctx.long_state[j], ctx.c, ctx.a);
-           xor_blocks_dst(ctx.c, ctx.b, &ctx.long_state[j]);
-
-           if ( cryptonightV7 )
-           {
-              uint8_t *lsa = (uint8_t*)&ctx.long_state[((uint64_t *)(ctx.a))[0] & 0x1FFFF0];
-              const uint8_t tmp = lsa[11];
-              const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-              lsa[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-           }
-
-           /* Iteration 2 */
-           mul_sum_xor_dst(ctx.c, ctx.a, &ctx.long_state[e2i(ctx.c)], tweak );
-
-           /* Iteration 3 */
-           j = e2i(ctx.a);
-           aesb_single_round(&ctx.long_state[j], ctx.b, ctx.a);
-           xor_blocks_dst(ctx.b, ctx.c, &ctx.long_state[j]);
-
-           if ( cryptonightV7 )
-           {
-              uint8_t *lsa = (uint8_t*)&ctx.long_state[((uint64_t *)(ctx.a))[0] & 0x1FFFF0];
-              const uint8_t tmp = lsa[11];
-              const uint8_t index = ( ( (tmp >> 3) & 6 ) | (tmp & 1) ) << 1;
-              lsa[11] = tmp ^ ( ( 0x75310 >> index) & 0x30 );
-           }
-
-           /* Iteration 4 */
-           mul_sum_xor_dst(ctx.b, ctx.a, &ctx.long_state[e2i(ctx.b)], tweak );
-
-	}
-
-    __builtin_prefetch( ctx.text,             0, 3 );
-    __builtin_prefetch( ctx.text       +  64, 0, 3 );
-    __builtin_prefetch( ctx.long_state,       1, 0 );
-    __builtin_prefetch( ctx.long_state +  64, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 128, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 192, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 256, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 320, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 384, 1, 0 );
-    __builtin_prefetch( ctx.long_state + 448, 1, 0 );
-
-	memcpy(ctx.text, ctx.state.init, INIT_SIZE_BYTE);
-	oaes_key_import_data(ctx.aes_ctx, &ctx.state.hs.b[32], AES_KEY_SIZE);
-	for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
-
-    __builtin_prefetch( ctx.long_state + i + 512, 1, 0 );
-    __builtin_prefetch( ctx.long_state + i + 576, 1, 0 );
-
-		xor_blocks(&ctx.text[0 * AES_BLOCK_SIZE], &ctx.long_state[i + 0 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[0 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[1 * AES_BLOCK_SIZE], &ctx.long_state[i + 1 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[1 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[2 * AES_BLOCK_SIZE], &ctx.long_state[i + 2 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[2 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[3 * AES_BLOCK_SIZE], &ctx.long_state[i + 3 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[3 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[4 * AES_BLOCK_SIZE], &ctx.long_state[i + 4 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[4 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[5 * AES_BLOCK_SIZE], &ctx.long_state[i + 5 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[5 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[6 * AES_BLOCK_SIZE], &ctx.long_state[i + 6 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[6 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-		xor_blocks(&ctx.text[7 * AES_BLOCK_SIZE], &ctx.long_state[i + 7 * AES_BLOCK_SIZE]);
-		aesb_pseudo_round_mut(&ctx.text[7 * AES_BLOCK_SIZE], ctx.aes_ctx->key->exp_data);
-	}
-	memcpy(ctx.state.init, ctx.text, INIT_SIZE_BYTE);
-//	hash_permutation(&ctx.state.hs);
-        keccakf( (uint64_t*)&ctx.state.hs.w, 24 );
-	/*memcpy(hash, &state, 32);*/
-	extra_hashes[ctx.state.hs.b[0] & 3](&ctx.state, 200, output);
-	oaes_free((OAES_CTX **) &ctx.aes_ctx);
-}
-
--- a/algo/cryptonight/cryptonight.h
+++ b/algo/cryptonight/cryptonight.h
@@ -1,51 +0,0 @@
-#ifndef __CRYPTONIGHT_H_INCLUDED
-#define __CRYPTONIGHT_H_INCLUDED
-
-#include <stddef.h>
-#include "crypto/oaes_lib.h"
-#include "miner.h"
-
-#define MEMORY         (1 << 21) /* 2 MiB */
-#define MEMORY_M128I   (MEMORY >> 4) // 2 MiB / 16 = 128 ki * __m128i
-#define ITER           (1 << 20)
-#define AES_BLOCK_SIZE  16
-#define AES_KEY_SIZE    32 /*16*/
-#define INIT_SIZE_BLK   8
-#define INIT_SIZE_BYTE (INIT_SIZE_BLK * AES_BLOCK_SIZE)	// 128
-#define INIT_SIZE_M128I (INIT_SIZE_BYTE >> 4) // 8
-
-
-#pragma pack(push, 1)
-union hash_state {
-  uint8_t b[200];
-  uint64_t w[25];
-};
-#pragma pack(pop)
-
-#pragma pack(push, 1)
-union cn_slow_hash_state {
-    union hash_state hs;
-    struct {
-        uint8_t k[64];
-        uint8_t init[INIT_SIZE_BYTE];
-    };
-};
-#pragma pack(pop)
-
-void do_blake_hash(const void* input, size_t len, char* output);
-void do_groestl_hash(const void* input, size_t len, char* output);
-void do_jh_hash(const void* input, size_t len, char* output);
-void do_skein_hash(const void* input, size_t len, char* output);
-void cryptonight_hash_ctx(void* output, const void* input, int len);
-void keccakf(uint64_t st[25], int rounds);
-extern void (* const extra_hashes[4])(const void *, size_t, char *);
-
-int scanhash_cryptonight( struct work *work, uint32_t max_nonce,
-                           uint64_t *hashes_done, struct thr_info *mythr );
-
-void cryptonight_hash_aes( void *restrict output, const void *input, int len );
-
-extern bool cryptonightV7;
-
-#endif
-
--- a/algo/cubehash/cube-hash-2way.c
+++ b/algo/cubehash/cube-hash-2way.c
@@ -179,14 +179,6 @@ int cube_4way_full( cube_4way_context *sp, void *output,  int hashbitlen,
    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] );
    h[ 0] = m512_const1_128( iv[0] );
    h[ 1] = m512_const1_128( iv[1] );
    h[ 2] = m512_const1_128( iv[2] );
@@ -447,14 +439,6 @@ int cube_2way_full( cube_2way_context *sp, void *output, int hashbitlen,
    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] );
    h[ 0] = m256_const1_128( iv[0] );
    h[ 1] = m256_const1_128( iv[1] );
    h[ 2] = m256_const1_128( iv[2] );
--- a/algo/cubehash/cube-hash-2way.h
+++ b/algo/cubehash/cube-hash-2way.h
@@ -28,6 +28,27 @@ int cube_4way_update_close( cube_4way_context *sp, void *output,
 int cube_4way_full( cube_4way_context *sp, void *output, int hashbitlen,
                    const void *data, size_t size );

+int cube_4x256_full( cube_4way_context *sp, void *output, int hashbitlen,
+                     const void *data, size_t size );
+
+#define cube512_4way_init( sp ) cube_4way_update( sp, 512 )
+#define cube512_4way_update cube_4way_update
+#define cube512_4way_update_close cube_4way_update
+#define cube512_4way_close cube_4way_update
+#define cube512_4way_full( sp, output, data, size ) \
+           cube_4way_full( sp, output, 512, data, size )
+#define cube512_4x256_full( sp, output, data, size ) \
+           cube_4x256_full( sp, output, 512, data, size )
+
+#define cube256_4way_init( sp ) cube_4way_update( sp, 256 )
+#define cube256_4way_update cube_4way_update
+#define cube256_4way_update_close cube_4way_update
+#define cube256_4way_close cube_4way_update
+#define cube256_4way_full( sp, output, data, size ) \
+           cube_4way_full( sp, output, 256, data, size )
+#define cube256_4x256_full( sp, output, data, size ) \
+           cube_4x256_full( sp, output, 256, data, size )
+
 #endif

 // 2x128, 2 way parallel SSE2
--- a/algo/cubehash/cubehash_sse2.c
+++ b/algo/cubehash/cubehash_sse2.c
@@ -230,11 +230,10 @@ int cubehashDigest( cubehashParam *sp, byte *digest )

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

-    sp->x[7] = _mm_xor_si128( sp->x[7], _mm_set_epi32( 1,0,0,0 ) );
+    sp->x[7] = _mm_xor_si128( sp->x[7], m128_const_64( 0x100000000, 0 ) );
    transform( sp );
    transform( sp );
    transform( sp );
@@ -276,11 +275,89 @@ int cubehashUpdateDigest( cubehashParam *sp, byte *digest,

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

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

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

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

+int cubehash_full( cubehashParam* sp, byte *digest, int hashbitlen,
+                   const byte *data, size_t size );
+
 #ifdef __cplusplus
 }
 #endif
--- a/algo/echo/echo-hash-4way.h
+++ b/algo/echo/echo-hash-4way.h
@@ -22,18 +22,26 @@ typedef struct
 } echo_4way_context __attribute__ ((aligned (64)));

 int echo_4way_init( echo_4way_context *state, int hashbitlen );
-
+#define echo512_4way_init( state ) echo_4way_init( state, 512 )
+#define echo256_4way_init( state ) echo_4way_init( state, 256 )

 int echo_4way_update( echo_4way_context *state, const void *data,
    unsigned int databitlen);
+#define echo512_4way_update echo_4way_update

 int echo_close( echo_4way_context *state, void *hashval );
+#define echo512_4way_close echo_4way_close

 int echo_4way_update_close( echo_4way_context *state, void *hashval,
                              const void *data, int databitlen );
+#define echo512_4way_update_close echo_4way_update_close

 int echo_4way_full( echo_4way_context *ctx, void *hashval, int nHashSize,
                    const void *data, int datalen );
+#define echo512_4way_full( state, hashval, data, datalen ) \
+           echo_4way_full( state, hashval, 512, data, datalen )
+#define echo256_4way_full( state, hashval, data, datalen ) \
+           echo_4way_full( state, hashval, 256, data, datalen )

 #endif 
 #endif
--- a/algo/echo/sph_echo.c
+++ b/algo/echo/sph_echo.c
@@ -36,6 +36,8 @@

 #include "sph_echo.h"

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

+#if !defined(__AES__)
+
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -316,5 +318,5 @@ void sph_echo512_addbits_and_close(
 #ifdef __cplusplus
 }
 #endif
-
+#endif // !AES
 #endif
--- a/algo/fugue/sph_fugue.h
+++ b/algo/fugue/sph_fugue.h
@@ -74,6 +74,14 @@ void sph_fugue512_close(void *cc, void *dst);
 void sph_fugue512_addbits_and_close(
 	void *cc, unsigned ub, unsigned n, void *dst);

+#define sph_fugue512_full( cc, dst, data, len ) \
+do{ \
+   sph_fugue512_init( cc ); \
+   sph_fugue512( cc, data, len ); \
+   sph_fugue512_close( cc, dst ); \
+}while(0)
+
+
 #ifdef __cplusplus
 }
 #endif
--- a/algo/groestl/aes_ni/groestl-asm-aes.h
+++ b/algo/groestl/aes_ni/groestl-asm-aes.h
--- a/algo/groestl/aes_ni/groestl-asm-avx.h
+++ b/algo/groestl/aes_ni/groestl-asm-avx.h
--- a/algo/groestl/aes_ni/groestl-asm-vperm.h
+++ b/algo/groestl/aes_ni/groestl-asm-vperm.h
--- a/algo/groestl/aes_ni/groestl-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl-intr-aes.h
@@ -1,3 +1,6 @@
+#if !defined GROESTL_INTR_AES_H__
+#define GROESTL_INTR_AES_H__
+
 /* groestl-intr-aes.h     Aug 2011
 *
 * Groestl implementation with intrinsics using ssse3, sse4.1, and aes
@@ -11,16 +14,51 @@
 #include <wmmintrin.h>
 #include "hash-groestl.h"

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

 #define tos(a)    #a
 #define tostr(a)  tos(a)
@@ -111,7 +149,7 @@ __m128i ALL_FF;
  \
  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
  /* compute w_i : add y_{i+4} */\
-  b1 = ALL_1B;\
+  b1 = m128_const1_64( 0x1b1b1b1b1b1b1b1b );\
  MUL2(a0, b0, b1);\
  a0 = _mm_xor_si128(a0, TEMP0);\
  MUL2(a1, b0, b1);\
@@ -152,25 +190,6 @@ __m128i ALL_FF;
 }/*MixBytes*/


-#define SET_CONSTANTS(){\
-  ALL_FF = _mm_set_epi32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff);\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x0306090c, 0x0f020508, 0x0b0e0104, 0x070a0d00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070a0d, 0x00030609, 0x0c0f0205, 0x080b0e01);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05080b0e, 0x0104070a, 0x0d000306, 0x090c0f02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x06090c0f, 0x0205080b, 0x0e010407, 0x0a0d0003);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x070a0d00, 0x0306090c, 0x0f020508, 0x0b0e0104);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x080b0e01, 0x04070a0d, 0x00030609, 0x0c0f0205);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x090c0f02, 0x05080b0e, 0x0104070a, 0x0d000306);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x0e010407, 0x0a0d0003, 0x06090c0f, 0x0205080b);\
-  for(i = 0; i < ROUNDS1024; i++)\
-  {\
-    ROUND_CONST_P[i] = _mm_set_epi32(0xf0e0d0c0 ^ (i * 0x01010101), 0xb0a09080 ^ (i * 0x01010101), 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_Q[i] = _mm_set_epi32(0x0f1f2f3f ^ (i * 0x01010101), 0x4f5f6f7f ^ (i * 0x01010101), 0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101));\
-  }\
-}while(0);\
-
 /* one round
 * a0-a7 = input rows
 * b0-b7 = output rows
@@ -194,30 +213,34 @@ __m128i ALL_FF;
  u8 round_counter = 0;\
  for(round_counter = 0; round_counter < 14; round_counter+=2) {\
    /* AddRoundConstant P1024 */\
-    xmm8 = _mm_xor_si128(xmm8, (ROUND_CONST_P[round_counter]));\
+    xmm8 = _mm_xor_si128( xmm8, \
+             casti_m128i( round_const_p, round_counter ) ); \
     /* ShiftBytes P1024 + pre-AESENCLAST */\
-    xmm8  = _mm_shuffle_epi8(xmm8,  (SUBSH_MASK[0]));\
-    xmm9  = _mm_shuffle_epi8(xmm9,  (SUBSH_MASK[1]));\
-    xmm10 = _mm_shuffle_epi8(xmm10, (SUBSH_MASK[2]));\
-    xmm11 = _mm_shuffle_epi8(xmm11, (SUBSH_MASK[3]));\
-    xmm12 = _mm_shuffle_epi8(xmm12, (SUBSH_MASK[4]));\
-    xmm13 = _mm_shuffle_epi8(xmm13, (SUBSH_MASK[5]));\
-    xmm14 = _mm_shuffle_epi8(xmm14, (SUBSH_MASK[6]));\
-    xmm15 = _mm_shuffle_epi8(xmm15, (SUBSH_MASK[7]));\
+    xmm8  = _mm_shuffle_epi8( xmm8,  SUBSH_MASK0 ); \
+    xmm9  = _mm_shuffle_epi8( xmm9,  SUBSH_MASK1 ); \
+    xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK2 ); \
+    xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK3 ); \
+    xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK4 ); \
+    xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK5 ); \
+    xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK6 ); \
+    xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK7 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
+    SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
+            xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7 ); \
    \
    /* AddRoundConstant P1024 */\
-    xmm0 = _mm_xor_si128(xmm0, (ROUND_CONST_P[round_counter+1]));\
-    xmm0 = _mm_shuffle_epi8(xmm0, (SUBSH_MASK[0]));\
-    xmm1 = _mm_shuffle_epi8(xmm1, (SUBSH_MASK[1]));\
-    xmm2 = _mm_shuffle_epi8(xmm2, (SUBSH_MASK[2]));\
-    xmm3 = _mm_shuffle_epi8(xmm3, (SUBSH_MASK[3]));\
-    xmm4 = _mm_shuffle_epi8(xmm4, (SUBSH_MASK[4]));\
-    xmm5 = _mm_shuffle_epi8(xmm5, (SUBSH_MASK[5]));\
-    xmm6 = _mm_shuffle_epi8(xmm6, (SUBSH_MASK[6]));\
-    xmm7 = _mm_shuffle_epi8(xmm7, (SUBSH_MASK[7]));\
-    SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
+    xmm0 = _mm_xor_si128( xmm0, \
+             casti_m128i( round_const_p, round_counter+1 ) ); \
+    xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK0 ); \
+    xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK1 ); \
+    xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK2 ); \
+    xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK3 ); \
+    xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK4 ); \
+    xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK5 ); \
+    xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK6 ); \
+    xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK7 ); \
+    SUBMIX( xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7, \
+            xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
  }\
 }

@@ -225,48 +248,52 @@ __m128i ALL_FF;
  u8 round_counter = 0;\
  for(round_counter = 0; round_counter < 14; round_counter+=2) {\
    /* AddRoundConstant Q1024 */\
-    xmm1 = ALL_FF;\
-    xmm8  = _mm_xor_si128(xmm8,  xmm1);\
-    xmm9  = _mm_xor_si128(xmm9,  xmm1);\
-    xmm10 = _mm_xor_si128(xmm10, xmm1);\
-    xmm11 = _mm_xor_si128(xmm11, xmm1);\
-    xmm12 = _mm_xor_si128(xmm12, xmm1);\
-    xmm13 = _mm_xor_si128(xmm13, xmm1);\
-    xmm14 = _mm_xor_si128(xmm14, xmm1);\
-    xmm15 = _mm_xor_si128(xmm15, (ROUND_CONST_Q[round_counter]));\
+    xmm1 = m128_neg1;\
+    xmm8  = _mm_xor_si128( xmm8,  xmm1 ); \
+    xmm9  = _mm_xor_si128( xmm9,  xmm1 ); \
+    xmm10 = _mm_xor_si128( xmm10, xmm1 ); \
+    xmm11 = _mm_xor_si128( xmm11, xmm1 ); \
+    xmm12 = _mm_xor_si128( xmm12, xmm1 ); \
+    xmm13 = _mm_xor_si128( xmm13, xmm1 ); \
+    xmm14 = _mm_xor_si128( xmm14, xmm1 ); \
+    xmm15 = _mm_xor_si128( xmm15, \
+              casti_m128i( round_const_q, round_counter ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm8  = _mm_shuffle_epi8(xmm8,  (SUBSH_MASK[1]));\
-    xmm9  = _mm_shuffle_epi8(xmm9,  (SUBSH_MASK[3]));\
-    xmm10 = _mm_shuffle_epi8(xmm10, (SUBSH_MASK[5]));\
-    xmm11 = _mm_shuffle_epi8(xmm11, (SUBSH_MASK[7]));\
-    xmm12 = _mm_shuffle_epi8(xmm12, (SUBSH_MASK[0]));\
-    xmm13 = _mm_shuffle_epi8(xmm13, (SUBSH_MASK[2]));\
-    xmm14 = _mm_shuffle_epi8(xmm14, (SUBSH_MASK[4]));\
-    xmm15 = _mm_shuffle_epi8(xmm15, (SUBSH_MASK[6]));\
+    xmm8  = _mm_shuffle_epi8( xmm8,  SUBSH_MASK1 ); \
+    xmm9  = _mm_shuffle_epi8( xmm9,  SUBSH_MASK3 ); \
+    xmm10 = _mm_shuffle_epi8( xmm10, SUBSH_MASK5 ); \
+    xmm11 = _mm_shuffle_epi8( xmm11, SUBSH_MASK7 ); \
+    xmm12 = _mm_shuffle_epi8( xmm12, SUBSH_MASK0 ); \
+    xmm13 = _mm_shuffle_epi8( xmm13, SUBSH_MASK2 ); \
+    xmm14 = _mm_shuffle_epi8( xmm14, SUBSH_MASK4 ); \
+    xmm15 = _mm_shuffle_epi8( xmm15, SUBSH_MASK6 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
+    SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
+            xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6 , xmm7 ); \
    \
    /* AddRoundConstant Q1024 */\
-    xmm9 = ALL_FF;\
-    xmm0 = _mm_xor_si128(xmm0,  xmm9);\
-    xmm1 = _mm_xor_si128(xmm1,  xmm9);\
-    xmm2 = _mm_xor_si128(xmm2,  xmm9);\
-    xmm3 = _mm_xor_si128(xmm3,  xmm9);\
-    xmm4 = _mm_xor_si128(xmm4,  xmm9);\
-    xmm5 = _mm_xor_si128(xmm5,  xmm9);\
-    xmm6 = _mm_xor_si128(xmm6,  xmm9);\
-    xmm7 = _mm_xor_si128(xmm7,  (ROUND_CONST_Q[round_counter+1]));\
+    xmm9 = m128_neg1;\
+    xmm0 = _mm_xor_si128( xmm0, xmm9 ); \
+    xmm1 = _mm_xor_si128( xmm1, xmm9 ); \
+    xmm2 = _mm_xor_si128( xmm2, xmm9 ); \
+    xmm3 = _mm_xor_si128( xmm3, xmm9 ); \
+    xmm4 = _mm_xor_si128( xmm4, xmm9 ); \
+    xmm5 = _mm_xor_si128( xmm5, xmm9 ); \
+    xmm6 = _mm_xor_si128( xmm6, xmm9 ); \
+    xmm7 = _mm_xor_si128( xmm7, \
+             casti_m128i( round_const_q, round_counter+1 ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm0 = _mm_shuffle_epi8(xmm0, (SUBSH_MASK[1]));\
-    xmm1 = _mm_shuffle_epi8(xmm1, (SUBSH_MASK[3]));\
-    xmm2 = _mm_shuffle_epi8(xmm2, (SUBSH_MASK[5]));\
-    xmm3 = _mm_shuffle_epi8(xmm3, (SUBSH_MASK[7]));\
-    xmm4 = _mm_shuffle_epi8(xmm4, (SUBSH_MASK[0]));\
-    xmm5 = _mm_shuffle_epi8(xmm5, (SUBSH_MASK[2]));\
-    xmm6 = _mm_shuffle_epi8(xmm6, (SUBSH_MASK[4]));\
-    xmm7 = _mm_shuffle_epi8(xmm7, (SUBSH_MASK[6]));\
+    xmm0 = _mm_shuffle_epi8( xmm0, SUBSH_MASK1 ); \
+    xmm1 = _mm_shuffle_epi8( xmm1, SUBSH_MASK3 ); \
+    xmm2 = _mm_shuffle_epi8( xmm2, SUBSH_MASK5 ); \
+    xmm3 = _mm_shuffle_epi8( xmm3, SUBSH_MASK7 ); \
+    xmm4 = _mm_shuffle_epi8( xmm4, SUBSH_MASK0 ); \
+    xmm5 = _mm_shuffle_epi8( xmm5, SUBSH_MASK2 ); \
+    xmm6 = _mm_shuffle_epi8( xmm6, SUBSH_MASK4 ); \
+    xmm7 = _mm_shuffle_epi8( xmm7, SUBSH_MASK6 ); \
    /* SubBytes + MixBytes */\
-    SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
+    SUBMIX( xmm0,  xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7, \
+            xmm8,  xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15 ); \
  }\
 }

@@ -278,7 +305,7 @@ __m128i ALL_FF;
 * clobbers: t0-t7
 */
 #define Matrix_Transpose(i0, i1, i2, i3, i4, i5, i6, i7, t0, t1, t2, t3, t4, t5, t6, t7){\
-  t0 = TRANSP_MASK;\
+  t0 = TRANSP_MASK; \
 \
  i6 = _mm_shuffle_epi8(i6, t0);\
  i0 = _mm_shuffle_epi8(i0, t0);\
@@ -366,7 +393,7 @@ __m128i ALL_FF;
  i4 = _mm_unpacklo_epi64(i4, i5);\
  t1 = _mm_unpackhi_epi64(t1, i5);\
  t2 = i6;\
-  o0 = TRANSP_MASK;\
+  o0 = TRANSP_MASK; \
  i6 = _mm_unpacklo_epi64(i6, i7);\
  t2 = _mm_unpackhi_epi64(t2, i7);\
  /* load transpose mask into a register, because it will be used 8 times */\
@@ -607,3 +634,4 @@ void OF1024( __m128i* chaining )
  return;
 }

+#endif
--- a/algo/groestl/aes_ni/groestl-intr-avx.h
+++ b/algo/groestl/aes_ni/groestl-intr-avx.h
--- a/algo/groestl/aes_ni/groestl-intr-vperm.h
+++ b/algo/groestl/aes_ni/groestl-intr-vperm.h
--- a/algo/groestl/aes_ni/groestl-version.h
+++ b/algo/groestl/aes_ni/groestl-version.h
@@ -1,10 +0,0 @@
-// specify assembly or intrinsics implementation
-//#define TASM
-#define TINTR
-
-// Not to be confused with AVX512VAES
-#define VAES
-// #define VAVX
-// #define VVPERM
-
-//#endif
--- a/algo/groestl/aes_ni/groestl256-asm-aes.h
+++ b/algo/groestl/aes_ni/groestl256-asm-aes.h
@@ -1,529 +0,0 @@
-/* groestl-asm-aes.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3, sse4.1, and aes
- * instructions.
- * Authors: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-/* global constants  */
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (16))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (16))) unsigned char ALL_1B[16];
-__attribute__ ((aligned (16))) unsigned char ALL_FF[16];
-
-/* temporary variables  */
-__attribute__ ((aligned (16))) unsigned char QTEMP[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP[3*16];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b */
-#define MUL2(i, j, k){\
-  asm("pxor xmm"tostr(j)", xmm"tostr(j)"");\
-  asm("pcmpgtb xmm"tostr(j)", xmm"tostr(i)"");\
-  asm("paddb xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("pand xmm"tostr(j)", xmm"tostr(k)"");\
-  asm("pxor xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* t_i = a_i + a_{i+1} */\
-  asm("movdqa xmm"tostr(b6)", xmm"tostr(a0)"");\
-  asm("movdqa xmm"tostr(b7)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(a0)", xmm"tostr(a1)"");\
-  asm("movdqa xmm"tostr(b0)", xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(a1)", xmm"tostr(a2)"");\
-  asm("movdqa xmm"tostr(b1)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(a3)"");\
-  asm("movdqa xmm"tostr(b2)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(a4)"");\
-  asm("movdqa xmm"tostr(b3)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(a5)"");\
-  asm("movdqa xmm"tostr(b4)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(a6)"");\
-  asm("movdqa xmm"tostr(b5)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a7)", xmm"tostr(b6)"");\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a6)"");\
-  /* spill values y_4, y_5 to memory */\
-  asm("movaps [TEMP+0*16], xmm"tostr(b0)"");\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a7)"");\
-  asm("movaps [TEMP+1*16], xmm"tostr(b1)"");\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a0)"");\
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  asm("movdqa xmm"tostr(b0)", xmm"tostr(a0)"");\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a1)"");\
-  asm("movdqa xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a2)"");\
-  asm("movaps [TEMP+2*16], xmm"tostr(a2)"");\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  asm("pxor xmm"tostr(a0)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(a1)", xmm"tostr(a4)"");\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(a5)"");\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(a6)"");\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(a7)"");\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm("pxor xmm"tostr(a7)", [TEMP+2*16]");\
-  \
-  /* compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  /* compute w_i : add y_{i+4} */\
-  asm("movaps xmm"tostr(b1)", [ALL_1B]");\
-  MUL2(a0, b0, b1);\
-  asm("pxor xmm"tostr(a0)", [TEMP+0*16]");\
-  MUL2(a1, b0, b1);\
-  asm("pxor xmm"tostr(a1)", [TEMP+1*16]");\
-  MUL2(a2, b0, b1);\
-  asm("pxor xmm"tostr(a2)", xmm"tostr(b2)"");\
-  MUL2(a3, b0, b1);\
-  asm("pxor xmm"tostr(a3)", xmm"tostr(b3)"");\
-  MUL2(a4, b0, b1);\
-  asm("pxor xmm"tostr(a4)", xmm"tostr(b4)"");\
-  MUL2(a5, b0, b1);\
-  asm("pxor xmm"tostr(a5)", xmm"tostr(b5)"");\
-  MUL2(a6, b0, b1);\
-  asm("pxor xmm"tostr(a6)", xmm"tostr(b6)"");\
-  MUL2(a7, b0, b1);\
-  asm("pxor xmm"tostr(a7)", xmm"tostr(b7)"");\
-  \
-  /* compute v_i : double w_i      */\
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  MUL2(a0, b0, b1);\
-  asm("pxor xmm"tostr(b5)", xmm"tostr(a0)"");\
-  MUL2(a1, b0, b1);\
-  asm("pxor xmm"tostr(b6)", xmm"tostr(a1)"");\
-  MUL2(a2, b0, b1);\
-  asm("pxor xmm"tostr(b7)", xmm"tostr(a2)"");\
-  MUL2(a5, b0, b1);\
-  asm("pxor xmm"tostr(b2)", xmm"tostr(a5)"");\
-  MUL2(a6, b0, b1);\
-  asm("pxor xmm"tostr(b3)", xmm"tostr(a6)"");\
-  MUL2(a7, b0, b1);\
-  asm("pxor xmm"tostr(b4)", xmm"tostr(a7)"");\
-  MUL2(a3, b0, b1);\
-  MUL2(a4, b0, b1);\
-  asm("movaps xmm"tostr(b0)", [TEMP+0*16]");\
-  asm("movaps xmm"tostr(b1)", [TEMP+1*16]");\
-  asm("pxor xmm"tostr(b0)", xmm"tostr(a3)"");\
-  asm("pxor xmm"tostr(b1)", xmm"tostr(a4)"");\
-}/*MixBytes*/
-
-#define SET_CONSTANTS(){\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0c0f0104070b0e00ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x03060a0d08020509ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0e090205000d0801ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x04070c0f0a03060bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x080b0306010f0a02ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x05000e090c04070dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0a0d040702090c03ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0601080b0e05000fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0b0e0500030a0d04ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0702090c0f060108ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0d080601040c0f05ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x00030b0e0907020aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0f0a0702050e0906ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x01040d080b00030cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x090c000306080b07ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x02050f0a0d01040eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}while(0);
-
-#define Push_All_Regs() do{\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}while(0);
-
-#define Pop_All_Regs() do{\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}while(0);
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant */\
-  asm ("movaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("pxor   xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  /* ShiftBytes + SubBytes (interleaved) */\
-  asm ("pxor xmm"tostr(b0)",  xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("aesenclast xmm"tostr(a0)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("aesenclast xmm"tostr(a1)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("aesenclast xmm"tostr(a2)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("aesenclast xmm"tostr(a3)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("aesenclast xmm"tostr(a4)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("aesenclast xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("aesenclast xmm"tostr(a6)", xmm"tostr(b0)"");\
-  asm ("pshufb     xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  asm ("aesenclast xmm"tostr(a7)", xmm"tostr(b0)"");\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("movaps xmm"tostr(t0)", [TRANSP_MASK]");\
-  \
-  asm ("pshufb xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i3)", xmm"tostr(t0)"");\
-  \
-  asm ("movdqa xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(i2)"");\
-  \
-  asm ("punpcklwd xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhwd xmm"tostr(o1)", xmm"tostr(i1)"");\
-  asm ("punpcklwd xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhwd xmm"tostr(t0)", xmm"tostr(i3)"");\
-  \
-  asm ("pshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("pshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("pshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("pshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-  \
-  asm ("movdqa xmm"tostr(o2)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(o3)", xmm"tostr(o1)"");\
-  \
-  asm ("punpckldq xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("punpckldq xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("punpckhdq xmm"tostr(o2)", xmm"tostr(i2)"");\
-  asm ("punpckhdq xmm"tostr(o3)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i4)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o4)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o5)", xmm"tostr(i2)"");\
-  asm ("movdqa     xmm"tostr(o6)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o4)", xmm"tostr(i6)"");\
-  asm ("punpckhqdq xmm"tostr(o5)", xmm"tostr(i6)"");\
-  asm ("movdqa     xmm"tostr(o7)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o7)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("movdqa     xmm"tostr(o0)", xmm"tostr(i0)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhqdq xmm"tostr(o0)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i3)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i4)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("pxor xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(i1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(i3)", xmm"tostr(i2)"");\
-  asm ("movdqa xmm"tostr(i5)", xmm"tostr(i4)"");\
-  asm ("movdqa xmm"tostr(i7)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i3)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i5)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i7)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("movaps xmm12, [rdi+0*16]");
-  asm ("movaps xmm13, [rdi+1*16]");
-  asm ("movaps xmm14, [rdi+2*16]");
-  asm ("movaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("movaps [rdi+0*16], xmm12");
-  asm ("movaps [rdi+1*16], xmm2");
-  asm ("movaps [rdi+2*16], xmm6");
-  asm ("movaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("movaps xmm12, [rsi+0*16]");
-  asm ("movaps xmm13, [rsi+1*16]");
-  asm ("movaps xmm14, [rsi+2*16]");
-  asm ("movaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (2x64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  asm ("movaps xmm8, [rdi+0*16]");
-  asm ("movaps xmm0, [rdi+1*16]");
-  asm ("movaps xmm4, [rdi+2*16]");
-  asm ("movaps xmm5, [rdi+3*16]");
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("pxor xmm8, xmm12");
-  asm ("pxor xmm0, xmm2");
-  asm ("pxor xmm4, xmm6");
-  asm ("pxor xmm5, xmm7");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("pxor xmm0, xmm8");
-  asm ("pxor xmm1, xmm10");
-  asm ("pxor xmm2, xmm12");
-  asm ("pxor xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("pxor xmm0, [rdi+0*16]");
-  asm ("pxor xmm1, [rdi+1*16]");
-  asm ("pxor xmm2, [rdi+2*16]");
-  asm ("pxor xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("movaps [rdi+0*16], xmm0");
-  asm ("movaps [rdi+1*16], xmm1");
-  asm ("movaps [rdi+2*16], xmm2");
-  asm ("movaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("movaps xmm8,  [rdi+0*16]");
-  asm ("movaps xmm10, [rdi+1*16]");
-  asm ("movaps xmm12, [rdi+2*16]");
-  asm ("movaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("pxor xmm8,  [rdi+0*16]");
-  asm ("pxor xmm10, [rdi+1*16]");
-  asm ("pxor xmm12, [rdi+2*16]");
-  asm ("pxor xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-
-  /* we only need to return the truncated half of the state */
-  asm ("movaps [rdi+2*16], xmm9");
-  asm ("movaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
--- a/algo/groestl/aes_ni/groestl256-asm-avx.h
+++ b/algo/groestl/aes_ni/groestl256-asm-avx.h
@@ -1,519 +0,0 @@
-/* groestl-asm-avx.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3, sse4.1, aes and avx
- * instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-
-/* global variables  */
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (32))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (32))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (32))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (32))) unsigned char ALL_1B[32];
-__attribute__ ((aligned (32))) unsigned char ALL_FF[32];
-
-/* temporary variables  */
-__attribute__ ((aligned (32))) unsigned char TEMP[6*32];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_CONSTANTS(){\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0c0f0104070b0e00ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x03060a0d08020509ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0e090205000d0801ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x04070c0f0a03060bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x080b0306010f0a02ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x05000e090c04070dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0a0d040702090c03ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0601080b0e05000fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0b0e0500030a0d04ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0702090c0f060108ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0d080601040c0f05ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x00030b0e0907020aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0f0a0702050e0906ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x01040d080b00030cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x090c000306080b07ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x02050f0a0d01040eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}while(0);
-
-#define Push_All_Regs() do{\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}while(0);
-
-#define Pop_All_Regs() do{\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}while(0);
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2(i, j, k, z){\
-  asm("vpcmpgtb xmm"tostr(j)", xmm"tostr(z)", xmm"tostr(i)"");\
-  asm("vpaddb xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("vpand xmm"tostr(j)", xmm"tostr(j)", xmm"tostr(k)"");\
-  asm("vpxor xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2v2(i, j, k, z){\
-  asm("vpblendvb xmm"tostr(j)", xmm"tostr(z)", xmm"tostr(k)", xmm"tostr(i)"");\
-  asm("vpaddb xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(i)"");\
-  asm("vpxor xmm"tostr(i)", xmm"tostr(i)", xmm"tostr(j)"");\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* xmm"tostr(8..xmm"tostr(15 = a2 a3... a0 a1 */\
-  asm("vmovdqa xmm"tostr(b0)", xmm"tostr(a2)"");\
-  asm("vmovdqa xmm"tostr(b1)", xmm"tostr(a3)"");\
-  asm("vmovdqa xmm"tostr(b2)", xmm"tostr(a4)"");\
-  asm("vmovdqa xmm"tostr(b3)", xmm"tostr(a5)"");\
-  asm("vmovdqa xmm"tostr(b4)", xmm"tostr(a6)"");\
-  asm("vmovdqa xmm"tostr(b5)", xmm"tostr(a7)"");\
-  asm("vmovdqa xmm"tostr(b6)", xmm"tostr(a0)"");\
-  asm("vmovdqa xmm"tostr(b7)", xmm"tostr(a1)"");\
-  \
-  /* t_i = a_i + a_{i+1} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b6)"");\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a3)"");\
-  \
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(b0)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(b1)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a2)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a5)"");\
-  \
-  /* spill values y_4, y_5 to memory */\
-  asm("vmovaps [TEMP+0*16], xmm"tostr(b0)"");\
-  asm("vmovaps [TEMP+1*16], xmm"tostr(b1)"");\
-  asm("vmovaps [TEMP+2*16], xmm"tostr(b2)"");\
-  \
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  asm("vmovdqa xmm"tostr(b0)", xmm"tostr(a0)"");\
-  asm("vmovdqa xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm("vmovaps [TEMP+3*16], xmm"tostr(a2)"");\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(a3)"");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(a4)"");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(a5)"");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", [TEMP+3*16]");\
-  \
-  /*compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  asm("vmovaps xmm"tostr(b1)", [ALL_1B]");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(b2)", xmm"tostr(b2)"");\
-  VMUL2(a7, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a0, b0, b1, b2);\
-  \
-  /* compute w_i :  add y_{i+4} */\
-  asm("vpxor xmm"tostr(a0)", xmm"tostr(a0)", [TEMP+0*16]");\
-  asm("vpxor xmm"tostr(a1)", xmm"tostr(a1)", [TEMP+1*16]");\
-  asm("vpxor xmm"tostr(a2)", xmm"tostr(a2)", [TEMP+2*16]");\
-  asm("vpxor xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b3)"");\
-  asm("vpxor xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b4)"");\
-  asm("vpxor xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b5)"");\
-  asm("vpxor xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b6)"");\
-  asm("vpxor xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b7)"");\
-  \
-  /*compute v_i: double w_i */\
-  VMUL2(a0, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  \
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  asm("vpxor xmm"tostr(b0)", xmm"tostr(a3)", [TEMP+0*16]");\
-  asm("vpxor xmm"tostr(b1)", xmm"tostr(a4)", [TEMP+1*16]");\
-  asm("vpxor xmm"tostr(b2)", xmm"tostr(a5)", [TEMP+2*16]");\
-  asm("vpxor xmm"tostr(b3)", xmm"tostr(b3)", xmm"tostr(a6)"");\
-  asm("vpxor xmm"tostr(b4)", xmm"tostr(b4)", xmm"tostr(a7)"");\
-  asm("vpxor xmm"tostr(b5)", xmm"tostr(b5)", xmm"tostr(a0)"");\
-  asm("vpxor xmm"tostr(b6)", xmm"tostr(b6)", xmm"tostr(a1)"");\
-  asm("vpxor xmm"tostr(b7)", xmm"tostr(b7)", xmm"tostr(a2)"");\
-}/*MixBytes*/
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant */\
-  asm ("vmovaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("vpxor   xmm"tostr(a0)", xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("vpxor   xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("vpxor   xmm"tostr(a7)", xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  /* ShiftBytes + SubBytes (interleaved) */\
-  asm ("vpxor xmm"tostr(b0)",  xmm"tostr(b0)",  xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a0)", xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("vaesenclast xmm"tostr(a0)", xmm"tostr(a0)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a1)", xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("vaesenclast xmm"tostr(a1)", xmm"tostr(a1)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a2)", xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("vaesenclast xmm"tostr(a2)", xmm"tostr(a2)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a3)", xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("vaesenclast xmm"tostr(a3)", xmm"tostr(a3)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a4)", xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("vaesenclast xmm"tostr(a4)", xmm"tostr(a4)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a5)", xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("vaesenclast xmm"tostr(a5)", xmm"tostr(a5)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a6)", xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("vaesenclast xmm"tostr(a6)", xmm"tostr(a6)", xmm"tostr(b0)"");\
-  asm ("vpshufb     xmm"tostr(a7)", xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  asm ("vaesenclast xmm"tostr(a7)", xmm"tostr(a7)", xmm"tostr(b0)"");\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
-
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("vmovaps xmm"tostr(t0)", [TRANSP_MASK]");\
-\
-  asm ("vpshufb xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i1)", xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpshufb xmm"tostr(i3)", xmm"tostr(i3)", xmm"tostr(t0)"");\
-\
-  asm ("vpunpckhwd xmm"tostr(o1)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklwd xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpckhwd xmm"tostr(t0)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklwd xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-\
-  asm ("vpshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("vpshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("vpshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("vpshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-\
-  asm ("vpunpckhdq xmm"tostr(o2)", xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("vpunpckhdq xmm"tostr(o3)", xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("vpunpckldq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("vpunpckldq xmm"tostr(o1)", xmm"tostr(o1)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("vpunpckhqdq xmm"tostr(o1)", xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("vpunpcklqdq xmm"tostr(o2)", xmm"tostr(i1)", xmm"tostr(i5)"");\
-  asm ("vpunpckhqdq xmm"tostr(o3)", xmm"tostr(i1)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(o4)", xmm"tostr(i2)", xmm"tostr(i6)"");\
-  asm ("vpunpckhqdq xmm"tostr(o5)", xmm"tostr(i2)", xmm"tostr(i6)"");\
-  asm ("vpunpcklqdq xmm"tostr(o6)", xmm"tostr(i3)", xmm"tostr(i7)"");\
-  asm ("vpunpckhqdq xmm"tostr(o7)", xmm"tostr(i3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("vpunpckhqdq xmm"tostr(o0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpckhqdq xmm"tostr(o1)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpckhqdq xmm"tostr(o2)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpckhqdq xmm"tostr(o3)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("vpxor xmm"tostr(t0)", xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i1)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i3)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i5)", xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("vpunpckhqdq xmm"tostr(i7)", xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("vpunpcklqdq xmm"tostr(i0)", xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("vpunpcklqdq xmm"tostr(i2)", xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("vpunpcklqdq xmm"tostr(i4)", xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("vpunpcklqdq xmm"tostr(i6)", xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("vmovaps xmm12, [rdi+0*16]");
-  asm ("vmovaps xmm13, [rdi+1*16]");
-  asm ("vmovaps xmm14, [rdi+2*16]");
-  asm ("vmovaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("vmovaps [rdi+0*16], xmm12");
-  asm ("vmovaps [rdi+1*16], xmm2");
-  asm ("vmovaps [rdi+2*16], xmm6");
-  asm ("vmovaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("vmovaps xmm12, [rsi+0*16]");
-  asm ("vmovaps xmm13, [rsi+1*16]");
-  asm ("vmovaps xmm14, [rsi+2*16]");
-  asm ("vmovaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value and xor message to CV to get input of P */
-  /* we first put two rows (2x64 bit) of the CV into one 128-bit xmm register */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("vpxor xmm8, xmm12, [rdi+0*16]");
-  asm ("vpxor xmm0, xmm2,  [rdi+1*16]");
-  asm ("vpxor xmm4, xmm6,  [rdi+2*16]");
-  asm ("vpxor xmm5, xmm7,  [rdi+3*16]");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("vpxor xmm0, xmm0, xmm8");
-  asm ("vpxor xmm1, xmm1, xmm10");
-  asm ("vpxor xmm2, xmm2, xmm12");
-  asm ("vpxor xmm3, xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("vpxor xmm0, xmm0, [rdi+0*16]");
-  asm ("vpxor xmm1, xmm1, [rdi+1*16]");
-  asm ("vpxor xmm2, xmm2, [rdi+2*16]");
-  asm ("vpxor xmm3, xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("vmovaps [rdi+0*16], xmm0");
-  asm ("vmovaps [rdi+1*16], xmm1");
-  asm ("vmovaps [rdi+2*16], xmm2");
-  asm ("vmovaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("vmovaps xmm8,  [rdi+0*16]");
-  asm ("vmovaps xmm10, [rdi+1*16]");
-  asm ("vmovaps xmm12, [rdi+2*16]");
-  asm ("vmovaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("vpxor xmm8,  xmm8,  [rdi+0*16]");
-  asm ("vpxor xmm10, xmm10, [rdi+1*16]");
-  asm ("vpxor xmm12, xmm12, [rdi+2*16]");
-  asm ("vpxor xmm14, xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-
-  /* we only need to return the truncated half of the state */
-  asm ("vmovaps [rdi+2*16], xmm9");
-  asm ("vmovaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
--- a/algo/groestl/aes_ni/groestl256-asm-vperm.h
+++ b/algo/groestl/aes_ni/groestl256-asm-vperm.h
@@ -1,856 +0,0 @@
-/* groestl-asm-vperm.h     Aug 2011
- *
- * Groestl implementation with inline assembly using ssse3 instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * Based on the vperm and aes_ni implementations of the hash function Groestl
- * by Cagdas Calik <ccalik@metu.edu.tr> http://www.metu.edu.tr/~ccalik/
- * Institute of Applied Mathematics, Middle East Technical University, Turkey
- *
- * This code is placed in the public domain
- */
-
-#include "hash-groestl256.h"
-
-/* global constants  */
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Lx[16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L0[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_L7[ROUNDS512*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_P[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char ROUND_CONST_Q[ROUNDS1024*16];
-__attribute__ ((aligned (16))) unsigned char TRANSP_MASK[16];
-__attribute__ ((aligned (16))) unsigned char SUBSH_MASK[8*16];
-__attribute__ ((aligned (16))) unsigned char ALL_0F[16];
-__attribute__ ((aligned (16))) unsigned char ALL_15[16];
-__attribute__ ((aligned (16))) unsigned char ALL_1B[16];
-__attribute__ ((aligned (16))) unsigned char ALL_63[16];
-__attribute__ ((aligned (16))) unsigned char ALL_FF[16];
-__attribute__ ((aligned (16))) unsigned char VPERM_IPT[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_OPT[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_INV[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB1[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB2[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SB4[2*16];
-__attribute__ ((aligned (16))) unsigned char VPERM_SBO[2*16];
-
-/* temporary variables  */
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL1[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL2[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP_MUL4[1*16];
-__attribute__ ((aligned (16))) unsigned char QTEMP[8*16];
-__attribute__ ((aligned (16))) unsigned char TEMP[8*16];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_SHARED_CONSTANTS(){\
-  ((u64*)TRANSP_MASK)[0] = 0x0d0509010c040800ULL;\
-  ((u64*)TRANSP_MASK)[1] = 0x0f070b030e060a02ULL;\
-  ((u64*)ALL_1B)[0] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_1B)[1] = 0x1b1b1b1b1b1b1b1bULL;\
-  ((u64*)ALL_63)[ 0] = 0x6363636363636363ULL;\
-  ((u64*)ALL_63)[ 1] = 0x6363636363636363ULL;\
-  ((u64*)ALL_0F)[ 0] = 0x0F0F0F0F0F0F0F0FULL;\
-  ((u64*)ALL_0F)[ 1] = 0x0F0F0F0F0F0F0F0FULL;\
-  ((u64*)VPERM_IPT)[ 0] = 0x4C01307D317C4D00ULL;\
-  ((u64*)VPERM_IPT)[ 1] = 0xCD80B1FCB0FDCC81ULL;\
-  ((u64*)VPERM_IPT)[ 2] = 0xC2B2E8985A2A7000ULL;\
-  ((u64*)VPERM_IPT)[ 3] = 0xCABAE09052227808ULL;\
-  ((u64*)VPERM_OPT)[ 0] = 0x01EDBD5150BCEC00ULL;\
-  ((u64*)VPERM_OPT)[ 1] = 0xE10D5DB1B05C0CE0ULL;\
-  ((u64*)VPERM_OPT)[ 2] = 0xFF9F4929D6B66000ULL;\
-  ((u64*)VPERM_OPT)[ 3] = 0xF7974121DEBE6808ULL;\
-  ((u64*)VPERM_INV)[ 0] = 0x01040A060F0B0780ULL;\
-  ((u64*)VPERM_INV)[ 1] = 0x030D0E0C02050809ULL;\
-  ((u64*)VPERM_INV)[ 2] = 0x0E05060F0D080180ULL;\
-  ((u64*)VPERM_INV)[ 3] = 0x040703090A0B0C02ULL;\
-  ((u64*)VPERM_SB1)[ 0] = 0x3618D415FAE22300ULL;\
-  ((u64*)VPERM_SB1)[ 1] = 0x3BF7CCC10D2ED9EFULL;\
-  ((u64*)VPERM_SB1)[ 2] = 0xB19BE18FCB503E00ULL;\
-  ((u64*)VPERM_SB1)[ 3] = 0xA5DF7A6E142AF544ULL;\
-  ((u64*)VPERM_SB2)[ 0] = 0x69EB88400AE12900ULL;\
-  ((u64*)VPERM_SB2)[ 1] = 0xC2A163C8AB82234AULL;\
-  ((u64*)VPERM_SB2)[ 2] = 0xE27A93C60B712400ULL;\
-  ((u64*)VPERM_SB2)[ 3] = 0x5EB7E955BC982FCDULL;\
-  ((u64*)VPERM_SB4)[ 0] = 0x3D50AED7C393EA00ULL;\
-  ((u64*)VPERM_SB4)[ 1] = 0xBA44FE79876D2914ULL;\
-  ((u64*)VPERM_SB4)[ 2] = 0xE1E937A03FD64100ULL;\
-  ((u64*)VPERM_SB4)[ 3] = 0xA876DE9749087E9FULL;\
-/*((u64*)VPERM_SBO)[ 0] = 0xCFE474A55FBB6A00ULL;\
-  ((u64*)VPERM_SBO)[ 1] = 0x8E1E90D1412B35FAULL;\
-  ((u64*)VPERM_SBO)[ 2] = 0xD0D26D176FBDC700ULL;\
-  ((u64*)VPERM_SBO)[ 3] = 0x15AABF7AC502A878ULL;*/\
-  ((u64*)ALL_15)[ 0] = 0x1515151515151515ULL;\
-  ((u64*)ALL_15)[ 1] = 0x1515151515151515ULL;\
-}/**/
-
-/* VPERM
- * Transform w/o settings c*
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2){\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(t1)", xmm"tostr(c0)"");\
-  asm ("pandn  xmm"tostr(t0)", xmm"tostr(a0)"");\
-  asm ("pandn  xmm"tostr(t1)", xmm"tostr(a1)"");\
-  asm ("psrld  xmm"tostr(t0)", 4");\
-  asm ("psrld  xmm"tostr(t1)", 4");\
-  asm ("pand   xmm"tostr(a0)", xmm"tostr(c0)"");\
-  asm ("pand   xmm"tostr(a1)", xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(t2)", xmm"tostr(c2)"");\
-  asm ("movdqa xmm"tostr(t3)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(t2)", xmm"tostr(a0)"");\
-  asm ("pshufb xmm"tostr(t3)", xmm"tostr(a1)"");\
-  asm ("movdqa xmm"tostr(a0)", xmm"tostr(c1)"");\
-  asm ("movdqa xmm"tostr(a1)", xmm"tostr(c1)"");\
-  asm ("pshufb xmm"tostr(a0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(a1)", xmm"tostr(t1)"");\
-  asm ("pxor   xmm"tostr(a0)", xmm"tostr(t2)"");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(t3)"");\
-}/**/
-
-#define VPERM_Transform_Set_Const(table, c0, c1, c2){\
-  asm ("movaps xmm"tostr(c0)", [ALL_0F]");\
-  asm ("movaps xmm"tostr(c1)", ["tostr(table)"+0*16]");\
-  asm ("movaps xmm"tostr(c2)", ["tostr(table)"+1*16]");\
-}/**/
-
-/* VPERM
- * Transform
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform(a0, a1, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Transform State
- * inputs:
- * a0-a3 = state
- * table = transformation table to use
- * t* = clobbers
- * outputs:
- * a0-a3 = transformed state
- * */
-#define VPERM_Transform_State(a0, a1, a2, a3, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-  VPERM_Transform_No_Const(a2, a3, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Add Constant to State
- * inputs:
- * a0-a7 = state
- * constant = constant to add
- * t0 = clobber
- * outputs:
- * a0-a7 = state + constant
- * */
-#define VPERM_Add_Constant(a0, a1, a2, a3, a4, a5, a6, a7, constant, t0){\
-  asm ("movaps xmm"tostr(t0)", ["tostr(constant)"]");\
-  asm ("pxor   xmm"tostr(a0)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a1)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a2)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a3)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a4)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a5)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a6)",  xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(a7)",  xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * Set Substitute Core Constants
- * */
-#define VPERM_Substitute_Core_Set_Const(c0, c1, c2){\
-  VPERM_Transform_Set_Const(VPERM_INV, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Substitute Core
- * first part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0 = 1 row
- * t*, c* = clobbers
- * outputs:
- * b0a, b0b = inputs for lookup step
- * */
-#define VPERM_Substitute_Core(a0, b0a, b0b, t0, t1, c0, c1, c2){\
-  asm ("movdqa xmm"tostr(t0)",  xmm"tostr(c0)"");\
-  asm ("pandn  xmm"tostr(t0)",  xmm"tostr(a0)"");\
-  asm ("psrld  xmm"tostr(t0)",  4");\
-  asm ("pand   xmm"tostr(a0)",  xmm"tostr(c0)"");\
-  asm ("movdqa xmm"tostr(b0a)", "tostr(c1)"");\
-  asm ("pshufb xmm"tostr(b0a)", xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(a0)",  xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(b0b)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0b)", xmm"tostr(t0)"");\
-  asm ("pxor   xmm"tostr(b0b)", xmm"tostr(b0a)"");\
-  asm ("movdqa xmm"tostr(t1)",  xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(t1)",  xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(t1)",  xmm"tostr(b0a)"");\
-  asm ("movdqa xmm"tostr(b0a)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0a)", xmm"tostr(b0b)"");\
-  asm ("pxor   xmm"tostr(b0a)", xmm"tostr(a0)"");\
-  asm ("movdqa xmm"tostr(b0b)", xmm"tostr(c2)"");\
-  asm ("pshufb xmm"tostr(b0b)", xmm"tostr(t1)"");\
-  asm ("pxor   xmm"tostr(b0b)", xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * Lookup
- * second part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0a, a0b = output of Substitution Core
- * table = lookup table to use (*1 / *2 / *4)
- * t0 = clobber
- * outputs:
- * b0 = output of sbox + multiplication
- * */
-#define VPERM_Lookup(a0a, a0b, table, b0, t0){\
-  asm ("movaps xmm"tostr(b0)", ["tostr(table)"+0*16]");\
-  asm ("movaps xmm"tostr(t0)", ["tostr(table)"+1*16]");\
-  asm ("pshufb xmm"tostr(b0)", xmm"tostr(a0b)"");\
-  asm ("pshufb xmm"tostr(t0)", xmm"tostr(a0a)"");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(t0)"");\
-}/**/
-
-/* VPERM
- * SubBytes and *2 / *4
- * this function is derived from:
- *   Constant-time SSSE3 AES core implementation
- *   by Mike Hamburg
- * and
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0-a7 = state
- * t*, c* = clobbers
- * outputs:
- * a0-a7 = state * 4
- * c2 = row0 * 2 -> b0
- * c1 = row7 * 2 -> b3
- * c0 = row7 * 1 -> b4
- * t2 = row4 * 1 -> b7
- * TEMP_MUL1 = row(i) * 1
- * TEMP_MUL2 = row(i) * 2
- *
- * call:VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7) */
-#define VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, t0, t1, t3, t4, c2, c1, c0, t2){\
-  /* set Constants */\
-  VPERM_Substitute_Core_Set_Const(c0, c1, c2);\
-  /* row 1 */\
-  VPERM_Substitute_Core(a1, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+1*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+1*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a1, t4);\
-  /* --- */\
-  /* row 2 */\
-  VPERM_Substitute_Core(a2, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+2*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+2*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a2, t4);\
-  /* --- */\
-  /* row 3 */\
-  VPERM_Substitute_Core(a3, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+3*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+3*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a3, t4);\
-  /* --- */\
-  /* row 5 */\
-  VPERM_Substitute_Core(a5, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+5*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+5*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a5, t4);\
-  /* --- */\
-  /* row 6 */\
-  VPERM_Substitute_Core(a6, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+6*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+6*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a6, t4);\
-  /* --- */\
-  /* row 7 */\
-  VPERM_Substitute_Core(a7, t0, t1, t3, t4, c0, xmm##c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  asm ("movaps [TEMP_MUL1+7*16], xmm"tostr(t2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c1, t4); /*c1 -> b3*/\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a7, t4);\
-  /* --- */\
-  /* row 4 */\
-  VPERM_Substitute_Core(a4, t0, t1, t3, t4, c0, [VPERM_INV+0*16], c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4); /*t2 -> b7*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  asm ("movaps [TEMP_MUL2+4*16], xmm"tostr(t3)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a4, t4);\
-  /* --- */\
-  /* row 0 */\
-  VPERM_Substitute_Core(a0, t0, t1, t3, t4, c0, [VPERM_INV+0*16], c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, c0, t4); /*c0 -> b4*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c2, t4); /*c2 -> b0*/\
-  asm ("movaps [TEMP_MUL2+0*16], xmm"tostr(c2)"");\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a0, t4);\
-  /* --- */\
-}/**/
-
-
-/* Optimized MixBytes
- * inputs:
- * a0-a7 = (row0-row7) * 4
- * b0 = row0 * 2
- * b3 = row7 * 2
- * b4 = row7 * 1
- * b7 = row4 * 1
- * all *1 and *2 values must also be in TEMP_MUL1, TEMP_MUL2
- * output: b0-b7
- * */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* save one value */\
-  asm ("movaps [TEMP_MUL4], xmm"tostr(a3)"");\
-  /* 1 */\
-  asm ("movdqa xmm"tostr(b1)", xmm"tostr(a0)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(b4)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b1)", [TEMP_MUL2+3*16]");\
-  asm ("movdqa xmm"tostr(b2)", xmm"tostr(b1)"");\
-  \
-  /* 2 */\
-  asm ("movdqa xmm"tostr(b5)", xmm"tostr(a1)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a4)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(b7)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(b3)""); /* -> helper! */\
-  asm ("movdqa xmm"tostr(b6)", xmm"tostr(b5)"");\
-  \
-  /* 4 */\
-  asm ("pxor   xmm"tostr(b7)", xmm"tostr(a6)"");\
-  /*asm ("pxor   xmm"tostr(b7)", [TEMP_MUL1+4*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b7)", [TEMP_MUL1+6*16]");\
-  asm ("pxor   xmm"tostr(b7)", [TEMP_MUL2+1*16]");\
-  asm ("pxor   xmm"tostr(b7)", xmm"tostr(b3)""); /* -> helper! */\
-  asm ("pxor   xmm"tostr(b2)", xmm"tostr(b7)"");\
-  \
-  /* 3 */\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(a7)"");\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL1+5*16]");\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL1+7*16]");\
-  /*asm ("pxor   xmm"tostr(b0)", [TEMP_MUL2+0*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b0)", [TEMP_MUL2+2*16]");\
-  asm ("movdqa xmm"tostr(b3)", xmm"tostr(b0)"");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(b0)"");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(b7)""); /* moved from 4 */\
-  \
-  /* 5 */\
-  asm ("pxor   xmm"tostr(b4)", xmm"tostr(a2)"");\
-  /*asm ("pxor   xmm"tostr(b4)", [TEMP_MUL1+0*16]"); -> helper! */\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL1+2*16]");\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL2+3*16]");\
-  asm ("pxor   xmm"tostr(b4)", [TEMP_MUL2+5*16]");\
-  asm ("pxor   xmm"tostr(b3)", xmm"tostr(b4)"");\
-  asm ("pxor   xmm"tostr(b6)", xmm"tostr(b4)"");\
-  \
-  /* 6 */\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL1+1*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL1+3*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL2+4*16]");\
-  asm ("pxor xmm"tostr(a3)", [TEMP_MUL2+6*16]");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a3)"");\
-  asm ("pxor xmm"tostr(b5)", xmm"tostr(a3)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a3)"");\
-  \
-  /* 7 */\
-  asm ("pxor xmm"tostr(a1)", [TEMP_MUL1+1*16]");\
-  asm ("pxor xmm"tostr(a1)", [TEMP_MUL2+4*16]");\
-  asm ("pxor xmm"tostr(b2)", xmm"tostr(a1)"");\
-  asm ("pxor xmm"tostr(b3)", xmm"tostr(a1)"");\
-  \
-  /* 8 */\
-  asm ("pxor xmm"tostr(a5)", [TEMP_MUL1+5*16]");\
-  asm ("pxor xmm"tostr(a5)", [TEMP_MUL2+0*16]");\
-  asm ("pxor xmm"tostr(b6)", xmm"tostr(a5)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a5)"");\
-  \
-  /* 9 */\
-  asm ("movaps xmm"tostr(a3)", [TEMP_MUL1+2*16]");\
-  asm ("pxor   xmm"tostr(a3)", [TEMP_MUL2+5*16]");\
-  asm ("pxor   xmm"tostr(b0)", xmm"tostr(a3)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* 10 */\
-  asm ("movaps xmm"tostr(a1)", [TEMP_MUL1+6*16]");\
-  asm ("pxor   xmm"tostr(a1)", [TEMP_MUL2+1*16]");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a1)"");\
-  asm ("pxor   xmm"tostr(b4)", xmm"tostr(a1)"");\
-  \
-  /* 11 */\
-  asm ("movaps xmm"tostr(a5)", [TEMP_MUL1+3*16]");\
-  asm ("pxor   xmm"tostr(a5)", [TEMP_MUL2+6*16]");\
-  asm ("pxor   xmm"tostr(b1)", xmm"tostr(a5)"");\
-  asm ("pxor   xmm"tostr(b6)", xmm"tostr(a5)"");\
-  \
-  /* 12 */\
-  asm ("movaps xmm"tostr(a3)", [TEMP_MUL1+7*16]");\
-  asm ("pxor   xmm"tostr(a3)", [TEMP_MUL2+2*16]");\
-  asm ("pxor   xmm"tostr(b2)", xmm"tostr(a3)"");\
-  asm ("pxor   xmm"tostr(b5)", xmm"tostr(a3)"");\
-  \
-  /* 13 */\
-  asm ("pxor xmm"tostr(b0)", [TEMP_MUL4]");\
-  asm ("pxor xmm"tostr(b0)", xmm"tostr(a4)"");\
-  asm ("pxor xmm"tostr(b1)", xmm"tostr(a4)"");\
-  asm ("pxor xmm"tostr(b3)", xmm"tostr(a6)"");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a0)"");\
-  asm ("pxor xmm"tostr(b4)", xmm"tostr(a7)"");\
-  asm ("pxor xmm"tostr(b5)", xmm"tostr(a0)"");\
-  asm ("pxor xmm"tostr(b7)", xmm"tostr(a2)"");\
-}/**/
-
-//#if (LENGTH <= 256)
-
-#define SET_CONSTANTS(){\
-  SET_SHARED_CONSTANTS();\
-  ((u64*)SUBSH_MASK)[ 0] = 0x0706050403020100ULL;\
-  ((u64*)SUBSH_MASK)[ 1] = 0x080f0e0d0c0b0a09ULL;\
-  ((u64*)SUBSH_MASK)[ 2] = 0x0007060504030201ULL;\
-  ((u64*)SUBSH_MASK)[ 3] = 0x0a09080f0e0d0c0bULL;\
-  ((u64*)SUBSH_MASK)[ 4] = 0x0100070605040302ULL;\
-  ((u64*)SUBSH_MASK)[ 5] = 0x0c0b0a09080f0e0dULL;\
-  ((u64*)SUBSH_MASK)[ 6] = 0x0201000706050403ULL;\
-  ((u64*)SUBSH_MASK)[ 7] = 0x0e0d0c0b0a09080fULL;\
-  ((u64*)SUBSH_MASK)[ 8] = 0x0302010007060504ULL;\
-  ((u64*)SUBSH_MASK)[ 9] = 0x0f0e0d0c0b0a0908ULL;\
-  ((u64*)SUBSH_MASK)[10] = 0x0403020100070605ULL;\
-  ((u64*)SUBSH_MASK)[11] = 0x09080f0e0d0c0b0aULL;\
-  ((u64*)SUBSH_MASK)[12] = 0x0504030201000706ULL;\
-  ((u64*)SUBSH_MASK)[13] = 0x0b0a09080f0e0d0cULL;\
-  ((u64*)SUBSH_MASK)[14] = 0x0605040302010007ULL;\
-  ((u64*)SUBSH_MASK)[15] = 0x0d0c0b0a09080f0eULL;\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ((u64*)ROUND_CONST_L0)[i*2+1] = 0xffffffffffffffffULL;\
-    ((u64*)ROUND_CONST_L0)[i*2+0] = (i * 0x0101010101010101ULL)  ^ 0x7060504030201000ULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+1] = (i * 0x0101010101010101ULL)  ^ 0x8f9fafbfcfdfefffULL;\
-    ((u64*)ROUND_CONST_L7)[i*2+0] = 0x0000000000000000ULL;\
-  }\
-  ((u64*)ROUND_CONST_Lx)[1] = 0xffffffffffffffffULL;\
-  ((u64*)ROUND_CONST_Lx)[0] = 0x0000000000000000ULL;\
-}/**/
-
-#define Push_All_Regs(){\
-/*  not using any...
-    asm("push rax");\
-    asm("push rbx");\
-    asm("push rcx");*/\
-}/**/
-
-#define Pop_All_Regs(){\
-/*  not using any...
-    asm("pop rcx");\
-    asm("pop rbx");\
-    asm("pop rax");*/\
-}/**/
-
-
-/* vperm:
- * transformation before rounds with ipt
- * first round add transformed constant
- * middle rounds: add constant XOR 0x15...15
- * last round: additionally add 0x15...15 after MB
- * transformation after rounds with opt
- */
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant + ShiftBytes (interleaved) */\
-  asm ("movaps xmm"tostr(b1)", [ROUND_CONST_Lx]");\
-  asm ("pxor   xmm"tostr(a0)", [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("pxor   xmm"tostr(a1)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a2)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a3)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a0)", [SUBSH_MASK+0*16]");\
-  asm ("pshufb xmm"tostr(a1)", [SUBSH_MASK+1*16]");\
-  asm ("pxor   xmm"tostr(a4)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a2)", [SUBSH_MASK+2*16]");\
-  asm ("pshufb xmm"tostr(a3)", [SUBSH_MASK+3*16]");\
-  asm ("pxor   xmm"tostr(a5)", xmm"tostr(b1)"");\
-  asm ("pxor   xmm"tostr(a6)", xmm"tostr(b1)"");\
-  asm ("pshufb xmm"tostr(a4)", [SUBSH_MASK+4*16]");\
-  asm ("pshufb xmm"tostr(a5)", [SUBSH_MASK+5*16]");\
-  asm ("pxor   xmm"tostr(a7)", [ROUND_CONST_L7+"tostr(i)"*16]");\
-  asm ("pshufb xmm"tostr(a6)", [SUBSH_MASK+6*16]");\
-  asm ("pshufb xmm"tostr(a7)", [SUBSH_MASK+7*16]");\
-  /* SubBytes + Multiplication by 2 and 4 */\
-  VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7);\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}/**/
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  VPERM_Add_Constant(8, 9, 10, 11, 12, 13, 14, 15, ALL_15, 0);\
-  ROUND(0, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(2, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(3, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(4, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(5, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(6, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  ROUND(8, 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7);\
-  ROUND(9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);\
-  VPERM_Add_Constant(8, 9, 10, 11, 12, 13, 14, 15, ALL_15, 0);\
-}
-
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  asm ("movaps xmm"tostr(t0)", [TRANSP_MASK]");\
-\
-  asm ("pshufb xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("pshufb xmm"tostr(i3)", xmm"tostr(t0)"");\
-\
-  asm ("movdqa xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(t0)", xmm"tostr(i2)"");\
-\
-  asm ("punpcklwd xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhwd xmm"tostr(o1)", xmm"tostr(i1)"");\
-  asm ("punpcklwd xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhwd xmm"tostr(t0)", xmm"tostr(i3)"");\
-\
-  asm ("pshufd xmm"tostr(i0)", xmm"tostr(i0)", 216");\
-  asm ("pshufd xmm"tostr(o1)", xmm"tostr(o1)", 216");\
-  asm ("pshufd xmm"tostr(i2)", xmm"tostr(i2)", 216");\
-  asm ("pshufd xmm"tostr(t0)", xmm"tostr(t0)", 216");\
-\
-  asm ("movdqa xmm"tostr(o2)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(o3)", xmm"tostr(o1)"");\
-\
-  asm ("punpckldq xmm"tostr(i0)", xmm"tostr(i2)"");\
-  asm ("punpckldq xmm"tostr(o1)", xmm"tostr(t0)"");\
-  asm ("punpckhdq xmm"tostr(o2)", xmm"tostr(i2)"");\
-  asm ("punpckhdq xmm"tostr(o3)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i0)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i4)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i4)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o4)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o5)", xmm"tostr(i2)"");\
-  asm ("movdqa     xmm"tostr(o6)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o4)", xmm"tostr(i6)"");\
-  asm ("punpckhqdq xmm"tostr(o5)", xmm"tostr(i6)"");\
-  asm ("movdqa     xmm"tostr(o7)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(o6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o7)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  asm ("movdqa     xmm"tostr(o0)", xmm"tostr(i0)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpckhqdq xmm"tostr(o0)", xmm"tostr(i1)"");\
-  asm ("movdqa     xmm"tostr(o1)", xmm"tostr(i2)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpckhqdq xmm"tostr(o1)", xmm"tostr(i3)"");\
-  asm ("movdqa     xmm"tostr(o2)", xmm"tostr(i4)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpckhqdq xmm"tostr(o2)", xmm"tostr(i5)"");\
-  asm ("movdqa     xmm"tostr(o3)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-  asm ("punpckhqdq xmm"tostr(o3)", xmm"tostr(i7)"");\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  asm ("pxor xmm"tostr(t0)", xmm"tostr(t0)"");\
-  asm ("movdqa xmm"tostr(i1)", xmm"tostr(i0)"");\
-  asm ("movdqa xmm"tostr(i3)", xmm"tostr(i2)"");\
-  asm ("movdqa xmm"tostr(i5)", xmm"tostr(i4)"");\
-  asm ("movdqa xmm"tostr(i7)", xmm"tostr(i6)"");\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i1)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i3)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i5)", xmm"tostr(t0)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(t0)"");\
-  asm ("punpckhqdq xmm"tostr(i7)", xmm"tostr(t0)"");\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  asm ("punpcklqdq xmm"tostr(i0)", xmm"tostr(i1)"");\
-  asm ("punpcklqdq xmm"tostr(i2)", xmm"tostr(i3)"");\
-  asm ("punpcklqdq xmm"tostr(i4)", xmm"tostr(i5)"");\
-  asm ("punpcklqdq xmm"tostr(i6)", xmm"tostr(i7)"");\
-}/**/
-
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst_CNT2(i, j){\
-  asm ("movaps xmm0, [ROUND_CONST_L0+"tostr(i)"*16]");\
-  asm ("movaps xmm1, [ROUND_CONST_L7+"tostr(i)"*16]");\
-  asm ("movaps xmm2, [ROUND_CONST_L0+"tostr(j)"*16]");\
-  asm ("movaps xmm3, [ROUND_CONST_L7+"tostr(j)"*16]");\
-  VPERM_Transform_State(0, 1, 2, 3, VPERM_IPT, 4, 5, 6, 7, 8, 9, 10);\
-  asm ("pxor xmm0, [ALL_15]");\
-  asm ("pxor xmm1, [ALL_15]");\
-  asm ("pxor xmm2, [ALL_15]");\
-  asm ("pxor xmm3, [ALL_15]");\
-  asm ("movaps [ROUND_CONST_L0+"tostr(i)"*16], xmm0");\
-  asm ("movaps [ROUND_CONST_L7+"tostr(i)"*16], xmm1");\
-  asm ("movaps [ROUND_CONST_L0+"tostr(j)"*16], xmm2");\
-  asm ("movaps [ROUND_CONST_L7+"tostr(j)"*16], xmm3");\
-}/**/
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst(){\
-  asm ("movaps xmm0, [ROUND_CONST_Lx]");\
-  VPERM_Transform(0, 1, VPERM_IPT, 4, 5, 6, 7, 8, 9, 10);\
-  asm ("pxor xmm0, [ALL_15]");\
-  asm ("movaps [ROUND_CONST_Lx], xmm0");\
-  VPERM_Transform_RoundConst_CNT2(0, 1);\
-  VPERM_Transform_RoundConst_CNT2(2, 3);\
-  VPERM_Transform_RoundConst_CNT2(4, 5);\
-  VPERM_Transform_RoundConst_CNT2(6, 7);\
-  VPERM_Transform_RoundConst_CNT2(8, 9);\
-}/**/
-
-void INIT256(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  asm volatile ("emms");
-
-  /* transform round constants into VPERM mode */
-  VPERM_Transform_RoundConst();
-
-  /* load IV into registers xmm12 - xmm15 */
-  asm ("movaps xmm12, [rdi+0*16]");
-  asm ("movaps xmm13, [rdi+1*16]");
-  asm ("movaps xmm14, [rdi+2*16]");
-  asm ("movaps xmm15, [rdi+3*16]");
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  VPERM_Transform_State(12, 13, 14, 15, VPERM_IPT, 1, 2, 3, 4, 5, 6, 7);
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* store transposed IV */
-  asm ("movaps [rdi+0*16], xmm12");
-  asm ("movaps [rdi+1*16], xmm2");
-  asm ("movaps [rdi+2*16], xmm6");
-  asm ("movaps [rdi+3*16], xmm7");
-
-  asm volatile ("emms");
-  asm (".att_syntax noprefix");
-}
-
-void TF512(u64* h, u64* m)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-  /* message M in rsi            */
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load message into registers xmm12 - xmm15 (Q = message) */
-  asm ("movaps xmm12, [rsi+0*16]");
-  asm ("movaps xmm13, [rsi+1*16]");
-  asm ("movaps xmm14, [rsi+2*16]");
-  asm ("movaps xmm15, [rsi+3*16]");
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  VPERM_Transform_State(12, 13, 14, 15, VPERM_IPT, 1, 2, 3, 4, 5, 6, 7);
-  Matrix_Transpose_A(12, 13, 14, 15, 2, 6, 7, 0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  asm ("movaps xmm8, [rdi+0*16]");
-  asm ("movaps xmm0, [rdi+1*16]");
-  asm ("movaps xmm4, [rdi+2*16]");
-  asm ("movaps xmm5, [rdi+3*16]");
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  asm ("pxor xmm8, xmm12");
-  asm ("pxor xmm0, xmm2");
-  asm ("pxor xmm4, xmm6");
-  asm ("pxor xmm5, xmm7");
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(8, 0, 4, 5, 12, 2, 6, 7, 9, 10, 11, 12, 13, 14, 15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  asm ("pxor xmm0, xmm8");
-  asm ("pxor xmm1, xmm10");
-  asm ("pxor xmm2, xmm12");
-  asm ("pxor xmm3, xmm14");
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  asm ("pxor xmm0, [rdi+0*16]");
-  asm ("pxor xmm1, [rdi+1*16]");
-  asm ("pxor xmm2, [rdi+2*16]");
-  asm ("pxor xmm3, [rdi+3*16]");
-
-  /* store CV */
-  asm ("movaps [rdi+0*16], xmm0");
-  asm ("movaps [rdi+1*16], xmm1");
-  asm ("movaps [rdi+2*16], xmm2");
-  asm ("movaps [rdi+3*16], xmm3");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-
-  return;
-}
-
-void OF512(u64* h)
-{
-  /* __cdecl calling convention: */
-  /* chaining value CV in rdi    */
-
-  asm (".intel_syntax noprefix");
-  Push_All_Regs();
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  asm ("movaps xmm8,  [rdi+0*16]");
-  asm ("movaps xmm10, [rdi+1*16]");
-  asm ("movaps xmm12, [rdi+2*16]");
-  asm ("movaps xmm14, [rdi+3*16]");
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(8, 9, 10, 11, 12, 13, 14, 15, 0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(8, 9, 10, 11, 12, 13, 14, 15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  asm ("pxor xmm8,  [rdi+0*16]");
-  asm ("pxor xmm10, [rdi+1*16]");
-  asm ("pxor xmm12, [rdi+2*16]");
-  asm ("pxor xmm14, [rdi+3*16]");
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(8, 10, 12, 14, 4, 9, 11, 0);
-  VPERM_Transform(9, 11, VPERM_OPT, 0, 1, 2, 3, 5, 6, 7);
-
-  /* we only need to return the truncated half of the state */
-  asm ("movaps [rdi+2*16], xmm9");
-  asm ("movaps [rdi+3*16], xmm11");
-
-  Pop_All_Regs();
-  asm (".att_syntax noprefix");
-
-  return;
-}
-
-
--- a/algo/groestl/aes_ni/groestl256-intr-aes.h
+++ b/algo/groestl/aes_ni/groestl256-intr-aes.h
@@ -11,17 +11,44 @@
 #include <wmmintrin.h>
 #include "hash-groestl256.h"

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

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

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

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

-#define SET_CONSTANTS(){\
-   ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}while(0); \
-
 /* one round
 * i = round number
 * a0-a7 = input rows
@@ -179,34 +185,34 @@ __m128i ALL_FF;
 */
 #define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
  /* AddRoundConstant */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a4 = _mm_xor_si128(a4, b1);\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
+  b1 = m128_const_64( 0xffffffffffffffff, 0 ); \
+  a0 = _mm_xor_si128( a0, casti_m128i( round_const_l0, i ) ); \
+  a1 = _mm_xor_si128( a1, b1 ); \
+  a2 = _mm_xor_si128( a2, b1 ); \
+  a3 = _mm_xor_si128( a3, b1 ); \
+  a4 = _mm_xor_si128( a4, b1 ); \
+  a5 = _mm_xor_si128( a5, b1 ); \
+  a6 = _mm_xor_si128( a6, b1 ); \
+  a7 = _mm_xor_si128( a7, casti_m128i( round_const_l7, i ) ); \
  \
  /* ShiftBytes + SubBytes (interleaved) */\
  b0 = _mm_xor_si128(b0,  b0);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a0 = _mm_aesenclast_si128(a0, b0);\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a1 = _mm_aesenclast_si128(a1, b0);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a2 = _mm_aesenclast_si128(a2, b0);\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a3 = _mm_aesenclast_si128(a3, b0);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a4 = _mm_aesenclast_si128(a4, b0);\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a5 = _mm_aesenclast_si128(a5, b0);\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a6 = _mm_aesenclast_si128(a6, b0);\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  a7 = _mm_aesenclast_si128(a7, b0);\
+  a0 = _mm_shuffle_epi8( a0, SUBSH_MASK0 ); \
+  a0 = _mm_aesenclast_si128( a0, b0 );\
+  a1 = _mm_shuffle_epi8( a1, SUBSH_MASK1 ); \
+  a1 = _mm_aesenclast_si128( a1, b0 );\
+  a2 = _mm_shuffle_epi8( a2, SUBSH_MASK2 ); \
+  a2 = _mm_aesenclast_si128( a2, b0 );\
+  a3 = _mm_shuffle_epi8( a3, SUBSH_MASK3 ); \
+  a3 = _mm_aesenclast_si128( a3, b0 );\
+  a4 = _mm_shuffle_epi8( a4, SUBSH_MASK4 ); \
+  a4 = _mm_aesenclast_si128( a4, b0 );\
+  a5 = _mm_shuffle_epi8( a5, SUBSH_MASK5 ); \
+  a5 = _mm_aesenclast_si128( a5, b0 );\
+  a6 = _mm_shuffle_epi8( a6, SUBSH_MASK6 ); \
+  a6 = _mm_aesenclast_si128( a6, b0 );\
+  a7 = _mm_shuffle_epi8( a7, SUBSH_MASK7 ); \
+  a7 = _mm_aesenclast_si128( a7, b0 );\
  \
  /* MixBytes */\
  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
@@ -234,8 +240,9 @@ __m128i ALL_FF;
 * outputs: i0, o1-o3
 * clobbers: t0
 */
+
 #define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
+  t0 = TRANSP_MASK; \
  \
  i0 = _mm_shuffle_epi8(i0, t0);\
  i1 = _mm_shuffle_epi8(i1, t0);\
--- a/algo/groestl/aes_ni/groestl256-intr-avx.h
+++ b/algo/groestl/aes_ni/groestl256-intr-avx.h
@@ -1,482 +0,0 @@
-/* groestl-intr-avx.h     Aug 2011
- *
- * Groestl implementation with intrinsics using ssse3, sse4.1, aes and avx
- * instructions.
- * Author: Günther A. Roland, Martin Schläffer, Krystian Matusiewicz
- *
- * This code is placed in the public domain
- */
-
-#include <smmintrin.h>
-#include <wmmintrin.h>
-#include <immintrin.h>
-#include "hash-groestl256.h"
-
-/* global constants  */
-__m128i ROUND_CONST_Lx;
-__m128i ROUND_CONST_L0[ROUNDS512];
-__m128i ROUND_CONST_L7[ROUNDS512];
-__m128i ROUND_CONST_P[ROUNDS1024];
-__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_FF;
-//#if LENGTH <= 256
-__m128i ALL_1B;
-//#else
-//__m256d ALL_1B;
-//#endif
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define insert_m128i_in_m256d(ymm, xmm, pos) (_mm256_castsi256_pd(_mm256_insertf128_si256(_mm256_castpd_si256(ymm), xmm, pos)))
-#define extract_m128i_from_m256d(ymm, pos) (_mm256_extractf128_si256(_mm256_castpd_si256(ymm), pos))
-
-#define SET_CONSTANTS(){\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  ALL_FF = _mm_set_epi32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff);\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  SUBSH_MASK[0] = _mm_set_epi32(0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}while(0);
-
-/* xmm[i] will be multiplied by 2
- * xmm[j] will be lost
- * xmm[k] has to be all 0x1b
- * xmm[z] has to be zero */
-#define VMUL2(i, j, k, z){\
-  j = _mm_cmpgt_epi8(z, i);\
-  i = _mm_add_epi8(i, i);\
-  j = _mm_and_si128(j, k);\
-  i = _mm_xor_si128(i, j);\
-}/**/
-
-/* Yet another implementation of MixBytes.
-   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
-   Input: a0, ..., a7
-   Output: b0, ..., b7 = MixBytes(a0,...,a7).
-   but we use the relations:
-   t_i = a_i + a_{i+3}
-   x_i = t_i + t_{i+3}
-   y_i = t_i + t+{i+2} + a_{i+6}
-   z_i = 2*x_i
-   w_i = z_i + y_{i+4}
-   v_i = 2*w_i
-   b_i = v_{i+3} + y_{i+4}
-   We keep building b_i in registers xmm8..xmm15 by first building y_{i+4} there
-   and then adding v_i computed in the meantime in registers xmm0..xmm7.
-   We almost fit into 16 registers, need only 3 spills to memory.
-   This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
-   K. Matusiewicz, 2011/05/29 */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* xmm"tostr(8..xmm"tostr(15 = a2 a3... a0 a1 */\
-  b0 = a2;\
-  b1 = a3;\
-  b2 = a4;\
-  b3 = a5;\
-  b4 = a6;\
-  b5 = a7;\
-  b6 = a0;\
-  b7 = a1;\
-  \
-  /* t_i = a_i + a_{i+1} */\
-  a0 = _mm_xor_si128(a0, a1);\
-  a1 = _mm_xor_si128(a1, a2);\
-  a2 = _mm_xor_si128(a2, a3);\
-  a3 = _mm_xor_si128(a3, a4);\
-  a4 = _mm_xor_si128(a4, a5);\
-  a5 = _mm_xor_si128(a5, a6);\
-  a6 = _mm_xor_si128(a6, a7);\
-  a7 = _mm_xor_si128(a7, b6);\
-  \
-  /* build y4 y5 y6 ... in regs xmm8, xmm9, xmm10 by adding t_i*/\
-  b0 = _mm_xor_si128(b0, a4);\
-  b1 = _mm_xor_si128(b1, a5);\
-  b2 = _mm_xor_si128(b2, a6);\
-  b3 = _mm_xor_si128(b3, a7);\
-  b4 = _mm_xor_si128(b4, a0);\
-  b5 = _mm_xor_si128(b5, a1);\
-  b6 = _mm_xor_si128(b6, a2);\
-  b7 = _mm_xor_si128(b7, a3);\
-  \
-  b0 = _mm_xor_si128(b0, a6);\
-  b1 = _mm_xor_si128(b1, a7);\
-  b2 = _mm_xor_si128(b2, a0);\
-  b3 = _mm_xor_si128(b3, a1);\
-  b4 = _mm_xor_si128(b4, a2);\
-  b5 = _mm_xor_si128(b5, a3);\
-  b6 = _mm_xor_si128(b6, a4);\
-  b7 = _mm_xor_si128(b7, a5);\
-  \
-  /* spill values y_4, y_5 to memory */\
-  TEMP0 = b0;\
-  TEMP1 = b1;\
-  TEMP2 = b2;\
-  \
-  /* save values t0, t1, t2 to xmm8, xmm9 and memory */\
-  b0 = a0;\
-  b1 = a1;\
-  TEMP3 = a2;\
-  \
-  /* compute x_i = t_i + t_{i+3} */\
-  a0 = _mm_xor_si128(a0, a3);\
-  a1 = _mm_xor_si128(a1, a4);\
-  a2 = _mm_xor_si128(a2, a5);\
-  a3 = _mm_xor_si128(a3, a6);\
-  a4 = _mm_xor_si128(a4, a7);\
-  a5 = _mm_xor_si128(a5, b0);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, TEMP3);\
-  \
-  /*compute z_i : double x_i using temp xmm8 and 1B xmm9 */\
-  b1 = ALL_1B;\
-  b2 = _mm_xor_si128(b2, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a0, b0, b1, b2);\
-  \
-  /* compute w_i :  add y_{i+4} */\
-  a0 = _mm_xor_si128(a0, TEMP0);\
-  a1 = _mm_xor_si128(a1, TEMP1);\
-  a2 = _mm_xor_si128(a2, TEMP2);\
-  a3 = _mm_xor_si128(a3, b3);\
-  a4 = _mm_xor_si128(a4, b4);\
-  a5 = _mm_xor_si128(a5, b5);\
-  a6 = _mm_xor_si128(a6, b6);\
-  a7 = _mm_xor_si128(a7, b7);\
-  \
-  /*compute v_i: double w_i */\
-  VMUL2(a0, b0, b1, b2);\
-  VMUL2(a1, b0, b1, b2);\
-  VMUL2(a2, b0, b1, b2);\
-  VMUL2(a3, b0, b1, b2);\
-  VMUL2(a4, b0, b1, b2);\
-  VMUL2(a5, b0, b1, b2);\
-  VMUL2(a6, b0, b1, b2);\
-  VMUL2(a7, b0, b1, b2);\
-  \
-  /* add to y_4 y_5 .. v3, v4, ... */\
-  b0 = _mm_xor_si128(a3, TEMP0);\
-  b1 = _mm_xor_si128(a4, TEMP1);\
-  b2 = _mm_xor_si128(a5, TEMP2);\
-  b3 = _mm_xor_si128(b3, a6);\
-  b4 = _mm_xor_si128(b4, a7);\
-  b5 = _mm_xor_si128(b5, a0);\
-  b6 = _mm_xor_si128(b6, a1);\
-  b7 = _mm_xor_si128(b7, a2);\
-}/*MixBytes*/
-
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* Add Round Constant */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a4 = _mm_xor_si128(a4, b1);\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
-  \
-  /* ShiftBytes + SubBytes (interleaved) */\
-  b0 = _mm_xor_si128(b0,  b0);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a0 = _mm_aesenclast_si128(a0, b0);\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a1 = _mm_aesenclast_si128(a1, b0);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a2 = _mm_aesenclast_si128(a2, b0);\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a3 = _mm_aesenclast_si128(a3, b0);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a4 = _mm_aesenclast_si128(a4, b0);\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a5 = _mm_aesenclast_si128(a5, b0);\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a6 = _mm_aesenclast_si128(a6, b0);\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  a7 = _mm_aesenclast_si128(a7, b0);\
-  \
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  ROUND(0, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(1, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(2, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(3, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(4, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(5, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(6, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(7, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(8, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(9, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-}
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
-  \
-  i0 = _mm_shuffle_epi8(i0, t0);\
-  i1 = _mm_shuffle_epi8(i1, t0);\
-  i2 = _mm_shuffle_epi8(i2, t0);\
-  i3 = _mm_shuffle_epi8(i3, t0);\
-  \
-  o1 = _mm_unpackhi_epi16(i0, i1);\
-  i0 = _mm_unpacklo_epi16(i0, i1);\
-  t0 = _mm_unpackhi_epi16(i2, i3);\
-  i2 = _mm_unpacklo_epi16(i2, i3);\
-  \
-  i0 = _mm_shuffle_epi32(i0, 216);\
-  o1 = _mm_shuffle_epi32(o1, 216);\
-  i2 = _mm_shuffle_epi32(i2, 216);\
-  t0 = _mm_shuffle_epi32(t0, 216);\
-  \
-  o2 = _mm_unpackhi_epi32(i0, i2);\
-  o3 = _mm_unpackhi_epi32(o1, t0);\
-  i0 = _mm_unpacklo_epi32(i0, i2);\
-  o1 = _mm_unpacklo_epi32(o1, t0);\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  o1 = _mm_unpackhi_epi64(i0, i4);\
-  i0 = _mm_unpacklo_epi64(i0, i4);\
-  o2 = _mm_unpacklo_epi64(i1, i5);\
-  o3 = _mm_unpackhi_epi64(i1, i5);\
-  o4 = _mm_unpacklo_epi64(i2, i6);\
-  o5 = _mm_unpackhi_epi64(i2, i6);\
-  o6 = _mm_unpacklo_epi64(i3, i7);\
-  o7 = _mm_unpackhi_epi64(i3, i7);\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  o0 = _mm_unpackhi_epi64(i0, i1);\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  o1 = _mm_unpackhi_epi64(i2, i3);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  o2 = _mm_unpackhi_epi64(i4, i5);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  o3 = _mm_unpackhi_epi64(i6, i7);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  t0 = _mm_xor_si128(t0, t0);\
-  i1 = _mm_unpackhi_epi64(i0, t0);\
-  i0 = _mm_unpacklo_epi64(i0, t0);\
-  i3 = _mm_unpackhi_epi64(i2, t0);\
-  i2 = _mm_unpacklo_epi64(i2, t0);\
-  i5 = _mm_unpackhi_epi64(i4, t0);\
-  i4 = _mm_unpacklo_epi64(i4, t0);\
-  i7 = _mm_unpackhi_epi64(i6, t0);\
-  i6 = _mm_unpacklo_epi64(i6, t0);\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-
-void INIT256(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, /*xmm1,*/ xmm2, /*xmm3, xmm4, xmm5,*/ xmm6, xmm7;
-  static __m128i /*xmm8, xmm9, xmm10, xmm11,*/ xmm12, xmm13, xmm14, xmm15;
-
-  /* load IV into registers xmm12 - xmm15 */
-  xmm12 = chaining[0];
-  xmm13 = chaining[1];
-  xmm14 = chaining[2];
-  xmm15 = chaining[3];
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* store transposed IV */
-  chaining[0] = xmm12;
-  chaining[1] = xmm2;
-  chaining[2] = xmm6;
-  chaining[3] = xmm7;
-}
-
-void TF512(u64* h, u64* m)
-{
-  __m128i* const chaining = (__m128i*) h;
-  __m128i* const message = (__m128i*) m;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP0;
-  static __m128i TEMP1;
-  static __m128i TEMP2;
-  static __m128i TEMP3;
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  /* load message into registers xmm12 - xmm15 */
-  xmm12 = message[0];
-  xmm13 = message[1];
-  xmm14 = message[2];
-  xmm15 = message[3];
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* load previous chaining value and xor message to CV to get input of P */
-  /* we first put two rows (2x64 bit) of the CV into one 128-bit xmm register */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  xmm8 = _mm_xor_si128(xmm12, chaining[0]);
-  xmm0 = _mm_xor_si128(xmm2,  chaining[1]);
-  xmm4 = _mm_xor_si128(xmm6,  chaining[2]);
-  xmm5 = _mm_xor_si128(xmm7,  chaining[3]);
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(xmm8, xmm0, xmm4, xmm5, xmm12, xmm2, xmm6, xmm7, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, xmm8);
-  xmm1 = _mm_xor_si128(xmm1, xmm10);
-  xmm2 = _mm_xor_si128(xmm2, xmm12);
-  xmm3 = _mm_xor_si128(xmm3, xmm14);
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, chaining[0]);
-  xmm1 = _mm_xor_si128(xmm1, chaining[1]);
-  xmm2 = _mm_xor_si128(xmm2, chaining[2]);
-  xmm3 = _mm_xor_si128(xmm3, chaining[3]);
-
-  /* store CV */
-  chaining[0] = xmm0;
-  chaining[1] = xmm1;
-  chaining[2] = xmm2;
-  chaining[3] = xmm3;
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-  return;
-}
-
-void OF512(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP0;
-  static __m128i TEMP1;
-  static __m128i TEMP2;
-  static __m128i TEMP3;
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = chaining[0];
-  xmm10 = chaining[1];
-  xmm12 = chaining[2];
-  xmm14 = chaining[3];
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = _mm_xor_si128(xmm8,  (chaining[0]));
-  xmm10 = _mm_xor_si128(xmm10, (chaining[1]));
-  xmm12 = _mm_xor_si128(xmm12, (chaining[2]));
-  xmm14 = _mm_xor_si128(xmm14, (chaining[3]));
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(xmm8, xmm10, xmm12, xmm14, xmm4, xmm9, xmm11, xmm0);
-
-  /* we only need to return the truncated half of the state */
-  chaining[2] = xmm9;
-  chaining[3] = xmm11;
-}
-
-
--- a/algo/groestl/aes_ni/groestl256-intr-vperm.h
+++ b/algo/groestl/aes_ni/groestl256-intr-vperm.h
@@ -1,793 +0,0 @@
-/* groestl-intr-vperm.h     Aug 2011
- *
- * Groestl implementation with intrinsics using ssse3 instructions.
- * Author: Günther A. Roland, Martin Schläffer
- *
- * Based on the vperm and aes_ni implementations of the hash function Groestl
- * by Cagdas Calik <ccalik@metu.edu.tr> http://www.metu.edu.tr/~ccalik/
- * Institute of Applied Mathematics, Middle East Technical University, Turkey
- *
- * This code is placed in the public domain
- */
-
-#include <tmmintrin.h>
-#include "hash-groestl256.h"
-
-/* global constants  */
-__m128i ROUND_CONST_Lx;
-__m128i ROUND_CONST_L0[ROUNDS512];
-__m128i ROUND_CONST_L7[ROUNDS512];
-__m128i ROUND_CONST_P[ROUNDS1024];
-__m128i ROUND_CONST_Q[ROUNDS1024];
-__m128i TRANSP_MASK;
-__m128i SUBSH_MASK[8];
-__m128i ALL_0F;
-__m128i ALL_15;
-__m128i ALL_1B;
-__m128i ALL_63;
-__m128i ALL_FF;
-__m128i VPERM_IPT[2];
-__m128i VPERM_OPT[2];
-__m128i VPERM_INV[2];
-__m128i VPERM_SB1[2];
-__m128i VPERM_SB2[2];
-__m128i VPERM_SB4[2];
-__m128i VPERM_SBO[2];
-
-
-#define tos(a)    #a
-#define tostr(a)  tos(a)
-
-#define SET_SHARED_CONSTANTS(){\
-  TRANSP_MASK = _mm_set_epi32(0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800);\
-  ALL_1B = _mm_set_epi32(0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b, 0x1b1b1b1b);\
-  ALL_63 = _mm_set_epi32(0x63636363, 0x63636363, 0x63636363, 0x63636363);\
-  ALL_0F = _mm_set_epi32(0x0f0f0f0f, 0x0f0f0f0f, 0x0f0f0f0f, 0x0f0f0f0f);\
-  ALL_15 = _mm_set_epi32(0x15151515, 0x15151515, 0x15151515, 0x15151515);\
-  VPERM_IPT[0] = _mm_set_epi32(0xCD80B1FC, 0xB0FDCC81, 0x4C01307D, 0x317C4D00);\
-  VPERM_IPT[1] = _mm_set_epi32(0xCABAE090, 0x52227808, 0xC2B2E898, 0x5A2A7000);\
-  VPERM_OPT[0] = _mm_set_epi32(0xE10D5DB1, 0xB05C0CE0, 0x01EDBD51, 0x50BCEC00);\
-  VPERM_OPT[1] = _mm_set_epi32(0xF7974121, 0xDEBE6808, 0xFF9F4929, 0xD6B66000);\
-  VPERM_INV[0] = _mm_set_epi32(0x030D0E0C, 0x02050809, 0x01040A06, 0x0F0B0780);\
-  VPERM_INV[1] = _mm_set_epi32(0x04070309, 0x0A0B0C02, 0x0E05060F, 0x0D080180);\
-  VPERM_SB1[0] = _mm_set_epi32(0x3BF7CCC1, 0x0D2ED9EF, 0x3618D415, 0xFAE22300);\
-  VPERM_SB1[1] = _mm_set_epi32(0xA5DF7A6E, 0x142AF544, 0xB19BE18F, 0xCB503E00);\
-  VPERM_SB2[0] = _mm_set_epi32(0xC2A163C8, 0xAB82234A, 0x69EB8840, 0x0AE12900);\
-  VPERM_SB2[1] = _mm_set_epi32(0x5EB7E955, 0xBC982FCD, 0xE27A93C6, 0x0B712400);\
-  VPERM_SB4[0] = _mm_set_epi32(0xBA44FE79, 0x876D2914, 0x3D50AED7, 0xC393EA00);\
-  VPERM_SB4[1] = _mm_set_epi32(0xA876DE97, 0x49087E9F, 0xE1E937A0, 0x3FD64100);\
-}/**/
-
-/* VPERM
- * Transform w/o settings c*
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2){\
-  t0 = c0;\
-  t1 = c0;\
-  t0 = _mm_andnot_si128(t0, a0);\
-  t1 = _mm_andnot_si128(t1, a1);\
-  t0 = _mm_srli_epi32(t0, 4);\
-  t1 = _mm_srli_epi32(t1, 4);\
-  a0 = _mm_and_si128(a0, c0);\
-  a1 = _mm_and_si128(a1, c0);\
-  t2 = c2;\
-  t3 = c2;\
-  t2 = _mm_shuffle_epi8(t2, a0);\
-  t3 = _mm_shuffle_epi8(t3, a1);\
-  a0 = c1;\
-  a1 = c1;\
-  a0 = _mm_shuffle_epi8(a0, t0);\
-  a1 = _mm_shuffle_epi8(a1, t1);\
-  a0 = _mm_xor_si128(a0, t2);\
-  a1 = _mm_xor_si128(a1, t3);\
-}/**/
-
-#define VPERM_Transform_Set_Const(table, c0, c1, c2){\
-  c0 = ALL_0F;\
-  c1 = ((__m128i*) table )[0];\
-  c2 = ((__m128i*) table )[1];\
-}/**/
-
-/* VPERM
- * Transform
- * transforms 2 rows to/from "vperm mode"
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0, a1 = 2 rows
- * table = transformation table to use
- * t*, c* = clobbers
- * outputs:
- * a0, a1 = 2 rows transformed with table
- * */
-#define VPERM_Transform(a0, a1, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Transform State
- * inputs:
- * a0-a3 = state
- * table = transformation table to use
- * t* = clobbers
- * outputs:
- * a0-a3 = transformed state
- * */
-#define VPERM_Transform_State(a0, a1, a2, a3, table, t0, t1, t2, t3, c0, c1, c2){\
-  VPERM_Transform_Set_Const(table, c0, c1, c2);\
-  VPERM_Transform_No_Const(a0, a1, t0, t1, t2, t3, c0, c1, c2);\
-  VPERM_Transform_No_Const(a2, a3, t0, t1, t2, t3, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Add Constant to State
- * inputs:
- * a0-a7 = state
- * constant = constant to add
- * t0 = clobber
- * outputs:
- * a0-a7 = state + constant
- * */
-#define VPERM_Add_Constant(a0, a1, a2, a3, a4, a5, a6, a7, constant, t0){\
-  t0 = constant;\
-  a0 = _mm_xor_si128(a0,  t0);\
-  a1 = _mm_xor_si128(a1,  t0);\
-  a2 = _mm_xor_si128(a2,  t0);\
-  a3 = _mm_xor_si128(a3,  t0);\
-  a4 = _mm_xor_si128(a4,  t0);\
-  a5 = _mm_xor_si128(a5,  t0);\
-  a6 = _mm_xor_si128(a6,  t0);\
-  a7 = _mm_xor_si128(a7,  t0);\
-}/**/
-
-/* VPERM
- * Set Substitute Core Constants
- * */
-#define VPERM_Substitute_Core_Set_Const(c0, c1, c2){\
-  VPERM_Transform_Set_Const(VPERM_INV, c0, c1, c2);\
-}/**/
-
-/* VPERM
- * Substitute Core
- * first part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0 = 1 row
- * t*, c* = clobbers
- * outputs:
- * b0a, b0b = inputs for lookup step
- * */
-#define VPERM_Substitute_Core(a0, b0a, b0b, t0, t1, c0, c1, c2){\
-  t0 = c0;\
-  t0 = _mm_andnot_si128(t0, a0);\
-  t0 = _mm_srli_epi32(t0, 4);\
-  a0 = _mm_and_si128(a0,  c0);\
-  b0a = c1;\
-  b0a = _mm_shuffle_epi8(b0a, a0);\
-  a0 = _mm_xor_si128(a0,  t0);\
-  b0b = c2;\
-  b0b = _mm_shuffle_epi8(b0b, t0);\
-  b0b = _mm_xor_si128(b0b, b0a);\
-  t1 = c2;\
-  t1 = _mm_shuffle_epi8(t1,  a0);\
-  t1 = _mm_xor_si128(t1,  b0a);\
-  b0a = c2;\
-  b0a = _mm_shuffle_epi8(b0a, b0b);\
-  b0a = _mm_xor_si128(b0a, a0);\
-  b0b = c2;\
-  b0b = _mm_shuffle_epi8(b0b, t1);\
-  b0b = _mm_xor_si128(b0b, t0);\
-}/**/
-
-/* VPERM
- * Lookup
- * second part of sbox inverse computation
- * this function is derived from:
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0a, a0b = output of Substitution Core
- * table = lookup table to use (*1 / *2 / *4)
- * t0 = clobber
- * outputs:
- * b0 = output of sbox + multiplication
- * */
-#define VPERM_Lookup(a0a, a0b, table, b0, t0){\
-  b0 = ((__m128i*) table )[0];\
-  t0 = ((__m128i*) table )[1];\
-  b0 = _mm_shuffle_epi8(b0, a0b);\
-  t0 = _mm_shuffle_epi8(t0, a0a);\
-  b0 = _mm_xor_si128(b0, t0);\
-}/**/
-
-/* VPERM
- * SubBytes and *2 / *4
- * this function is derived from:
- *   Constant-time SSSE3 AES core implementation
- *   by Mike Hamburg
- * and
- *   vperm and aes_ni implementations of hash function Grostl
- *   by Cagdas CALIK
- * inputs:
- * a0-a7 = state
- * t*, c* = clobbers
- * outputs:
- * a0-a7 = state * 4
- * c2 = row0 * 2 -> b0
- * c1 = row7 * 2 -> b3
- * c0 = row7 * 1 -> b4
- * t2 = row4 * 1 -> b7
- * TEMP_MUL1 = row(i) * 1
- * TEMP_MUL2 = row(i) * 2
- *
- * call:VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7) */
-#define VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, t0, t1, t3, t4, c2, c1, c0, t2){\
-  /* set Constants */\
-  VPERM_Substitute_Core_Set_Const(c0, c1, c2);\
-  /* row 1 */\
-  VPERM_Substitute_Core(a1, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[1] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[1] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a1, t4);\
-  /* --- */\
-  /* row 2 */\
-  VPERM_Substitute_Core(a2, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[2] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[2] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a2, t4);\
-  /* --- */\
-  /* row 3 */\
-  VPERM_Substitute_Core(a3, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[3] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[3] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a3, t4);\
-  /* --- */\
-  /* row 5 */\
-  VPERM_Substitute_Core(a5, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[5] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[5] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a5, t4);\
-  /* --- */\
-  /* row 6 */\
-  VPERM_Substitute_Core(a6, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[6] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[6] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a6, t4);\
-  /* --- */\
-  /* row 7 */\
-  VPERM_Substitute_Core(a7, t0, t1, t3, t4, c0, c1, c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4);\
-  TEMP_MUL1[7] = t2;\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c1, t4); /*c1 -> b3*/\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a7, t4);\
-  /* --- */\
-  /* row 4 */\
-  VPERM_Substitute_Core(a4, t0, t1, t3, t4, c0, (VPERM_INV[0]), c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, t2, t4); /*t2 -> b7*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, t3, t4);\
-  TEMP_MUL2[4] = t3;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a4, t4);\
-  /* --- */\
-  /* row 0 */\
-  VPERM_Substitute_Core(a0, t0, t1, t3, t4, c0, (VPERM_INV[0]), c2);\
-  VPERM_Lookup(t0, t1, VPERM_SB1, c0, t4); /*c0 -> b4*/\
-  VPERM_Lookup(t0, t1, VPERM_SB2, c2, t4); /*c2 -> b0*/\
-  TEMP_MUL2[0] = c2;\
-  VPERM_Lookup(t0, t1, VPERM_SB4, a0, t4);\
-  /* --- */\
-}/**/
-
-
-/* Optimized MixBytes
- * inputs:
- * a0-a7 = (row0-row7) * 4
- * b0 = row0 * 2
- * b3 = row7 * 2
- * b4 = row7 * 1
- * b7 = row4 * 1
- * all *1 and *2 values must also be in TEMP_MUL1, TEMP_MUL2
- * output: b0-b7
- * */
-#define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* save one value */\
-  TEMP_MUL4 = a3;\
-  /* 1 */\
-  b1 = a0;\
-  b1 = _mm_xor_si128(b1, a5);\
-  b1 = _mm_xor_si128(b1, b4); /* -> helper! */\
-  b1 = _mm_xor_si128(b1, (TEMP_MUL2[3]));\
-  b2 = b1;\
-  \
-  /* 2 */\
-  b5 = a1;\
-  b5 = _mm_xor_si128(b5, a4);\
-  b5 = _mm_xor_si128(b5, b7); /* -> helper! */\
-  b5 = _mm_xor_si128(b5, b3); /* -> helper! */\
-  b6 = b5;\
-  \
-  /* 4 */\
-  b7 = _mm_xor_si128(b7, a6);\
-  /*b7 = _mm_xor_si128(b7, (TEMP_MUL1[4])); -> helper! */\
-  b7 = _mm_xor_si128(b7, (TEMP_MUL1[6]));\
-  b7 = _mm_xor_si128(b7, (TEMP_MUL2[1]));\
-  b7 = _mm_xor_si128(b7, b3); /* -> helper! */\
-  b2 = _mm_xor_si128(b2, b7);\
-  \
-  /* 3 */\
-  b0 = _mm_xor_si128(b0, a7);\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL1[5]));\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL1[7]));\
-  /*b0 = _mm_xor_si128(b0, (TEMP_MUL2[0])); -> helper! */\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL2[2]));\
-  b3 = b0;\
-  b1 = _mm_xor_si128(b1, b0);\
-  b0 = _mm_xor_si128(b0, b7); /* moved from 4 */\
-  \
-  /* 5 */\
-  b4 = _mm_xor_si128(b4, a2);\
-  /*b4 = _mm_xor_si128(b4, (TEMP_MUL1[0])); -> helper! */\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL1[2]));\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL2[3]));\
-  b4 = _mm_xor_si128(b4, (TEMP_MUL2[5]));\
-  b3 = _mm_xor_si128(b3, b4);\
-  b6 = _mm_xor_si128(b6, b4);\
-  \
-  /* 6 */\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL1[1]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL1[3]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[4]));\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[6]));\
-  b4 = _mm_xor_si128(b4, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  b7 = _mm_xor_si128(b7, a3);\
-  \
-  /* 7 */\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL1[1]));\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL2[4]));\
-  b2 = _mm_xor_si128(b2, a1);\
-  b3 = _mm_xor_si128(b3, a1);\
-  \
-  /* 8 */\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL1[5]));\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL2[0]));\
-  b6 = _mm_xor_si128(b6, a5);\
-  b7 = _mm_xor_si128(b7, a5);\
-  \
-  /* 9 */\
-  a3 = TEMP_MUL1[2];\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[5]));\
-  b0 = _mm_xor_si128(b0, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  \
-  /* 10 */\
-  a1 = TEMP_MUL1[6];\
-  a1 = _mm_xor_si128(a1, (TEMP_MUL2[1]));\
-  b1 = _mm_xor_si128(b1, a1);\
-  b4 = _mm_xor_si128(b4, a1);\
-  \
-  /* 11 */\
-  a5 = TEMP_MUL1[3];\
-  a5 = _mm_xor_si128(a5, (TEMP_MUL2[6]));\
-  b1 = _mm_xor_si128(b1, a5);\
-  b6 = _mm_xor_si128(b6, a5);\
-  \
-  /* 12 */\
-  a3 = TEMP_MUL1[7];\
-  a3 = _mm_xor_si128(a3, (TEMP_MUL2[2]));\
-  b2 = _mm_xor_si128(b2, a3);\
-  b5 = _mm_xor_si128(b5, a3);\
-  \
-  /* 13 */\
-  b0 = _mm_xor_si128(b0, (TEMP_MUL4));\
-  b0 = _mm_xor_si128(b0, a4);\
-  b1 = _mm_xor_si128(b1, a4);\
-  b3 = _mm_xor_si128(b3, a6);\
-  b4 = _mm_xor_si128(b4, a0);\
-  b4 = _mm_xor_si128(b4, a7);\
-  b5 = _mm_xor_si128(b5, a0);\
-  b7 = _mm_xor_si128(b7, a2);\
-}/**/
-
-#define SET_CONSTANTS(){\
-  SET_SHARED_CONSTANTS();\
-  SUBSH_MASK[0] = _mm_set_epi32(0x080f0e0d, 0x0c0b0a09, 0x07060504, 0x03020100);\
-  SUBSH_MASK[1] = _mm_set_epi32(0x0a09080f, 0x0e0d0c0b, 0x00070605, 0x04030201);\
-  SUBSH_MASK[2] = _mm_set_epi32(0x0c0b0a09, 0x080f0e0d, 0x01000706, 0x05040302);\
-  SUBSH_MASK[3] = _mm_set_epi32(0x0e0d0c0b, 0x0a09080f, 0x02010007, 0x06050403);\
-  SUBSH_MASK[4] = _mm_set_epi32(0x0f0e0d0c, 0x0b0a0908, 0x03020100, 0x07060504);\
-  SUBSH_MASK[5] = _mm_set_epi32(0x09080f0e, 0x0d0c0b0a, 0x04030201, 0x00070605);\
-  SUBSH_MASK[6] = _mm_set_epi32(0x0b0a0908, 0x0f0e0d0c, 0x05040302, 0x01000706);\
-  SUBSH_MASK[7] = _mm_set_epi32(0x0d0c0b0a, 0x09080f0e, 0x06050403, 0x02010007);\
-  for(i = 0; i < ROUNDS512; i++)\
-  {\
-    ROUND_CONST_L0[i] = _mm_set_epi32(0xffffffff, 0xffffffff, 0x70605040 ^ (i * 0x01010101), 0x30201000 ^ (i * 0x01010101));\
-    ROUND_CONST_L7[i] = _mm_set_epi32(0x8f9fafbf ^ (i * 0x01010101), 0xcfdfefff ^ (i * 0x01010101), 0x00000000, 0x00000000);\
-  }\
-  ROUND_CONST_Lx = _mm_set_epi32(0xffffffff, 0xffffffff, 0x00000000, 0x00000000);\
-}/**/
-
-/* vperm:
- * transformation before rounds with ipt
- * first round add transformed constant
- * middle rounds: add constant XOR 0x15...15
- * last round: additionally add 0x15...15 after MB
- * transformation after rounds with opt
- */
-/* one round
- * i = round number
- * a0-a7 = input rows
- * b0-b7 = output rows
- */
-#define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
-  /* AddRoundConstant + ShiftBytes (interleaved) */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm_xor_si128(a0, (ROUND_CONST_L0[i]));\
-  a1 = _mm_xor_si128(a1, b1);\
-  a2 = _mm_xor_si128(a2, b1);\
-  a3 = _mm_xor_si128(a3, b1);\
-  a0 = _mm_shuffle_epi8(a0, (SUBSH_MASK[0]));\
-  a1 = _mm_shuffle_epi8(a1, (SUBSH_MASK[1]));\
-  a4 = _mm_xor_si128(a4, b1);\
-  a2 = _mm_shuffle_epi8(a2, (SUBSH_MASK[2]));\
-  a3 = _mm_shuffle_epi8(a3, (SUBSH_MASK[3]));\
-  a5 = _mm_xor_si128(a5, b1);\
-  a6 = _mm_xor_si128(a6, b1);\
-  a4 = _mm_shuffle_epi8(a4, (SUBSH_MASK[4]));\
-  a5 = _mm_shuffle_epi8(a5, (SUBSH_MASK[5]));\
-  a7 = _mm_xor_si128(a7, (ROUND_CONST_L7[i]));\
-  a6 = _mm_shuffle_epi8(a6, (SUBSH_MASK[6]));\
-  a7 = _mm_shuffle_epi8(a7, (SUBSH_MASK[7]));\
-  /* SubBytes + Multiplication by 2 and 4 */\
-  VPERM_SUB_MULTIPLY(a0, a1, a2, a3, a4, a5, a6, a7, b1, b2, b5, b6, b0, b3, b4, b7);\
-  /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
-}/**/
-
-/* 10 rounds, P and Q in parallel */
-#define ROUNDS_P_Q(){\
-  VPERM_Add_Constant(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, ALL_15, xmm0);\
-  ROUND(0, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(1, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(2, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(3, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(4, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(5, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(6, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(7, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  ROUND(8, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
-  ROUND(9, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
-  VPERM_Add_Constant(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, ALL_15, xmm0);\
-}
-
-
-/* Matrix Transpose Step 1
- * input is a 512-bit state with two columns in one xmm
- * output is a 512-bit state with two rows in one xmm
- * inputs: i0-i3
- * outputs: i0, o1-o3
- * clobbers: t0
- */
-#define Matrix_Transpose_A(i0, i1, i2, i3, o1, o2, o3, t0){\
-  t0 = TRANSP_MASK;\
-\
-  i0 = _mm_shuffle_epi8(i0, t0);\
-  i1 = _mm_shuffle_epi8(i1, t0);\
-  i2 = _mm_shuffle_epi8(i2, t0);\
-  i3 = _mm_shuffle_epi8(i3, t0);\
-\
-  o1 = i0;\
-  t0 = i2;\
-\
-  i0 = _mm_unpacklo_epi16(i0, i1);\
-  o1 = _mm_unpackhi_epi16(o1, i1);\
-  i2 = _mm_unpacklo_epi16(i2, i3);\
-  t0 = _mm_unpackhi_epi16(t0, i3);\
-\
-  i0 = _mm_shuffle_epi32(i0, 216);\
-  o1 = _mm_shuffle_epi32(o1, 216);\
-  i2 = _mm_shuffle_epi32(i2, 216);\
-  t0 = _mm_shuffle_epi32(t0, 216);\
-\
-  o2 = i0;\
-  o3 = o1;\
-\
-  i0 = _mm_unpacklo_epi32(i0, i2);\
-  o1 = _mm_unpacklo_epi32(o1, t0);\
-  o2 = _mm_unpackhi_epi32(o2, i2);\
-  o3 = _mm_unpackhi_epi32(o3, t0);\
-}/**/
-
-/* Matrix Transpose Step 2
- * input are two 512-bit states with two rows in one xmm
- * output are two 512-bit states with one row of each state in one xmm
- * inputs: i0-i3 = P, i4-i7 = Q
- * outputs: (i0, o1-o7) = (P|Q)
- * possible reassignments: (output reg = input reg)
- * * i1 -> o3-7
- * * i2 -> o5-7
- * * i3 -> o7
- * * i4 -> o3-7
- * * i5 -> o6-7
- */
-#define Matrix_Transpose_B(i0, i1, i2, i3, i4, i5, i6, i7, o1, o2, o3, o4, o5, o6, o7){\
-  o1 = i0;\
-  o2 = i1;\
-  i0 = _mm_unpacklo_epi64(i0, i4);\
-  o1 = _mm_unpackhi_epi64(o1, i4);\
-  o3 = i1;\
-  o4 = i2;\
-  o2 = _mm_unpacklo_epi64(o2, i5);\
-  o3 = _mm_unpackhi_epi64(o3, i5);\
-  o5 = i2;\
-  o6 = i3;\
-  o4 = _mm_unpacklo_epi64(o4, i6);\
-  o5 = _mm_unpackhi_epi64(o5, i6);\
-  o7 = i3;\
-  o6 = _mm_unpacklo_epi64(o6, i7);\
-  o7 = _mm_unpackhi_epi64(o7, i7);\
-}/**/
-
-/* Matrix Transpose Inverse Step 2
- * input are two 512-bit states with one row of each state in one xmm
- * output are two 512-bit states with two rows in one xmm
- * inputs: i0-i7 = (P|Q)
- * outputs: (i0, i2, i4, i6) = P, (o0-o3) = Q
- */
-#define Matrix_Transpose_B_INV(i0, i1, i2, i3, i4, i5, i6, i7, o0, o1, o2, o3){\
-  o0 = i0;\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  o0 = _mm_unpackhi_epi64(o0, i1);\
-  o1 = i2;\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  o1 = _mm_unpackhi_epi64(o1, i3);\
-  o2 = i4;\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  o2 = _mm_unpackhi_epi64(o2, i5);\
-  o3 = i6;\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-  o3 = _mm_unpackhi_epi64(o3, i7);\
-}/**/
-
-/* Matrix Transpose Output Step 2
- * input is one 512-bit state with two rows in one xmm
- * output is one 512-bit state with one row in the low 64-bits of one xmm
- * inputs: i0,i2,i4,i6 = S
- * outputs: (i0-7) = (0|S)
- */
-#define Matrix_Transpose_O_B(i0, i1, i2, i3, i4, i5, i6, i7, t0){\
-  t0 = _mm_xor_si128(t0, t0);\
-  i1 = i0;\
-  i3 = i2;\
-  i5 = i4;\
-  i7 = i6;\
-  i0 = _mm_unpacklo_epi64(i0, t0);\
-  i1 = _mm_unpackhi_epi64(i1, t0);\
-  i2 = _mm_unpacklo_epi64(i2, t0);\
-  i3 = _mm_unpackhi_epi64(i3, t0);\
-  i4 = _mm_unpacklo_epi64(i4, t0);\
-  i5 = _mm_unpackhi_epi64(i5, t0);\
-  i6 = _mm_unpacklo_epi64(i6, t0);\
-  i7 = _mm_unpackhi_epi64(i7, t0);\
-}/**/
-
-/* Matrix Transpose Output Inverse Step 2
- * input is one 512-bit state with one row in the low 64-bits of one xmm
- * output is one 512-bit state with two rows in one xmm
- * inputs: i0-i7 = (0|S)
- * outputs: (i0, i2, i4, i6) = S
- */
-#define Matrix_Transpose_O_B_INV(i0, i1, i2, i3, i4, i5, i6, i7){\
-  i0 = _mm_unpacklo_epi64(i0, i1);\
-  i2 = _mm_unpacklo_epi64(i2, i3);\
-  i4 = _mm_unpacklo_epi64(i4, i5);\
-  i6 = _mm_unpacklo_epi64(i6, i7);\
-}/**/
-
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst_CNT2(i, j){\
-  xmm0 = ROUND_CONST_L0[i];\
-  xmm1 = ROUND_CONST_L7[i];\
-  xmm2 = ROUND_CONST_L0[j];\
-  xmm3 = ROUND_CONST_L7[j];\
-  VPERM_Transform_State(xmm0, xmm1, xmm2, xmm3, VPERM_IPT, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10);\
-  xmm0 = _mm_xor_si128(xmm0, (ALL_15));\
-  xmm1 = _mm_xor_si128(xmm1, (ALL_15));\
-  xmm2 = _mm_xor_si128(xmm2, (ALL_15));\
-  xmm3 = _mm_xor_si128(xmm3, (ALL_15));\
-  ROUND_CONST_L0[i] = xmm0;\
-  ROUND_CONST_L7[i] = xmm1;\
-  ROUND_CONST_L0[j] = xmm2;\
-  ROUND_CONST_L7[j] = xmm3;\
-}/**/
-
-/* transform round constants into VPERM mode */
-#define VPERM_Transform_RoundConst(){\
-  xmm0 = ROUND_CONST_Lx;\
-  VPERM_Transform(xmm0, xmm1, VPERM_IPT, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10);\
-  xmm0 = _mm_xor_si128(xmm0, (ALL_15));\
-  ROUND_CONST_Lx = xmm0;\
-  VPERM_Transform_RoundConst_CNT2(0, 1);\
-  VPERM_Transform_RoundConst_CNT2(2, 3);\
-  VPERM_Transform_RoundConst_CNT2(4, 5);\
-  VPERM_Transform_RoundConst_CNT2(6, 7);\
-  VPERM_Transform_RoundConst_CNT2(8, 9);\
-}/**/
-
-void INIT256(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, /*xmm11,*/ xmm12, xmm13, xmm14, xmm15;
-
-  /* transform round constants into VPERM mode */
-  VPERM_Transform_RoundConst();
-
-  /* load IV into registers xmm12 - xmm15 */
-  xmm12 = chaining[0];
-  xmm13 = chaining[1];
-  xmm14 = chaining[2];
-  xmm15 = chaining[3];
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  VPERM_Transform_State(xmm12, xmm13, xmm14, xmm15, VPERM_IPT, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* store transposed IV */
-  chaining[0] = xmm12;
-  chaining[1] = xmm2;
-  chaining[2] = xmm6;
-  chaining[3] = xmm7;
-}
-
-void TF512(u64* h, u64* m)
-{
-  __m128i* const chaining = (__m128i*) h;
-  __m128i* const message = (__m128i*) m;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP_MUL1[8];
-  static __m128i TEMP_MUL2[8];
-  static __m128i TEMP_MUL4;
-
-#ifdef IACA_TRACE
-  IACA_START;
-#endif
-
-  /* load message into registers xmm12 - xmm15 */
-  xmm12 = message[0];
-  xmm13 = message[1];
-  xmm14 = message[2];
-  xmm15 = message[3];
-
-  /* transform message M from column ordering into row ordering */
-  /* we first put two rows (64 bit) of the message into one 128-bit xmm register */
-  VPERM_Transform_State(xmm12, xmm13, xmm14, xmm15, VPERM_IPT, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* load previous chaining value */
-  /* we first put two rows (64 bit) of the CV into one 128-bit xmm register */
-  xmm8 = chaining[0];
-  xmm0 = chaining[1];
-  xmm4 = chaining[2];
-  xmm5 = chaining[3];
-
-  /* xor message to CV get input of P */
-  /* result: CV+M in xmm8, xmm0, xmm4, xmm5 */
-  xmm8 = _mm_xor_si128(xmm8, xmm12);
-  xmm0 = _mm_xor_si128(xmm0, xmm2);
-  xmm4 = _mm_xor_si128(xmm4, xmm6);
-  xmm5 = _mm_xor_si128(xmm5, xmm7);
-
-  /* there are now 2 rows of the Groestl state (P and Q) in each xmm register */
-  /* unpack to get 1 row of P (64 bit) and Q (64 bit) into one xmm register */
-  /* result: the 8 rows of P and Q in xmm8 - xmm12 */
-  Matrix_Transpose_B(xmm8, xmm0, xmm4, xmm5, xmm12, xmm2, xmm6, xmm7, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* compute the two permutations P and Q in parallel */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P or two rows of Q in one xmm register */
-  Matrix_Transpose_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3);
-
-  /* xor output of P and Q */
-  /* result: P(CV+M)+Q(M) in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, xmm8);
-  xmm1 = _mm_xor_si128(xmm1, xmm10);
-  xmm2 = _mm_xor_si128(xmm2, xmm12);
-  xmm3 = _mm_xor_si128(xmm3, xmm14);
-
-  /* xor CV (feed-forward) */
-  /* result: P(CV+M)+Q(M)+CV in xmm0...xmm3 */
-  xmm0 = _mm_xor_si128(xmm0, (chaining[0]));
-  xmm1 = _mm_xor_si128(xmm1, (chaining[1]));
-  xmm2 = _mm_xor_si128(xmm2, (chaining[2]));
-  xmm3 = _mm_xor_si128(xmm3, (chaining[3]));
-
-  /* store CV */
-  chaining[0] = xmm0;
-  chaining[1] = xmm1;
-  chaining[2] = xmm2;
-  chaining[3] = xmm3;
-
-#ifdef IACA_TRACE
-  IACA_END;
-#endif
-
-  return;
-}
-
-void OF512(u64* h)
-{
-  __m128i* const chaining = (__m128i*) h;
-  static __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
-  static __m128i xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15;
-  static __m128i TEMP_MUL1[8];
-  static __m128i TEMP_MUL2[8];
-  static __m128i TEMP_MUL4;
-
-  /* load CV into registers xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = chaining[0];
-  xmm10 = chaining[1];
-  xmm12 = chaining[2];
-  xmm14 = chaining[3];
-
-  /* there are now 2 rows of the CV in one xmm register */
-  /* unpack to get 1 row of P (64 bit) into one half of an xmm register */
-  /* result: the 8 input rows of P in xmm8 - xmm15 */
-  Matrix_Transpose_O_B(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0);
-
-  /* compute the permutation P */
-  /* result: the output of P(CV) in xmm8 - xmm15 */
-  ROUNDS_P_Q();
-
-  /* unpack again to get two rows of P in one xmm register */
-  /* result: P(CV) in xmm8, xmm10, xmm12, xmm14 */
-  Matrix_Transpose_O_B_INV(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);
-
-  /* xor CV to P output (feed-forward) */
-  /* result: P(CV)+CV in xmm8, xmm10, xmm12, xmm14 */
-  xmm8 = _mm_xor_si128(xmm8,  (chaining[0]));
-  xmm10 = _mm_xor_si128(xmm10, (chaining[1]));
-  xmm12 = _mm_xor_si128(xmm12, (chaining[2]));
-  xmm14 = _mm_xor_si128(xmm14, (chaining[3]));
-
-  /* transform state back from row ordering into column ordering */
-  /* result: final hash value in xmm9, xmm11 */
-  Matrix_Transpose_A(xmm8, xmm10, xmm12, xmm14, xmm4, xmm9, xmm11, xmm0);
-  VPERM_Transform(xmm9, xmm11, VPERM_OPT, xmm0, xmm1, xmm2, xmm3, xmm5, xmm6, xmm7);
-
-  /* we only need to return the truncated half of the state */
-  chaining[2] = xmm9;
-  chaining[3] = xmm11;
-
-  return;
-}//OF512()
-
-
-
--- a/algo/groestl/aes_ni/hash-groestl.c
+++ b/algo/groestl/aes_ni/hash-groestl.c
@@ -16,48 +16,13 @@

 #ifdef __AES__

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

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

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

  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return FAIL_GR;
@@ -70,8 +35,6 @@ HashReturn_gr init_groestl( hashState_groestl* ctx, int hashlen )

  // The only non-zero in the IV is len. It can be hard coded.
  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
-//  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
-//  INIT(ctx->chaining);

  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;
@@ -92,8 +55,6 @@ HashReturn_gr reinit_groestl( hashState_groestl* ctx )
     ctx->buffer[i]   = _mm_setzero_si128();
  }
  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
-//  ((u64*)ctx->chaining)[COLS-1] = U64BIG((u64)LENGTH);
-//  INIT(ctx->chaining);
  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;

@@ -109,7 +70,7 @@ HashReturn_gr reinit_groestl( hashState_groestl* ctx )
 // 5. Midstate will work at reduced impact than full hash, if total hash
 //    (midstate + tail) is less than 1 block.
 //    This, unfortunately, is the case with all current users.
-// 6. the morefull blocks the bigger the gain
+// 6. the more full blocks the bigger the gain

 // use only for midstate precalc
 HashReturn_gr update_groestl( hashState_groestl* ctx, const void* input,
@@ -143,12 +104,11 @@ HashReturn_gr update_groestl( hashState_groestl* ctx, const void* input,
 // deprecated do not use
 HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
 {
-   const int len = (int)ctx->databitlen / 128;  // bits to __m128i 
-   const int blocks = ctx->blk_count + 1;       // adjust for final block
-
-   const int rem_ptr = ctx->rem_ptr;      // end of data start of padding
-   const int hashlen_m128i = ctx->hashlen / 16;  // bytes to __m128i
-   const int hash_offset = SIZE512 - hashlen_m128i;  // where in buffer
+   const int len = (int)ctx->databitlen / 128; // bits to __m128i 
+   const uint64_t blocks = ctx->blk_count + 1; // adjust for final block
+   const int rem_ptr = ctx->rem_ptr;           // end of data start of padding
+   const int hashlen_m128i = ctx->hashlen / 16;     // bytes to __m128i
+   const int hash_offset = SIZE512 - hashlen_m128i; // where in buffer
   int i;

   // first pad byte = 0x80, last pad byte = block count
@@ -157,21 +117,18 @@ HashReturn_gr final_groestl( hashState_groestl* ctx, void* output )
   if ( rem_ptr == len - 1 )
   {
       // only 128 bits left in buffer, all padding at once
-       ctx->buffer[rem_ptr] = _mm_set_epi8( blocks,0,0,0, 0,0,0,0,
-                                                  0,0,0,0, 0,0,0,0x80 );
+      ctx->buffer[rem_ptr] = _mm_set_epi64x( blocks << 56, 0x80 );
   }
   else
   {
       // add first padding
-       ctx->buffer[rem_ptr] = _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                            0,0,0,0, 0,0,0,0x80 );
+       ctx->buffer[rem_ptr] = m128_const_64( 0, 0x80 );
       // add zero padding
       for ( i = rem_ptr + 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();

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

   // digest final padding block and do output transform
@@ -189,21 +146,20 @@ int groestl512_full( hashState_groestl* ctx, void* output,
                                const void* input, uint64_t databitlen )
 {

-  int i;
-
-  ctx->hashlen = 64;
-  SET_CONSTANTS();
-
-  for ( i = 0; i < SIZE512; i++ )
-  {
-     ctx->chaining[i] = _mm_setzero_si128();
-     ctx->buffer[i]   = _mm_setzero_si128();
-  }
-  ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
-  ctx->buf_ptr = 0;
-  ctx->rem_ptr = 0;
+   int i;
+   ctx->hashlen = 64;

+   for ( i = 0; i < SIZE512; i++ )
+   {
+      ctx->chaining[i] = _mm_setzero_si128();
+      ctx->buffer[i]   = _mm_setzero_si128();
+   }
+   ctx->chaining[ 6 ] = m128_const_64( 0x0200000000000000, 0 );
+   ctx->buf_ptr = 0;
+   ctx->rem_ptr = 0;

+   // --- update ---
+   
   const int len = (int)databitlen / 128;
   const int hashlen_m128i = ctx->hashlen / 16;   // bytes to __m128i
   const int hash_offset = SIZE512 - hashlen_m128i;
@@ -211,8 +167,6 @@ int groestl512_full( hashState_groestl* ctx, void* output,
   uint64_t blocks = len / SIZE512;
   __m128i* in = (__m128i*)input;

-   // --- update ---
-
   // digest any full blocks, process directly from input 
   for ( i = 0; i < blocks; i++ )
      TF1024( ctx->chaining, &in[ i * SIZE512 ] );
@@ -231,26 +185,22 @@ int groestl512_full( hashState_groestl* ctx, void* output,
   if ( i == len -1 )
   {
       // only 128 bits left in buffer, all padding at once
-       ctx->buffer[i] = _mm_set_epi8( blocks,0,0,0, 0,0,0,0,
-                                           0,0,0,0, 0,0,0,0x80 );
+      ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0x80 );
   }
   else
   {
       // add first padding
-       ctx->buffer[i] = _mm_set_epi8( 0,0,0,0, 0,0,0,0,
-                                      0,0,0,0, 0,0,0,0x80 );
+       ctx->buffer[i] = m128_const_64( 0, 0x80 );
       // add zero padding
       for ( i += 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();

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

   // digest final padding block and do output transform
   TF1024( ctx->chaining, ctx->buffer );
-
   OF1024( ctx->chaining );

   // store hash result in output 
@@ -268,7 +218,7 @@ HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
   const int hashlen_m128i = ctx->hashlen / 16;   // bytes to __m128i
   const int hash_offset = SIZE512 - hashlen_m128i;
   int rem = ctx->rem_ptr;
-   int blocks = len / SIZE512;
+   uint64_t blocks = len / SIZE512;
   __m128i* in = (__m128i*)input;
   int i;

@@ -292,26 +242,22 @@ HashReturn_gr update_and_final_groestl( hashState_groestl* ctx, void* output,
   if ( i == len -1 )
   {        
       // only 128 bits left in buffer, all padding at once
-       ctx->buffer[i] = _mm_set_epi8( blocks,0,0,0, 0,0,0,0,
-                                           0,0,0,0, 0,0,0,0x80 );
+      ctx->buffer[i] = _mm_set_epi64x( blocks << 56, 0x80 );
   }   
   else
   {
       // add first padding
-       ctx->buffer[i] = _mm_set_epi8( 0,0,0,0, 0,0,0,0, 
-                                      0,0,0,0, 0,0,0,0x80 );
+       ctx->buffer[i] = m128_const_64( 0, 0x80 );
       // add zero padding
       for ( i += 1; i < SIZE512 - 1; i++ )
           ctx->buffer[i] = _mm_setzero_si128();

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

   // digest final padding block and do output transform
   TF1024( ctx->chaining, ctx->buffer );
-
   OF1024( ctx->chaining );

   // store hash result in output 
--- a/algo/groestl/aes_ni/hash-groestl256.c
+++ b/algo/groestl/aes_ni/hash-groestl256.c
@@ -13,41 +13,7 @@

 #ifdef __AES__

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

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

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

  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return FAIL_GR;
@@ -249,6 +214,98 @@ HashReturn_gr update_and_final_groestl256( hashState_groestl256* ctx,
   return SUCCESS_GR;
 }

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

+int groestl256_full( hashState_groestl256* ctx,
+                   void* output, const void* input, DataLength_gr databitlen );
+
 #endif /* __hash_h */
--- a/algo/groestl/groestl-4way.c
+++ b/algo/groestl/groestl-4way.c
@@ -53,7 +53,7 @@ int scanhash_groestl_4way( struct work *work, uint32_t max_nonce,
        if ( fulltest( hash+(lane<<3), ptarget) && !opt_benchmark )
        {
           pdata[19] = n + lane;
-           submit_lane_solution( work, hash+(lane<<3), mythr, lane );
+           submit_solution( work, hash+(lane<<3), mythr );
        }
        n += 4;
     } while ( ( n < last_nonce ) && !work_restart[thr_id].restart );
--- a/algo/groestl/groestl.c
+++ b/algo/groestl/groestl.c
@@ -1,4 +1,7 @@
 #include "groestl-gate.h"
+
+#if !defined(GROESTL_8WAY) && !defined(GROESTLX16R_4WAY)
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <stdint.h>
@@ -88,4 +91,4 @@ int scanhash_groestl( struct work *work, uint32_t max_nonce,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
-
+#endif
--- a/algo/groestl/groestl256-hash-4way.c
+++ b/algo/groestl/groestl256-hash-4way.c
@@ -23,7 +23,6 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )
  int i;

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

  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return 1;
@@ -36,9 +35,6 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )

  // The only non-zero in the IV is len. It can be hard coded.
  ctx->chaining[ 3 ] = m512_const2_64( 0, 0x0100000000000000 );
-//  uint64_t len = U64BIG((uint64_t)LENGTH);
-//  ctx->chaining[ COLS/2 -1 ] = _mm512_set4_epi64( len, 0, len, 0 );
-//  INIT256_4way(ctx->chaining);

  ctx->buf_ptr = 0;
  ctx->rem_ptr = 0;
@@ -46,6 +42,77 @@ int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )
  return 0;
 }

+int groestl256_4way_full( groestl256_4way_context* ctx, void* output,
+                                const void* input, uint64_t databitlen )
+{
+   const int len = (int)databitlen / 128;
+   const int hashlen_m128i = 32 / 16;   // bytes to __m128i
+   const int hash_offset = SIZE256 - hashlen_m128i;
+   int rem = ctx->rem_ptr;
+   int blocks = len / SIZE256;
+   __m512i* in = (__m512i*)input;
+   int i;
+
+  if (ctx->chaining == NULL || ctx->buffer == NULL)
+    return 1;
+
+  for ( i = 0; i < SIZE256; i++ )
+  {
+     ctx->chaining[i] = m512_zero;
+     ctx->buffer[i]   = m512_zero;
+  }
+
+  // The only non-zero in the IV is len. It can be hard coded.
+  ctx->chaining[ 3 ] = m512_const2_64( 0, 0x0100000000000000 );
+  ctx->buf_ptr = 0;
+  ctx->rem_ptr = 0;
+   
+   // --- update ---
+
+   // digest any full blocks, process directly from input 
+   for ( i = 0; i < blocks; i++ )
+      TF512_4way( ctx->chaining, &in[ i * SIZE256 ] );
+   ctx->buf_ptr = blocks * SIZE256;
+
+   // copy any remaining data to buffer, it may already contain data
+   // from a previous update for a midstate precalc
+   for ( i = 0; i < len % SIZE256; i++ )
+       ctx->buffer[ rem + i ] = in[ ctx->buf_ptr + i ];
+   i += rem;    // use i as rem_ptr in final
+
+   //--- final ---
+
+   blocks++;      // adjust for final block
+
+   if ( i == SIZE256 - 1 )
+   {        
+       // only 1 vector left in buffer, all padding at once
+      ctx->buffer[i] = m512_const2_64( (uint64_t)blocks << 56, 0x80 ); 
+   }   
+   else
+   {
+       // add first padding
+       ctx->buffer[i] = m512_const4_64( 0, 0x80, 0, 0x80 );
+       // add zero padding
+       for ( i += 1; i < SIZE256 - 1; i++ )
+           ctx->buffer[i] = m512_zero;
+
+       // add length padding, second last byte is zero unless blocks > 255
+      ctx->buffer[i] = m512_const2_64( (uint64_t)blocks << 56, 0 );
+   }
+
+// digest final padding block and do output transform
+   TF512_4way( ctx->chaining, ctx->buffer );
+
+   OF512_4way( ctx->chaining );
+
+   // store hash result in output 
+   for ( i = 0; i < hashlen_m128i; i++ )
+      casti_m512i( output, i ) = ctx->chaining[ hash_offset + i ];
+
+   return 0;
+}
+
 int groestl256_4way_update_close( groestl256_4way_context* ctx, void* output,
                                const void* input, uint64_t databitlen )
 {
@@ -75,11 +142,11 @@ int groestl256_4way_update_close( groestl256_4way_context* ctx, void* output,
   blocks++;      // adjust for final block

   if ( i == SIZE256 - 1 )
-   {        
+   {
       // only 1 vector left in buffer, all padding at once
       ctx->buffer[i] = m512_const1_128( _mm_set_epi8(
                      blocks, blocks>>8,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0x80 ) );
-   }   
+   }
   else
   {
       // add first padding
--- a/algo/groestl/groestl256-hash-4way.h
+++ b/algo/groestl/groestl256-hash-4way.h
@@ -71,5 +71,8 @@ int groestl256_4way_init( groestl256_4way_context*, uint64_t );
 int groestl256_4way_update_close( groestl256_4way_context*,  void*,
                                        const void*, uint64_t );

+int groestl256_4way_full( groestl256_4way_context*, void*,
+                          const void*, uint64_t );
+
 #endif
 #endif 
--- a/algo/groestl/groestl256-intr-4way.h
+++ b/algo/groestl/groestl256-intr-4way.h
@@ -14,17 +14,78 @@
 #include "groestl256-hash-4way.h"

 #if defined(__VAES__)
+static const __m128i round_const_l0[] __attribute__ ((aligned (64))) =
+{
+   { 0x7060504030201000, 0xffffffffffffffff },
+   { 0x7161514131211101, 0xffffffffffffffff },
+   { 0x7262524232221202, 0xffffffffffffffff },
+   { 0x7363534333231303, 0xffffffffffffffff },
+   { 0x7464544434241404, 0xffffffffffffffff },
+   { 0x7565554535251505, 0xffffffffffffffff },
+   { 0x7666564636261606, 0xffffffffffffffff },
+   { 0x7767574737271707, 0xffffffffffffffff },
+   { 0x7868584838281808, 0xffffffffffffffff },
+   { 0x7969594939291909, 0xffffffffffffffff }
+};

-/* global constants  */
-__m512i ROUND_CONST_Lx;
-__m512i ROUND_CONST_L0[ROUNDS512];
-__m512i ROUND_CONST_L7[ROUNDS512];
-//__m512i ROUND_CONST_P[ROUNDS1024];
-//__m512i ROUND_CONST_Q[ROUNDS1024];
-__m512i TRANSP_MASK;
-__m512i SUBSH_MASK[8];
-__m512i ALL_1B;
-__m512i ALL_FF;
+static const __m128i round_const_l7[] __attribute__ ((aligned (64))) =
+{
+   { 0x0000000000000000, 0x8f9fafbfcfdfefff },
+   { 0x0000000000000000, 0x8e9eaebecedeeefe },
+   { 0x0000000000000000, 0x8d9dadbdcdddedfd },
+   { 0x0000000000000000, 0x8c9cacbcccdcecfc },
+   { 0x0000000000000000, 0x8b9babbbcbdbebfb },
+   { 0x0000000000000000, 0x8a9aaabacadaeafa },
+   { 0x0000000000000000, 0x8999a9b9c9d9e9f9 },
+   { 0x0000000000000000, 0x8898a8b8c8d8e8f8 },
+   { 0x0000000000000000, 0x8797a7b7c7d7e7f7 },
+   { 0x0000000000000000, 0x8696a6b6c6d6e6f6 }
+};
+
+static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
+                                     0x1d1519111c141810, 0x1f171b131e161a12,
+                                     0x2d2529212c242820, 0x2f272b232e262a22,
+                                     0x3d3539313c343830, 0x3f373b333e363a32 };
+
+static const __m512i SUBSH_MASK0 = { 0x0c0f0104070b0e00, 0x03060a0d08020509,
+                                     0x1c1f1114171b1e10, 0x13161a1d18121519,
+                                     0x2c2f2124272b2e20, 0x23262a2d28222529,
+                                     0x3c3f3134373b3e30, 0x33363a3d38323539 };
+
+static const __m512i SUBSH_MASK1 = { 0x0e090205000d0801, 0x04070c0f0a03060b,
+                                     0x1e191215101d1801, 0x14171c1f1a13161b,
+                                     0x2e292225202d2821, 0x24272c2f2a23262b,
+                                     0x3e393235303d3831, 0x34373c3f3a33363b };
+
+static const __m512i SUBSH_MASK2 = { 0x080b0306010f0a02, 0x05000e090c04070d,
+                                     0x181b1316111f1a12, 0x15101e191c14171d,
+                                     0x282b2326212f2a22, 0x25202e292c24272d,
+                                     0x383b3336313f3a32, 0x35303e393c34373d };
+
+static const __m512i SUBSH_MASK3 = { 0x0a0d040702090c03, 0x0601080b0e05000f,
+                                     0x1a1d141712191c13, 0x1611181b1e15101f,
+                                     0x2a2d242722292c23, 0x2621282b2e25202f,
+                                     0x3a3d343732393c33, 0x3631383b3e35303f };
+
+static const __m512i SUBSH_MASK4 = { 0x0b0e0500030a0d04, 0x0702090c0f060108,
+                                     0x1b1e1510131a1d14, 0x1712191c1f161118,
+                                     0x2b2e2520232a2d24, 0x2722292c2f262128,
+                                     0x3b3e3530333a3d34, 0x3732393c3f363138 };
+
+static const __m512i SUBSH_MASK5 = { 0x0d080601040c0f05, 0x00030b0e0907020a,
+                                     0x1d181611141c1f15, 0x10131b1e1917121a,
+                                     0x2d282621242c2f25, 0x20232b2e2927222a,
+                                     0x3d383631343c3f35, 0x30333b3e3937323a };
+
+static const __m512i SUBSH_MASK6 = { 0x0f0a0702050e0906, 0x01040d080b00030c,
+                                     0x1f1a1712151e1916, 0x11141d181b10131c,
+                                     0x2f2a2722252e2926, 0x21242d282b20232c,
+                                     0x3f3a3732353e3936, 0x31343d383b30333c };
+
+static const __m512i SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e,
+                                     0x191c101316181b17, 0x12151f1a1d11141e,
+                                     0x292c202326282b27, 0x22252f2a2d21242e,
+                                     0x393c303336383b37, 0x32353f3a3d31343e };

 #define tos(a)    #a
 #define tostr(a)  tos(a)
@@ -40,8 +101,6 @@ __m512i ALL_FF;
  i = _mm512_xor_si512(i, j);\
 } 

- /**/
-
 /* Yet another implementation of MixBytes.
   This time we use the formulae (3) from the paper "Byte Slicing Groestl".
   Input: a0, ..., a7
@@ -155,95 +214,36 @@ __m512i ALL_FF;
  b1 = _mm512_xor_si512(b1, a4);\
 }/*MixBytes*/

-// calculate the round constants seperately and load at startup
-
-#define SET_CONSTANTS(){\
-  ALL_1B = _mm512_set1_epi32( 0x1b1b1b1b );\
-  TRANSP_MASK   = _mm512_set_epi32( \
-                         0x3f373b33, 0x3e363a32, 0x3d353931, 0x3c343830, \
-                         0x2f272b23, 0x2e262a22, 0x2d252921, 0x2c242820, \
-                         0x1f171b13, 0x1e161a12, 0x1d151911, 0x1c141810, \
-                         0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800 ); \
-  SUBSH_MASK[0] = _mm512_set_epi32( \
-                         0x33363a3d, 0x38323539, 0x3c3f3134, 0x373b3e30, \
-                         0x23262a2d, 0x28222529, 0x2c2f2124, 0x272b2e20, \
-                         0x13161a1d, 0x18121519, 0x1c1f1114, 0x171b1e10, \
-                         0x03060a0d, 0x08020509, 0x0c0f0104, 0x070b0e00 ); \
-  SUBSH_MASK[1] = _mm512_set_epi32( \
-                         0x34373c3f, 0x3a33363b, 0x3e393235, 0x303d3831, \
-                         0x24272c2f, 0x2a23262b, 0x2e292225, 0x202d2821, \
-                         0x14171c1f, 0x1a13161b, 0x1e191215, 0x101d1801, \
-                         0x04070c0f, 0x0a03060b, 0x0e090205, 0x000d0801 );\
-  SUBSH_MASK[2] = _mm512_set_epi32( \
-                         0x35303e39, 0x3c34373d, 0x383b3336, 0x313f3a32, \
-                         0x25202e29, 0x2c24272d, 0x282b2326, 0x212f2a22, \
-                         0x15101e19, 0x1c14171d, 0x181b1316, 0x111f1a12, \
-                         0x05000e09, 0x0c04070d, 0x080b0306, 0x010f0a02 );\
-  SUBSH_MASK[3] = _mm512_set_epi32( \
-                         0x3631383b, 0x3e35303f, 0x3a3d3437, 0x32393c33, \
-                         0x2621282b, 0x2e25202f, 0x2a2d2427, 0x22292c23, \
-                         0x1611181b, 0x1e15101f, 0x1a1d1417, 0x12191c13, \
-                         0x0601080b, 0x0e05000f, 0x0a0d0407, 0x02090c03 );\
-  SUBSH_MASK[4] = _mm512_set_epi32( \
-                         0x3732393c, 0x3f363138, 0x3b3e3530, 0x333a3d34, \
-                         0x2722292c, 0x2f262128, 0x2b2e2520, 0x232a2d24, \
-                         0x1712191c, 0x1f161118, 0x1b1e1510, 0x131a1d14, \
-                         0x0702090c, 0x0f060108, 0x0b0e0500, 0x030a0d04 );\
-  SUBSH_MASK[5] = _mm512_set_epi32( \
-                         0x30333b3e, 0x3937323a, 0x3d383631, 0x343c3f35, \
-                         0x20232b2e, 0x2927222a, 0x2d282621, 0x242c2f25, \
-                         0x10131b1e, 0x1917121a, 0x1d181611, 0x141c1f15, \
-                         0x00030b0e, 0x0907020a, 0x0d080601, 0x040c0f05 );\
-  SUBSH_MASK[6] = _mm512_set_epi32( \
-                         0x31343d38, 0x3b30333c, 0x3f3a3732, 0x353e3936, \
-                         0x21242d28, 0x2b20232c, 0x2f2a2722, 0x252e2926, \
-                         0x11141d18, 0x1b10131c, 0x1f1a1712, 0x151e1916, \
-                         0x01040d08, 0x0b00030c, 0x0f0a0702, 0x050e0906 );\
-  SUBSH_MASK[7] = _mm512_set_epi32( \
-                         0x32353f3a, 0x3d31343e, 0x393c3033, 0x36383b37, \
-                         0x22252f2a, 0x2d21242e, 0x292c2023, 0x26282b27, \
-                         0x12151f1a, 0x1d11141e, 0x191c1013, 0x16181b17, \
-                         0x02050f0a, 0x0d01040e, 0x090c0003, 0x06080b07 );\
-  for ( i = 0; i < ROUNDS512; i++ ) \
-  {\
-    ROUND_CONST_L0[i] = _mm512_set4_epi32( 0xffffffff, 0xffffffff, \
-          0x70605040 ^ ( i * 0x01010101 ), 0x30201000 ^ ( i * 0x01010101 ) ); \
-    ROUND_CONST_L7[i] = _mm512_set4_epi32( 0x8f9fafbf ^ ( i * 0x01010101 ), \
-          0xcfdfefff ^ ( i * 0x01010101 ), 0x00000000, 0x00000000 ); \
-  }\
-  ROUND_CONST_Lx = _mm512_set4_epi32( 0xffffffff, 0xffffffff, \
-                                      0x00000000, 0x00000000 ); \
-}while(0);\

 #define ROUND(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
  /* AddRoundConstant */\
-  b1 = ROUND_CONST_Lx;\
-  a0 = _mm512_xor_si512( a0, (ROUND_CONST_L0[i]) );\
+  b1 = m512_const2_64( 0xffffffffffffffff, 0 ); \
+  a0 = _mm512_xor_si512( a0, m512_const1_128( round_const_l0[i] ) );\
  a1 = _mm512_xor_si512( a1, b1 );\
  a2 = _mm512_xor_si512( a2, b1 );\
  a3 = _mm512_xor_si512( a3, b1 );\
  a4 = _mm512_xor_si512( a4, b1 );\
  a5 = _mm512_xor_si512( a5, b1 );\
  a6 = _mm512_xor_si512( a6, b1 );\
-  a7 = _mm512_xor_si512( a7, (ROUND_CONST_L7[i]) );\
+  a7 = _mm512_xor_si512( a7, m512_const1_128( round_const_l7[i] ) );\
  \
  /* ShiftBytes + SubBytes (interleaved) */\
  b0 = _mm512_xor_si512( b0, b0 );\
-  a0 = _mm512_shuffle_epi8( a0, (SUBSH_MASK[0]) );\
+  a0 = _mm512_shuffle_epi8( a0, SUBSH_MASK0 );\
  a0 = _mm512_aesenclast_epi128(a0, b0 );\
-  a1 = _mm512_shuffle_epi8( a1, (SUBSH_MASK[1]) );\
+  a1 = _mm512_shuffle_epi8( a1, SUBSH_MASK1 );\
  a1 = _mm512_aesenclast_epi128(a1, b0 );\
-  a2 = _mm512_shuffle_epi8( a2, (SUBSH_MASK[2]) );\
+  a2 = _mm512_shuffle_epi8( a2, SUBSH_MASK2 );\
  a2 = _mm512_aesenclast_epi128(a2, b0 );\
-  a3 = _mm512_shuffle_epi8( a3, (SUBSH_MASK[3]) );\
+  a3 = _mm512_shuffle_epi8( a3, SUBSH_MASK3 );\
  a3 = _mm512_aesenclast_epi128(a3, b0 );\
-  a4 = _mm512_shuffle_epi8( a4, (SUBSH_MASK[4]) );\
+  a4 = _mm512_shuffle_epi8( a4, SUBSH_MASK4 );\
  a4 = _mm512_aesenclast_epi128(a4, b0 );\
-  a5 = _mm512_shuffle_epi8( a5, (SUBSH_MASK[5]) );\
+  a5 = _mm512_shuffle_epi8( a5, SUBSH_MASK5 );\
  a5 = _mm512_aesenclast_epi128(a5, b0 );\
-  a6 = _mm512_shuffle_epi8( a6, (SUBSH_MASK[6]) );\
+  a6 = _mm512_shuffle_epi8( a6, SUBSH_MASK6 );\
  a6 = _mm512_aesenclast_epi128(a6, b0 );\
-  a7 = _mm512_shuffle_epi8( a7, (SUBSH_MASK[7]) );\
+  a7 = _mm512_shuffle_epi8( a7, SUBSH_MASK7 );\
  a7 = _mm512_aesenclast_epi128( a7, b0 );\
  \
  /* MixBytes */\
@@ -390,29 +390,6 @@ __m512i ALL_FF;
 }/**/


-
-void INIT256_4way( __m512i* chaining )
-{
-  static __m512i xmm0, xmm2, xmm6, xmm7;
-  static __m512i xmm12, xmm13, xmm14, xmm15;
-
-  /* load IV into registers xmm12 - xmm15 */
-  xmm12 = chaining[0];
-  xmm13 = chaining[1];
-  xmm14 = chaining[2];
-  xmm15 = chaining[3];
-
-  /* transform chaining value from column ordering into row ordering */
-  /* we put two rows (64 bit) of the IV into one 128-bit XMM register */
-  Matrix_Transpose_A(xmm12, xmm13, xmm14, xmm15, xmm2, xmm6, xmm7, xmm0);
-
-  /* store transposed IV */
-  chaining[0] = xmm12;
-  chaining[1] = xmm2;
-  chaining[2] = xmm6;
-  chaining[3] = xmm7;
-}
-
 void TF512_4way( __m512i* chaining, __m512i* message )
 {
  static __m512i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
--- a/algo/groestl/groestl512-hash-4way.c
+++ b/algo/groestl/groestl512-hash-4way.c
@@ -19,10 +19,6 @@

 int groestl512_4way_init( groestl512_4way_context* ctx, uint64_t hashlen )
 {
-  int i;
-
-  SET_CONSTANTS();
-
  if (ctx->chaining == NULL || ctx->buffer == NULL)
    return 1;

@@ -99,7 +95,6 @@ int groestl512_4way_full( groestl512_4way_context* ctx, void* output,

   // --- init ---

-   SET_CONSTANTS();
   memset_zero_512( ctx->chaining, SIZE512 );
   memset_zero_512( ctx->buffer, SIZE512 );
   ctx->chaining[ 6 ] = m512_const2_64( 0x0200000000000000, 0 );
--- a/algo/groestl/groestl512-intr-4way.h
+++ b/algo/groestl/groestl512-intr-4way.h
@@ -15,16 +15,86 @@

 #if defined(__VAES__)

-/* global constants  */
-__m512i ROUND_CONST_Lx;
-//__m128i ROUND_CONST_L0[ROUNDS512];
-//__m128i ROUND_CONST_L7[ROUNDS512];
-__m512i ROUND_CONST_P[ROUNDS1024];
-__m512i ROUND_CONST_Q[ROUNDS1024];
-__m512i TRANSP_MASK;
-__m512i SUBSH_MASK[8];
-__m512i ALL_1B;
-__m512i ALL_FF;
+static const __m128i round_const_p[] __attribute__ ((aligned (64))) =
+{
+   { 0x7060504030201000, 0xf0e0d0c0b0a09080 },
+   { 0x7161514131211101, 0xf1e1d1c1b1a19181 }, 
+   { 0x7262524232221202, 0xf2e2d2c2b2a29282 },
+   { 0x7363534333231303, 0xf3e3d3c3b3a39383 },
+   { 0x7464544434241404, 0xf4e4d4c4b4a49484 },
+   { 0x7565554535251505, 0xf5e5d5c5b5a59585 },
+   { 0x7666564636261606, 0xf6e6d6c6b6a69686 },
+   { 0x7767574737271707, 0xf7e7d7c7b7a79787 },
+   { 0x7868584838281808, 0xf8e8d8c8b8a89888 },
+   { 0x7969594939291909, 0xf9e9d9c9b9a99989 },
+   { 0x7a6a5a4a3a2a1a0a, 0xfaeadacabaaa9a8a },
+   { 0x7b6b5b4b3b2b1b0b, 0xfbebdbcbbbab9b8b },
+   { 0x7c6c5c4c3c2c1c0c, 0xfcecdcccbcac9c8c },
+   { 0x7d6d5d4d3d2d1d0d, 0xfdedddcdbdad9d8d }
+};
+
+static const __m128i round_const_q[] __attribute__ ((aligned (64))) =
+{
+   { 0x8f9fafbfcfdfefff, 0x0f1f2f3f4f5f6f7f },
+   { 0x8e9eaebecedeeefe, 0x0e1e2e3e4e5e6e7e },
+   { 0x8d9dadbdcdddedfd, 0x0d1d2d3d4d5d6d7d },
+   { 0x8c9cacbcccdcecfc, 0x0c1c2c3c4c5c6c7c },
+   { 0x8b9babbbcbdbebfb, 0x0b1b2b3b4b5b6b7b },
+   { 0x8a9aaabacadaeafa, 0x0a1a2a3a4a5a6a7a },
+   { 0x8999a9b9c9d9e9f9, 0x0919293949596979 },
+   { 0x8898a8b8c8d8e8f8, 0x0818283848586878 },
+   { 0x8797a7b7c7d7e7f7, 0x0717273747576777 },
+   { 0x8696a6b6c6d6e6f6, 0x0616263646566676 },
+   { 0x8595a5b5c5d5e5f5, 0x0515253545556575 },
+   { 0x8494a4b4c4d4e4f4, 0x0414243444546474 },
+   { 0x8393a3b3c3d3e3f3, 0x0313233343536373 },
+   { 0x8292a2b2c2d2e2f2, 0x0212223242526272 }
+};
+
+static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
+                                     0x1d1519111c141810, 0x1f171b131e161a12,
+                                     0x2d2529212c242820, 0x2f272b232e262a22,
+                                     0x3d3539313c343830, 0x3f373b333e363a32 };
+
+static const __m512i SUBSH_MASK0 = { 0x0b0e0104070a0d00, 0x0306090c0f020508,
+                                     0x1b1e1114171a1d10, 0x1316191c1f121518,
+                                     0x2b2e2124272a2d20, 0x2326292c2f222528,
+                                     0x3b3e3134373a3d30, 0x3336393c3f323538 };
+
+static const __m512i SUBSH_MASK1 = { 0x0c0f0205080b0e01, 0x04070a0d00030609,
+                                     0x1c1f1215181b1e11, 0x14171a1d10131619,
+                                     0x2c2f2225282b2e21, 0x24272a2d20232629,
+                                     0x3c3f3235383b3e31, 0x34373a3d30333639 };
+
+static const __m512i SUBSH_MASK2 = { 0x0d000306090c0f02, 0x05080b0e0104070a,
+                                     0x1d101316191c1f12, 0x15181b1e1114171a,
+                                     0x2d202326292c2f22, 0x25282b2e2124272a,
+                                     0x3d303336393c3f32, 0x35383b3e3134373a };
+
+static const __m512i SUBSH_MASK3 = { 0x0e0104070a0d0003, 0x06090c0f0205080b,
+                                     0x1e1114171a1d1013, 0x16191c1f1215181b,
+                                     0x2e2124272a2d2023, 0x26292c2f2225282b,
+                                     0x3e3134373a3d3033, 0x36393c3f3235383b };
+
+static const __m512i SUBSH_MASK4 = { 0x0f0205080b0e0104, 0x070a0d000306090c,
+                                     0x1f1215181b1e1114, 0x171a1d101316191c,
+                                     0x2f2225282b2e2124, 0x272a2d202326292c,
+                                     0x3f3235383b3e3134, 0x373a3d303336393c };
+
+static const __m512i SUBSH_MASK5 = { 0x000306090c0f0205, 0x080b0e0104070a0d,
+                                     0x101316191c1f1215, 0x181b1e1114171a1d,
+                                     0x202326292c2f2225, 0x282b2e2124272a2d,
+                                     0x303336393c3f3235, 0x383b3e3134373a3d };
+
+static const __m512i SUBSH_MASK6 = { 0x0104070a0d000306, 0x090c0f0205080b0e,
+                                     0x1114171a1d101316, 0x191c1f1215181b1e,
+                                     0x2124272a2d202326, 0x292c2f2225282b2e,
+                                     0x3134373a3d303336, 0x393c3f3235383b3e };
+
+static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
+                                     0x16191c1f1215181b, 0x1e1114171a1d1013,
+                                     0x26292c2f2225282b, 0x2e2124272a2d2023,
+                                     0x36393c3f3235383b, 0x3e3134373a3d3033 };

 #define tos(a)    #a
 #define tostr(a)  tos(a)
@@ -155,69 +225,6 @@ __m512i ALL_FF;
  b1 = _mm512_xor_si512(b1, a4);\
 }/*MixBytes*/

-// calculate the round constants seperately and load at startup
-
-#define SET_CONSTANTS(){\
-  ALL_FF = _mm512_set1_epi32( 0xffffffff );\
-  ALL_1B = _mm512_set1_epi32( 0x1b1b1b1b );\
-  TRANSP_MASK   = _mm512_set_epi32( \
-                         0x3f373b33, 0x3e363a32, 0x3d353931, 0x3c343830, \
-                         0x2f272b23, 0x2e262a22, 0x2d252921, 0x2c242820, \
-                         0x1f171b13, 0x1e161a12, 0x1d151911, 0x1c141810, \
-                         0x0f070b03, 0x0e060a02, 0x0d050901, 0x0c040800 ); \
-  SUBSH_MASK[0] = _mm512_set_epi32( \
-                         0x3336393c, 0x3f323538, 0x3b3e3134, 0x373a3d30, \
-                         0x2326292c, 0x2f222528, 0x2b2e2124, 0x272a2d20, \
-                         0x1316191c, 0x1f121518, 0x1b1e1114, 0x171a1d10, \
-                         0x0306090c, 0x0f020508, 0x0b0e0104, 0x070a0d00 ); \
-  SUBSH_MASK[1] = _mm512_set_epi32( \
-                         0x34373a3d, 0x30333639, 0x3c3f3235, 0x383b3e31, \
-                         0x24272a2d, 0x20232629, 0x2c2f2225, 0x282b2e21, \
-                         0x14171a1d, 0x10131619, 0x1c1f1215, 0x181b1e11, \
-                         0x04070a0d, 0x00030609, 0x0c0f0205, 0x080b0e01 ); \
-  SUBSH_MASK[2] = _mm512_set_epi32( \
-                         0x35383b3e, 0x3134373a, 0x3d303336, 0x393c3f32, \
-                         0x25282b2e, 0x2124272a, 0x2d202326, 0x292c2f22, \
-                         0x15181b1e, 0x1114171a, 0x1d101316, 0x191c1f12, \
-                         0x05080b0e, 0x0104070a, 0x0d000306, 0x090c0f02 ); \
-  SUBSH_MASK[3] = _mm512_set_epi32( \
-                         0x36393c3f, 0x3235383b, 0x3e313437, 0x3a3d3033, \
-                         0x26292c2f, 0x2225282b, 0x2e212427, 0x2a2d2023, \
-                         0x16191c1f, 0x1215181b, 0x1e111417, 0x1a1d1013, \
-                         0x06090c0f, 0x0205080b, 0x0e010407, 0x0a0d0003 ); \
-  SUBSH_MASK[4] = _mm512_set_epi32( \
-                         0x373a3d30, 0x3336393c, 0x3f323538, 0x3b3e3134, \
-                         0x272a2d20, 0x2326292c, 0x2f222528, 0x2b2e2124, \
-                         0x171a1d10, 0x1316191c, 0x1f121518, 0x1b1e1114, \
-                         0x070a0d00, 0x0306090c, 0x0f020508, 0x0b0e0104 ); \
-  SUBSH_MASK[5] = _mm512_set_epi32( \
-                         0x383b3e31, 0x34373a3d, 0x30333639, 0x3c3f3235, \
-                         0x282b2e21, 0x24272a2d, 0x20232629, 0x2c2f2225, \
-                         0x181b1e11, 0x14171a1d, 0x10131619, 0x1c1f1215, \
-                         0x080b0e01, 0x04070a0d, 0x00030609, 0x0c0f0205 ); \
-  SUBSH_MASK[6] = _mm512_set_epi32( \
-                         0x393c3f32, 0x35383b3e, 0x3134373a, 0x3d303336, \
-                         0x292c2f22, 0x25282b2e, 0x2124272a, 0x2d202326, \
-                         0x191c1f12, 0x15181b1e, 0x1114171a, 0x1d101316, \
-                         0x090c0f02, 0x05080b0e, 0x0104070a, 0x0d000306 ); \
-  SUBSH_MASK[7] = _mm512_set_epi32( \
-                         0x3e313437, 0x3a3d3033, 0x36393c3f, 0x3235383b, \
-                         0x2e212427, 0x2a2d2023, 0x26292c2f, 0x2225282b, \
-                         0x1e111417, 0x1a1d1013, 0x16191c1f, 0x1215181b, \
-                         0x0e010407, 0x0a0d0003, 0x06090c0f, 0x0205080b ); \
-  for( i = 0; i < ROUNDS1024; i++ ) \
-  { \
-    ROUND_CONST_P[i] = _mm512_set4_epi32( 0xf0e0d0c0 ^ (i * 0x01010101), \
-                                          0xb0a09080 ^ (i * 0x01010101), \
-                                          0x70605040 ^ (i * 0x01010101), \
-                                          0x30201000 ^ (i * 0x01010101) ); \
-    ROUND_CONST_Q[i] = _mm512_set4_epi32( 0x0f1f2f3f ^ (i * 0x01010101), \
-                                          0x4f5f6f7f ^ (i * 0x01010101), \
-                                          0x8f9fafbf ^ (i * 0x01010101), \
-                                          0xcfdfefff ^ (i * 0x01010101));\
-  } \
-}while(0);\
-
 /* one round
 * a0-a7 = input rows
 * b0-b7 = output rows
@@ -242,30 +249,32 @@ __m512i ALL_FF;
  for ( round_counter = 0; round_counter < 14; round_counter += 2 ) \
  { \
    /* AddRoundConstant P1024 */\
-    xmm8 = _mm512_xor_si512( xmm8, ( ROUND_CONST_P[ round_counter ] ) );\
+    xmm8 = _mm512_xor_si512( xmm8, m512_const1_128( \
+             casti_m128i( round_const_p, round_counter ) ) ); \
    /* ShiftBytes P1024 + pre-AESENCLAST */\
-    xmm8  = _mm512_shuffle_epi8( xmm8,  ( SUBSH_MASK[0] ) );\
-    xmm9  = _mm512_shuffle_epi8( xmm9,  ( SUBSH_MASK[1] ) );\
-    xmm10 = _mm512_shuffle_epi8( xmm10, ( SUBSH_MASK[2] ) );\
-    xmm11 = _mm512_shuffle_epi8( xmm11, ( SUBSH_MASK[3] ) );\
-    xmm12 = _mm512_shuffle_epi8( xmm12, ( SUBSH_MASK[4] ) );\
-    xmm13 = _mm512_shuffle_epi8( xmm13, ( SUBSH_MASK[5] ) );\
-    xmm14 = _mm512_shuffle_epi8( xmm14, ( SUBSH_MASK[6] ) );\
-    xmm15 = _mm512_shuffle_epi8( xmm15, ( SUBSH_MASK[7] ) );\
+    xmm8  = _mm512_shuffle_epi8( xmm8,  SUBSH_MASK0 ); \
+    xmm9  = _mm512_shuffle_epi8( xmm9,  SUBSH_MASK1 );\
+    xmm10 = _mm512_shuffle_epi8( xmm10, SUBSH_MASK2 );\
+    xmm11 = _mm512_shuffle_epi8( xmm11, SUBSH_MASK3 );\
+    xmm12 = _mm512_shuffle_epi8( xmm12, SUBSH_MASK4 );\
+    xmm13 = _mm512_shuffle_epi8( xmm13, SUBSH_MASK5 );\
+    xmm14 = _mm512_shuffle_epi8( xmm14, SUBSH_MASK6 );\
+    xmm15 = _mm512_shuffle_epi8( xmm15, SUBSH_MASK7 );\
    /* SubBytes + MixBytes */\
    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
    \
     /* AddRoundConstant P1024 */\
-    xmm0 = _mm512_xor_si512( xmm0, ( ROUND_CONST_P[ round_counter+1 ] ) );\
+    xmm0 = _mm512_xor_si512( xmm0, m512_const1_128( \
+             casti_m128i( round_const_p, round_counter+1 ) ) ); \
    /* ShiftBytes P1024 + pre-AESENCLAST */\
-    xmm0 = _mm512_shuffle_epi8( xmm0, ( SUBSH_MASK[0] ) );\
-    xmm1 = _mm512_shuffle_epi8( xmm1, ( SUBSH_MASK[1] ) );\
-    xmm2 = _mm512_shuffle_epi8( xmm2, ( SUBSH_MASK[2] ) );\
-    xmm3 = _mm512_shuffle_epi8( xmm3, ( SUBSH_MASK[3] ) );\
-    xmm4 = _mm512_shuffle_epi8( xmm4, ( SUBSH_MASK[4] ) );\
-    xmm5 = _mm512_shuffle_epi8( xmm5, ( SUBSH_MASK[5] ) );\
-    xmm6 = _mm512_shuffle_epi8( xmm6, ( SUBSH_MASK[6] ) );\
-    xmm7 = _mm512_shuffle_epi8( xmm7, ( SUBSH_MASK[7] ) );\
+    xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK0 );\
+    xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK1 );\
+    xmm2 = _mm512_shuffle_epi8( xmm2, SUBSH_MASK2 );\
+    xmm3 = _mm512_shuffle_epi8( xmm3, SUBSH_MASK3 );\
+    xmm4 = _mm512_shuffle_epi8( xmm4, SUBSH_MASK4 );\
+    xmm5 = _mm512_shuffle_epi8( xmm5, SUBSH_MASK5 );\
+    xmm6 = _mm512_shuffle_epi8( xmm6, SUBSH_MASK6 );\
+    xmm7 = _mm512_shuffle_epi8( xmm7, SUBSH_MASK7 );\
    /* SubBytes + MixBytes */\
     SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
  }\
@@ -284,16 +293,17 @@ __m512i ALL_FF;
    xmm12 = _mm512_xor_si512( xmm12, xmm1 );\
    xmm13 = _mm512_xor_si512( xmm13, xmm1 );\
    xmm14 = _mm512_xor_si512( xmm14, xmm1 );\
-    xmm15 = _mm512_xor_si512( xmm15, ( ROUND_CONST_Q[ round_counter ] ) );\
+    xmm15 = _mm512_xor_si512( xmm15, m512_const1_128( \
+                 casti_m128i( round_const_q, round_counter ) ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm8  = _mm512_shuffle_epi8( xmm8,  ( SUBSH_MASK[1] ) );\
-    xmm9  = _mm512_shuffle_epi8( xmm9,  ( SUBSH_MASK[3] ) );\
-    xmm10 = _mm512_shuffle_epi8( xmm10, ( SUBSH_MASK[5] ) );\
-    xmm11 = _mm512_shuffle_epi8( xmm11, ( SUBSH_MASK[7] ) );\
-    xmm12 = _mm512_shuffle_epi8( xmm12, ( SUBSH_MASK[0] ) );\
-    xmm13 = _mm512_shuffle_epi8( xmm13, ( SUBSH_MASK[2] ) );\
-    xmm14 = _mm512_shuffle_epi8( xmm14, ( SUBSH_MASK[4] ) );\
-    xmm15 = _mm512_shuffle_epi8( xmm15, ( SUBSH_MASK[6] ) );\
+    xmm8  = _mm512_shuffle_epi8( xmm8,  SUBSH_MASK1 );\
+    xmm9  = _mm512_shuffle_epi8( xmm9,  SUBSH_MASK3 );\
+    xmm10 = _mm512_shuffle_epi8( xmm10, SUBSH_MASK5 );\
+    xmm11 = _mm512_shuffle_epi8( xmm11, SUBSH_MASK7 );\
+    xmm12 = _mm512_shuffle_epi8( xmm12, SUBSH_MASK0 );\
+    xmm13 = _mm512_shuffle_epi8( xmm13, SUBSH_MASK2 );\
+    xmm14 = _mm512_shuffle_epi8( xmm14, SUBSH_MASK4 );\
+    xmm15 = _mm512_shuffle_epi8( xmm15, SUBSH_MASK6 );\
    /* SubBytes + MixBytes */\
    SUBMIX(xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7);\
    \
@@ -306,16 +316,17 @@ __m512i ALL_FF;
    xmm4 = _mm512_xor_si512( xmm4, xmm9 );\
    xmm5 = _mm512_xor_si512( xmm5, xmm9 );\
    xmm6 = _mm512_xor_si512( xmm6, xmm9 );\
-    xmm7 = _mm512_xor_si512( xmm7, ( ROUND_CONST_Q[ round_counter+1 ] ) );\
+    xmm7 = _mm512_xor_si512( xmm7, m512_const1_128( \
+             casti_m128i( round_const_q, round_counter+1 ) ) ); \
    /* ShiftBytes Q1024 + pre-AESENCLAST */\
-    xmm0 = _mm512_shuffle_epi8( xmm0, ( SUBSH_MASK[1] ) );\
-    xmm1 = _mm512_shuffle_epi8( xmm1, ( SUBSH_MASK[3] ) );\
-    xmm2 = _mm512_shuffle_epi8( xmm2, ( SUBSH_MASK[5] ) );\
-    xmm3 = _mm512_shuffle_epi8( xmm3, ( SUBSH_MASK[7] ) );\
-    xmm4 = _mm512_shuffle_epi8( xmm4, ( SUBSH_MASK[0] ) );\
-    xmm5 = _mm512_shuffle_epi8( xmm5, ( SUBSH_MASK[2] ) );\
-    xmm6 = _mm512_shuffle_epi8( xmm6, ( SUBSH_MASK[4] ) );\
-    xmm7 = _mm512_shuffle_epi8( xmm7, ( SUBSH_MASK[6] ) );\
+    xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK1 );\
+    xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK3 );\
+    xmm2 = _mm512_shuffle_epi8( xmm2, SUBSH_MASK5 );\
+    xmm3 = _mm512_shuffle_epi8( xmm3, SUBSH_MASK7 );\
+    xmm4 = _mm512_shuffle_epi8( xmm4, SUBSH_MASK0 );\
+    xmm5 = _mm512_shuffle_epi8( xmm5, SUBSH_MASK2 );\
+    xmm6 = _mm512_shuffle_epi8( xmm6, SUBSH_MASK4 );\
+    xmm7 = _mm512_shuffle_epi8( xmm7, SUBSH_MASK6 );\
    /* SubBytes + MixBytes */\
    SUBMIX(xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15);\
  }\
--- a/algo/groestl/myr-groestl.c
+++ b/algo/groestl/myr-groestl.c
@@ -1,4 +1,7 @@
 #include "myrgr-gate.h"
+
+#if !defined(MYRGR_8WAY) && !defined(MYRGR_4WAY)
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <stdint.h>
@@ -86,3 +89,4 @@ int scanhash_myriad( struct work *work, uint32_t max_nonce,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
+#endif
--- a/algo/groestl/myrgr-4way.c
+++ b/algo/groestl/myrgr-4way.c
@@ -143,7 +143,7 @@ int scanhash_myriad_8way( struct work *work, uint32_t max_nonce,
         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
         {
            pdata[19] = n + lane;
-            submit_lane_solution( work, lane_hash, mythr, lane );
+            submit_solution( work, lane_hash, mythr );
         }
      }
      n += 8;
@@ -226,7 +226,7 @@ int scanhash_myriad_4way( struct work *work, uint32_t max_nonce,
         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
         {
            pdata[19] = n + lane;
-            submit_lane_solution( work, lane_hash, mythr, lane );
+            submit_solution( work, lane_hash, mythr );
         }
      }
      n += 4;
--- a/algo/groestl/sph_groestl.c
+++ b/algo/groestl/sph_groestl.c
@@ -35,6 +35,8 @@

 #include "sph_groestl.h"

+#if !defined(__AES__)
+
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -3116,4 +3118,6 @@ sph_groestl512_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)

 #ifdef __cplusplus
 }
+
+#endif  // !AES
 #endif
--- a/algo/groestl/sph_groestl.h
+++ b/algo/groestl/sph_groestl.h
@@ -42,6 +42,7 @@ extern "C"{
 #include <stddef.h>
 #include "algo/sha/sph_types.h"

+#if !defined(__AES__)   
 /**
 * Output size (in bits) for Groestl-224.
 */
@@ -326,4 +327,5 @@ void sph_groestl512_addbits_and_close(
 }
 #endif

+#endif  // !AES
 #endif
--- a/algo/heavy/bastion.c
+++ b/algo/heavy/bastion.c
@@ -1,156 +0,0 @@
-#include "algo-gate-api.h"
-#include <stdio.h>
-#include <string.h>
-#include <openssl/sha.h>
-#include <stdint.h>
-#include <stdlib.h>
-#include "sph_hefty1.h"
-#include "algo/luffa/sph_luffa.h"
-#include "algo/fugue/sph_fugue.h"
-#include "algo/skein/sph_skein.h"
-#include "algo/whirlpool/sph_whirlpool.h"
-#include "algo/shabal/sph_shabal.h"
-#include "algo/echo/sph_echo.h"
-#include "algo/hamsi/sph_hamsi.h"
-#include "algo/luffa/luffa_for_sse2.h"
-#ifdef __AES__
-  #include "algo/echo/aes_ni/hash_api.h"
-#endif
-
-void bastionhash(void *output, const void *input)
-{
-	unsigned char hash[64] __attribute__ ((aligned (64)));
-
-#ifdef __AES__
-   hashState_echo          ctx_echo;
-#else
-   sph_echo512_context     ctx_echo;
-#endif
-   hashState_luffa         ctx_luffa;
-	sph_fugue512_context ctx_fugue;
-	sph_whirlpool_context ctx_whirlpool;
-	sph_shabal512_context ctx_shabal;
-   sph_hamsi512_context ctx_hamsi;
-	sph_skein512_context ctx_skein;
-
-	HEFTY1(input, 80, hash);
-
-        init_luffa( &ctx_luffa, 512 );
-        update_and_final_luffa( &ctx_luffa, (BitSequence*)hash,
-                                (const BitSequence*)hash, 64 );
-
-	if (hash[0] & 0x8)
-	{
-		sph_fugue512_init(&ctx_fugue);
-		sph_fugue512(&ctx_fugue, hash, 64);
-		sph_fugue512_close(&ctx_fugue, hash);
-	} else {
-   sph_skein512_init( &ctx_skein );
-   sph_skein512( &ctx_skein, hash, 64 );
-   sph_skein512_close( &ctx_skein, hash );
-	}
-
-	sph_whirlpool_init(&ctx_whirlpool);
-	sph_whirlpool(&ctx_whirlpool, hash, 64);
-	sph_whirlpool_close(&ctx_whirlpool, hash);
-
-	sph_fugue512_init(&ctx_fugue);
-	sph_fugue512(&ctx_fugue, hash, 64);
-	sph_fugue512_close(&ctx_fugue, hash);
-
-	if (hash[0] & 0x8)
-	{
-#ifdef __AES__
-      init_echo( &ctx_echo, 512 );
-      update_final_echo ( &ctx_echo,(BitSequence*)hash,
-                              (const BitSequence*)hash, 512 );
-#else
-		sph_echo512_init(&ctx_echo);
-		sph_echo512(&ctx_echo, hash, 64);
-		sph_echo512_close(&ctx_echo, hash);
-#endif
-	} else {
-      init_luffa( &ctx_luffa, 512 );
-      update_and_final_luffa( &ctx_luffa, (BitSequence*)hash,
-                                    (const BitSequence*)hash, 64 );
-	}
-
-	sph_shabal512_init(&ctx_shabal);
-	sph_shabal512(&ctx_shabal, hash, 64);
-	sph_shabal512_close(&ctx_shabal, hash);
-
-   sph_skein512_init( &ctx_skein );
-   sph_skein512( &ctx_skein, hash, 64 );
-   sph_skein512_close( &ctx_skein, hash );
-
-	if (hash[0] & 0x8)
-	{
-		sph_shabal512_init(&ctx_shabal);
-		sph_shabal512(&ctx_shabal, hash, 64);
-		sph_shabal512_close(&ctx_shabal, hash);
-	} else {
-		sph_whirlpool_init(&ctx_whirlpool);
-		sph_whirlpool(&ctx_whirlpool, hash, 64);
-		sph_whirlpool_close(&ctx_whirlpool, hash);
-	}
-
-	sph_shabal512_init(&ctx_shabal);
-	sph_shabal512(&ctx_shabal, hash, 64);
-	sph_shabal512_close(&ctx_shabal, hash);
-
-	if (hash[0] & 0x8)
-	{
-		sph_hamsi512_init(&ctx_hamsi);
-		sph_hamsi512(&ctx_hamsi, hash, 64);
-		sph_hamsi512_close(&ctx_hamsi, hash);
-	} else {
-      init_luffa( &ctx_luffa, 512 );
-      update_and_final_luffa( &ctx_luffa, (BitSequence*)hash,
-                                    (const BitSequence*)hash, 64 );
-	}
-
-	memcpy(output, hash, 32);
-}
-
-int scanhash_bastion( struct work *work, uint32_t max_nonce,
-      uint64_t *hashes_done, struct thr_info *mythr)
-{
-	uint32_t _ALIGN(64) hash32[8];
-	uint32_t _ALIGN(64) 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[19];
-
-	uint32_t n = first_nonce;
-
-	for (int i=0; i < 19; i++) 
-		be32enc(&endiandata[i], pdata[i]);
-
-	do {
-		be32enc(&endiandata[19], n);
-		bastionhash(hash32, endiandata);
-		if (hash32[7] < Htarg && fulltest(hash32, ptarget)) {
-			pdata[19] = n;
-         submit_solution( work, hash32, mythr );
-		}
-		n++;
-
-	} while (n < max_nonce && !work_restart[thr_id].restart);
-
-	*hashes_done = n - first_nonce + 1;
-	pdata[19] = n;
-
-	return 0;
-}
-
-bool register_bastion_algo( algo_gate_t* gate )
-{
-  gate->optimizations = SSE2_OPT | AES_OPT;
-  gate->scanhash = (void*)&scanhash_bastion;
-  gate->hash     = (void*)&bastionhash;
-  return true;
-};
-
--- a/algo/heavy/heavy.c
+++ b/algo/heavy/heavy.c
@@ -1,111 +0,0 @@
-#include <string.h>
-#include <openssl/sha.h>
-#include <stdint.h>
-
-#include "algo-gate-api.h"
-#include "sph_hefty1.h"
-#include "algo/keccak/sph_keccak.h"
-#include "algo/blake/sph_blake.h"
-#include "algo/groestl/sph_groestl.h"
-
-/* Combines top 64-bits from each hash into a single hash */
-static void combine_hashes(uint32_t *out, uint32_t *hash1, uint32_t *hash2, uint32_t *hash3, uint32_t *hash4)
-{
-    uint32_t *hash[4] = { hash1, hash2, hash3, hash4 };
-
-    /* Transpose first 64 bits of each hash into out */
-    memset(out, 0, 32);
-    int bits = 0;
-    for (unsigned int i = 7; i >= 6; i--) {
-        for (uint32_t mask = 0x80000000; mask; mask >>= 1) {
-            for (unsigned int k = 0; k < 4; k++) {
-                out[(255 - bits)/32] <<= 1;
-                if ((hash[k][i] & mask) != 0)
-                    out[(255 - bits)/32] |= 1;
-                bits++;
-            }
-        }
-    }
-}
-
-extern void heavyhash(unsigned char* output, const unsigned char* input, int len)
-{
-    unsigned char hash1[32];
-    HEFTY1(input, len, hash1);
-
-// HEFTY1 is new, so take an extra security measure to eliminate
-//     * the possiblity of collisions:
-//     *
-//     *     Hash(x) = SHA256(x + HEFTY1(x))
-//     *
-//     * N.B. '+' is concatenation.
-//
-    unsigned char hash2[32];;
-    SHA256_CTX ctx;
-    SHA256_Init(&ctx);
-    SHA256_Update(&ctx, input, len);
-    SHA256_Update(&ctx, hash1, sizeof(hash1));
-    SHA256_Final(hash2, &ctx);
-
-//   * Additional security: Do not rely on a single cryptographic hash
-//     * function.  Instead, combine the outputs of 4 of the most secure
-//     * cryptographic hash functions-- SHA256, KECCAK512, GROESTL512
-//     * and BLAKE512.
-
-
-    uint32_t hash3[16];
-    sph_keccak512_context keccakCtx;
-    sph_keccak512_init(&keccakCtx);
-    sph_keccak512(&keccakCtx, input, len);
-    sph_keccak512(&keccakCtx, hash1, sizeof(hash1));
-    sph_keccak512_close(&keccakCtx, (void *)&hash3);
-
-    uint32_t hash4[16];
-    sph_groestl512_context groestlCtx;
-    sph_groestl512_init(&groestlCtx);
-    sph_groestl512(&groestlCtx, input, len);
-    sph_groestl512(&groestlCtx, hash1, sizeof(hash1));
-    sph_groestl512_close(&groestlCtx, (void *)&hash4);
-
-    uint32_t hash5[16];
-    sph_blake512_context blakeCtx;
-    sph_blake512_init(&blakeCtx);
-    sph_blake512(&blakeCtx, input, len);
-    sph_blake512(&blakeCtx, (unsigned char *)&hash1, sizeof(hash1));
-    sph_blake512_close(&blakeCtx, (void *)&hash5);
-
-    uint32_t *final = (uint32_t *)output;
-    combine_hashes(final, (uint32_t *)hash2, hash3, hash4, hash5);
-
-}
-
-int scanhash_heavy( uint32_t *pdata, const uint32_t *ptarget,
-            uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr)
-{
-    uint32_t hash[8];
-    uint32_t start_nonce = pdata[19];
-    int thr_id = mythr->id;  // thr_id arg is deprecated
-    
-    do {
-        heavyhash((unsigned char *)hash, (unsigned char *)pdata, 80);
-    
-        if (hash[7] <= ptarget[7]) {
-            if (fulltest(hash, ptarget)) {
-                *hashes_done = pdata[19] - start_nonce;
-                return 1;
-                break;
-            }
-        }
-        pdata[19]++;
-    } while (pdata[19] < max_nonce && !work_restart[thr_id].restart);
-    *hashes_done = pdata[19] - start_nonce;
-    return 0;
-}
-
-bool register_heavy_algo( algo_gate_t* gate )
-{
-    gate->scanhash = (void*)&scanhash_heavy;
-    gate->hash     = (void*)&heavyhash;
-    return true;
-};
-
--- a/algo/hodl/hodl-gate.c
+++ b/algo/hodl/hodl-gate.c
@@ -99,9 +99,13 @@ void hodl_build_block_header( struct work* g_work, uint32_t version,
 // called only by thread 0, saves a backup of g_work
 void hodl_get_new_work( struct work* work, struct work* g_work)
 {
-     work_free( &hodl_work );
-     work_copy( &hodl_work, g_work );
-     hodl_work.data[ algo_gate.nonce_index ] = ( clock() + rand() ) % 9999;
+   pthread_rwlock_rdlock( &g_work_lock );
+
+   work_free( &hodl_work );
+   work_copy( &hodl_work, g_work );
+   hodl_work.data[ algo_gate.nonce_index ] = ( clock() + rand() ) % 9999;
+
+   pthread_rwlock_unlock( &g_work_lock );
 }

 json_t *hodl_longpoll_rpc_call( CURL *curl, int *err, char* lp_url )
@@ -144,7 +148,7 @@ int hodl_scanhash( struct work* work, uint32_t max_nonce,
 #if defined(__AES__)
  GenRandomGarbage( (CacheEntry*)hodl_scratchbuf, work->data, mythr->id );
  pthread_barrier_wait( &hodl_barrier );
-  return scanhash_hodl_wolf( work, max_nonce, hashes_done, thr_info );
+  return scanhash_hodl_wolf( work, max_nonce, hashes_done, mythr );
 #endif
  return false;
 }
@@ -155,11 +159,10 @@ bool register_hodl_algo( algo_gate_t* gate )
  applog( LOG_ERR, "Only CPUs with AES are supported, use legacy version.");
  return false;
 #endif
-//  if ( TOTAL_CHUNKS % opt_n_threads )
-//  {
-//     applog(LOG_ERR,"Thread count must be power of 2.");
-//     return false;
-//  }
+
+  if ( GARBAGE_SIZE % opt_n_threads )
+     applog( LOG_WARNING,"WARNING: Thread count must be power of 2. Miner may crash or produce invalid hash!" );
+
  pthread_barrier_init( &hodl_barrier, NULL, opt_n_threads );
  gate->optimizations         = SSE42_OPT | AES_OPT | AVX2_OPT;
  gate->scanhash              = (void*)&hodl_scanhash;
@@ -171,7 +174,7 @@ bool register_hodl_algo( algo_gate_t* gate )
  gate->resync_threads        = (void*)&hodl_resync_threads;
  gate->do_this_thread        = (void*)&hodl_do_this_thread;
  gate->work_cmp_size         = 76;
-  hodl_scratchbuf = (unsigned char*)malloc( 1 << 30 );
+  hodl_scratchbuf = (unsigned char*)_mm_malloc( 1 << 30, 64 );
  allow_getwork = false;
  opt_target_factor = 8388608.0;
  return ( hodl_scratchbuf != NULL );
--- a/algo/hodl/hodl-wolf.c
+++ b/algo/hodl/hodl-wolf.c
@@ -70,7 +70,7 @@ int scanhash_hodl_wolf( struct work* work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    int threadNumber = mythr->id;
    CacheEntry *Garbage = (CacheEntry*)hodl_scratchbuf;
-    CacheEntry Cache[AES_PARALLEL_N];
+    CacheEntry Cache[AES_PARALLEL_N] __attribute__ ((aligned (64)));
    __m128i* data[AES_PARALLEL_N];
    const __m128i* next[AES_PARALLEL_N];
    uint32_t CollisionCount = 0;
@@ -129,9 +129,10 @@ int scanhash_hodl_wolf( struct work* work, uint32_t max_nonce,
 	      if( FinalPoW[7] <= ptarget[7] )
 	      {
 	          pdata[20] = swab32( BlockHdr[20] );
-		  pdata[21] = swab32( BlockHdr[21] );
-		  *hashes_done = CollisionCount;
-		  return(1);
+             pdata[21] = swab32( BlockHdr[21] );
+		       *hashes_done = CollisionCount;
+             submit_solution( work, FinalPoW, mythr );
+             return(0);
 	      }
 	   }
 	}
@@ -198,7 +199,8 @@ int scanhash_hodl_wolf( struct work* work, uint32_t max_nonce,
                  pdata[20] = swab32( BlockHdr[20] );
                  pdata[21] = swab32( BlockHdr[21] );
                  *hashes_done = CollisionCount;
-                  return(1);
+                  submit_solution( work, FinalPoW, mythr );
+                  return(0);
              }
           }
        }
--- a/algo/jh/jha-4way.c
+++ b/algo/jh/jha-4way.c
@@ -129,7 +129,7 @@ int scanhash_jha_4way( struct work *work, uint32_t max_nonce,
                 if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
                 {
                    pdata[19] = n+i;
-                    submit_lane_solution( work, lane_hash, mythr, i );
+                    submit_solution( work, lane_hash, mythr );
                 }
              }
              n += 4;
--- a/algo/jh/jha.c
+++ b/algo/jh/jha.c
@@ -1,4 +1,7 @@
 #include "jha-gate.h"
+
+#if !defined(JHA_8WAY) && !defined(JHA_4WAY)
+
 #include <stdlib.h>
 #include <stdint.h>
 #include <string.h>
@@ -133,3 +136,4 @@ int scanhash_jha( struct work *work, uint32_t max_nonce,
 	return 0;
 }

+#endif
--- a/algo/keccak/keccak-4way.c
+++ b/algo/keccak/keccak-4way.c
@@ -28,30 +28,32 @@ int scanhash_keccak_8way( struct work *work, uint32_t max_nonce,
   const uint32_t first_nonce = pdata[19];
   __m512i  *noncev = (__m512i*)vdata + 9;   // aligned
   const uint32_t Htarg = ptarget[7];
-   int thr_id = mythr->id;  
+   const int thr_id = mythr->id;  
+   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 {
-       *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 );
-
      keccakhash_8way( hash, vdata );

      for ( int lane = 0; lane < 8; lane++ )
-      if ( hash7[ lane<<1 ] <= Htarg ) 
+      if unlikely( hash7[ lane<<1 ] <= Htarg && !bench ) 
      {
          extr_lane_8x64( lane_hash, hash, lane, 256 );
-          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+          if ( valid_hash( lane_hash, ptarget ) )
          {
-              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              pdata[19] = bswap_32( n + lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
+      *noncev = _mm512_add_epi32( *noncev,
+                                  m512_const1_64( 0x0000000800000000 ) );
      n += 8;

   } while ( (n < max_nonce-8) && !work_restart[thr_id].restart);
-
+   pdata[19] = n;
   *hashes_done = n - first_nonce + 1;
   return 0;
 }
@@ -79,29 +81,30 @@ int scanhash_keccak_4way( struct work *work, uint32_t max_nonce,
   const uint32_t first_nonce = pdata[19];
   __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
   const uint32_t Htarg = ptarget[7];
-   int thr_id = mythr->id;
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+   *noncev = mm256_intrlv_blend_32(
+                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
   do {
-       *noncev = mm256_intrlv_blend_32( mm256_bswap_32(
-                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
-	
      keccakhash_4way( hash, vdata );

      for ( int lane = 0; lane < 4; lane++ )
-      if ( hash7[ lane<<1 ] <= Htarg )
+      if unlikely( hash7[ lane<<1 ] <= Htarg && !bench )
      {
          extr_lane_4x64( lane_hash, hash, lane, 256 );
-          if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+          if ( valid_hash( lane_hash, ptarget ))
          {
-              pdata[19] = n + lane;
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              pdata[19] = bswap_32( n + lane );
+              submit_solution( work, lane_hash, mythr );
          }
      }
+      *noncev = _mm256_add_epi32( *noncev,
+                                  m256_const1_64( 0x0000000400000000 ) );
      n += 4;
-
   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
-
+   pdata[19] = n;
   *hashes_done = n - first_nonce + 1;
   return 0;
 }
--- a/algo/keccak/keccak-gate.c
+++ b/algo/keccak/keccak-gate.c
@@ -1,5 +1,9 @@
 #include "keccak-gate.h"
+#include "sph_keccak.h"

+int hard_coded_eb = 1;
+
+// KECCAK

 bool register_keccak_algo( algo_gate_t* gate )
 {
@@ -19,6 +23,8 @@ bool register_keccak_algo( algo_gate_t* gate )
  return true;
 };

+// KECCAKC
+
 bool register_keccakc_algo( algo_gate_t* gate )
 {
  gate->optimizations = AVX2_OPT | AVX512_OPT;
@@ -37,3 +43,50 @@ bool register_keccakc_algo( algo_gate_t* gate )
  return true;
 };

+// SHA3D
+
+void sha3d( void *state, const void *input, int len )
+{
+	uint32_t _ALIGN(64) buffer[16], hash[16];
+	sph_keccak_context ctx_keccak;
+
+	sph_keccak256_init( &ctx_keccak );
+	sph_keccak256 ( &ctx_keccak, input, len );
+	sph_keccak256_close( &ctx_keccak, (void*) buffer );
+
+   sph_keccak256_init( &ctx_keccak );
+	sph_keccak256 ( &ctx_keccak, buffer, 32 );
+	sph_keccak256_close( &ctx_keccak, (void*) hash );
+
+	memcpy(state, hash, 32);
+}
+
+void sha3d_gen_merkle_root( char* merkle_root, struct stratum_ctx* sctx )
+{
+  sha3d( merkle_root, sctx->job.coinbase, (int) sctx->job.coinbase_size );
+  for ( int i = 0; i < sctx->job.merkle_count; i++ )
+  {
+     memcpy( merkle_root + 32, sctx->job.merkle[i], 32 );
+     sha256d( merkle_root, merkle_root, 64 );
+  }
+}
+
+bool register_sha3d_algo( algo_gate_t* gate )
+{
+  hard_coded_eb = 6;
+//  opt_extranonce = false;
+  gate->optimizations = AVX2_OPT | AVX512_OPT;
+  gate->gen_merkle_root = (void*)&sha3d_gen_merkle_root;
+#if defined (KECCAK_8WAY)
+  gate->scanhash  = (void*)&scanhash_sha3d_8way;
+  gate->hash      = (void*)&sha3d_hash_8way;
+#elif defined (KECCAK_4WAY)
+  gate->scanhash  = (void*)&scanhash_sha3d_4way;
+  gate->hash      = (void*)&sha3d_hash_4way;
+#else
+  gate->scanhash  = (void*)&scanhash_sha3d;
+  gate->hash      = (void*)&sha3d_hash;
+#endif
+  return true;
+};
+
--- a/algo/keccak/keccak-gate.h
+++ b/algo/keccak/keccak-gate.h
@@ -10,24 +10,37 @@
  #define KECCAK_4WAY 1
 #endif

+extern int hard_coded_eb;
+
 #if defined(KECCAK_8WAY)

 void keccakhash_8way( void *state, const void *input );
 int scanhash_keccak_8way( struct work *work, uint32_t max_nonce,
                         uint64_t *hashes_done, struct thr_info *mythr );

+void sha3d_hash_8way( void *state, const void *input );
+int scanhash_sha3d_8way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+
 #elif defined(KECCAK_4WAY)

 void keccakhash_4way( void *state, const void *input );
 int scanhash_keccak_4way( struct work *work, uint32_t max_nonce,
                         uint64_t *hashes_done, struct thr_info *mythr );

+void sha3d_hash_4way( void *state, const void *input );
+int scanhash_sha3d_4way( struct work *work, uint32_t max_nonce,
+                         uint64_t *hashes_done, struct thr_info *mythr );
+
 #else

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

-#endif
+void sha3d_hash( void *state, const void *input );
+int scanhash_sha3d( struct work *work, uint32_t max_nonce,
+                    uint64_t *hashes_done, struct thr_info *mythr );

 #endif
+#endif
--- a/algo/keccak/keccak-hash-4way.c
+++ b/algo/keccak/keccak-hash-4way.c
@@ -1,6 +1,9 @@
 #include <stddef.h>
 #include <stdint.h>
 #include "keccak-hash-4way.h"
+#include "keccak-gate.h"
+
+#if defined(__AVX2__)

 static const uint64_t RC[] = {
        0x0000000000000001, 0x0000000000008082,
@@ -168,7 +171,7 @@ static void keccak64_8way_close( keccak64_ctx_m512i *kc, void *dst,
    size_t j;
    size_t m512_len = byte_len >> 3;

-    eb = 0x100  >> 8;
+    eb = hard_coded_eb;
    if ( kc->ptr == (lim - 8) )
    {
        const uint64_t t = eb | 0x8000000000000000;
@@ -238,7 +241,7 @@ keccak512_8way_close(void *cc, void *dst)

 #endif  // AVX512

-#if defined(__AVX2__)
+// AVX2

 #define INPUT_BUF(size)   do { \
    size_t j; \
@@ -349,7 +352,7 @@ static void keccak64_close( keccak64_ctx_m256i *kc, void *dst, size_t byte_len,
    size_t j;
    size_t m256_len = byte_len >> 3;

-    eb = 0x100  >> 8;
+    eb = hard_coded_eb;
    if ( kc->ptr == (lim - 8) )
    {
        const uint64_t t = eb | 0x8000000000000000;
--- a/algo/keccak/keccak.c
+++ b/algo/keccak/keccak.c
@@ -1,4 +1,6 @@
-#include "algo-gate-api.h"
+#include "keccak-gate.h"
+
+#if !defined(KECCAK_8WAY) && !defined(KECCAK_4WAY)

 #include <stdlib.h>
 #include <string.h>
@@ -18,36 +20,35 @@ void keccakhash(void *state, const void *input)
 	memcpy(state, hash, 32);
 }

-int scanhash_keccak( struct work *work,
-	uint32_t max_nonce, uint64_t *hashes_done, struct thr_info *mythr )
+int scanhash_keccak( 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(64) hash64[8];
+   uint32_t _ALIGN(64) endiandata[32];
+   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;
+   const int thr_id = mythr->id;

-	uint32_t _ALIGN(32) hash64[8];
-	uint32_t endiandata[32];
+   for ( int i=0; i < 19; i++ )
+      be32enc( &endiandata[i], pdata[i] );

-        for (int i=0; i < 19; i++) 
-                be32enc(&endiandata[i], pdata[i]);
+   do {
+      be32enc( &endiandata[19], n );
+      keccakhash( hash64, endiandata );
+      if ( valid_hash( hash64, ptarget ) && !opt_benchmark )
+      {
+         pdata[19] = n;
+         submit_solution( work, hash64, mythr );
+      }
+      n++;
+   } while ( n < last_nonce && !work_restart[thr_id].restart );

-	do {
-	
-		pdata[19] = ++n;
-		be32enc(&endiandata[19], n); 
-		keccakhash(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;
+   *hashes_done = n - first_nonce;
+   pdata[19] = n;
+   return 0;
 }

+#endif
--- a/algo/keccak/sha3d-4way.c
+++ b/algo/keccak/sha3d-4way.c
@@ -0,0 +1,126 @@
+#include "keccak-gate.h"
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include "sph_keccak.h"
+#include "keccak-hash-4way.h"
+
+#if defined(KECCAK_8WAY)
+
+void sha3d_hash_8way(void *state, const void *input)
+{
+    uint32_t buffer[16*8] __attribute__ ((aligned (128)));
+    keccak256_8way_context ctx;
+
+    keccak256_8way_init( &ctx );
+    keccak256_8way_update( &ctx, input, 80 );
+    keccak256_8way_close( &ctx, buffer );
+
+    keccak256_8way_init( &ctx );
+    keccak256_8way_update( &ctx, buffer, 32 );
+    keccak256_8way_close( &ctx, state );
+}
+
+int scanhash_sha3d_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];
+   const int thr_id = mythr->id;  
+   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 {
+      sha3d_hash_8way( hash, vdata );
+
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg && !bench ) )
+      {
+          extr_lane_8x64( lane_hash, hash, lane, 256 );
+          if ( 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;
+}
+
+#elif defined(KECCAK_4WAY)
+
+void sha3d_hash_4way(void *state, const void *input)
+{
+    uint32_t buffer[16*4] __attribute__ ((aligned (64)));
+    keccak256_4way_context ctx;
+
+    keccak256_4way_init( &ctx );
+    keccak256_4way_update( &ctx, input, 80 );
+    keccak256_4way_close( &ctx, buffer );
+
+    keccak256_4way_init( &ctx );
+    keccak256_4way_update( &ctx, buffer, 32 );
+    keccak256_4way_close( &ctx, state );
+}
+
+int scanhash_sha3d_4way( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t vdata[24*4] __attribute__ ((aligned (64)));
+   uint32_t hash[16*4] __attribute__ ((aligned (32)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[25]);   // 3*8+1
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   uint32_t n = pdata[19];
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 4;
+   __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
+   const uint32_t Htarg = ptarget[7];
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+
+   mm256_bswap32_intrlv80_4x64( vdata, pdata );
+   *noncev = mm256_intrlv_blend_32( 
+                _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
+   do {
+      sha3d_hash_4way( hash, vdata );
+
+      for ( int lane = 0; lane < 4; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg && !bench ) )
+      {
+          extr_lane_4x64( lane_hash, hash, 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;
+}
+
+#endif
--- a/algo/keccak/sha3d.c
+++ b/algo/keccak/sha3d.c
@@ -0,0 +1,54 @@
+#include "keccak-gate.h"
+
+#if !defined(KECCAK_8WAY) && !defined(KECCAK_4WAY)
+
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include "sph_keccak.h"
+
+void sha3d_hash(void *state, const void *input)
+{
+    uint32_t buffer[16];
+    sph_keccak256_context ctx_keccak;
+   
+    sph_keccak256_init( &ctx_keccak );
+    sph_keccak256 ( &ctx_keccak, input, 80 );
+    sph_keccak256_close( &ctx_keccak, buffer );
+    sph_keccak256_init( &ctx_keccak );
+    sph_keccak256 ( &ctx_keccak, buffer, 32 );
+    sph_keccak256_close( &ctx_keccak, state );
+}
+
+int scanhash_sha3d( struct work *work, uint32_t max_nonce,
+                    uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t _ALIGN(64) hash64[8];
+   uint32_t _ALIGN(64) endiandata[32];
+   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;
+   const int thr_id = mythr->id;
+
+   for ( int i=0; i < 19; i++ ) 
+      be32enc( &endiandata[i], pdata[i] );
+
+	do {
+		be32enc( &endiandata[19], n ); 
+		sha3d_hash( hash64, endiandata );
+      if ( valid_hash( hash64, ptarget ) && !opt_benchmark )
+      {
+         pdata[19] = n;
+         submit_solution( work, hash64, mythr );
+		}
+      n++;
+   } while ( n < last_nonce && !work_restart[thr_id].restart );
+	
+	*hashes_done = n - first_nonce;
+	pdata[19] = n;
+	return 0;
+}
+
+#endif
--- a/algo/keccak/sph_keccak.c
+++ b/algo/keccak/sph_keccak.c
@@ -32,8 +32,8 @@

 #include <stddef.h>
 #include <string.h>
-
 #include "sph_keccak.h"
+#include "keccak-gate.h"

 #ifdef __cplusplus
 extern "C"{
@@ -1616,7 +1616,7 @@ keccak_core(sph_keccak_context *kc, const void *data, size_t len, size_t lim)
 		} u; \
 		size_t j; \
 \
-		eb = (0x100 | (ub & 0xFF)) >> (8 - n); \
+		eb = hard_coded_eb; \
 		if (kc->ptr == (lim - 1)) { \
 			if (n == 7) { \
 				u.tmp[0] = eb; \
--- a/algo/luffa/luffa.c
+++ b/algo/luffa/luffa.c
@@ -1,63 +0,0 @@
-#include "algo-gate-api.h"
-
-#include <stdlib.h>
-#include <stdint.h>
-#include <string.h>
-#include <stdio.h>
-
-#include "sph_luffa.h"
-
-void luffahash(void *output, const void *input)
-{
-	unsigned char _ALIGN(128) hash[64];
-	sph_luffa512_context ctx_luffa;
-
-	sph_luffa512_init(&ctx_luffa);
-	sph_luffa512 (&ctx_luffa, input, 80);
-	sph_luffa512_close(&ctx_luffa, (void*) hash);
-
-	memcpy(output, hash, 32);
-}
-
-int scanhash_luffa(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 _ALIGN(64) hash64[8];
-	uint32_t _ALIGN(64) endiandata[20];
-
-	const uint32_t Htarg = ptarget[7];
-	const uint32_t first_nonce = pdata[19];
-
-	uint32_t n = first_nonce;
-
-        for (int i=0; i < 19; i++) 
-                be32enc(&endiandata[i], pdata[i]);
-
-	do {
-		be32enc(&endiandata[19], n);
-		luffahash(hash64, endiandata);
-		if (hash64[7] < Htarg && fulltest(hash64, ptarget)) {
-			*hashes_done = n - first_nonce + 1;
-			pdata[19] = n;
-			return true;
-		}
-		n++;
-
-	} while (n < max_nonce && !work_restart[thr_id].restart);
-
-	*hashes_done = n - first_nonce + 1;
-	pdata[19] = n;
-
-	return 0;
-}
-
-bool register_luffa_algo( algo_gate_t* gate )
-{
-    gate->scanhash = (void*)&scanhash_luffa;
-    gate->hash     = (void*)&luffahash;
-    return true;
-};
-
--- a/algo/luffa/luffa_for_sse2.c
+++ b/algo/luffa/luffa_for_sse2.c
@@ -344,17 +344,12 @@ HashReturn update_and_final_luffa( hashState_luffa *state, BitSequence* output,

    // 16 byte partial block exists for 80 byte len
    if ( state->rembytes  )
-    {
-      // padding of partial block
-      rnd512( state, _mm_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 ),
+       // padding of partial block
+       rnd512( state, m128_const_64( 0, 0x80000000 ),
                      mm128_bswap_32( cast_m128i( data ) ) );
-    }
    else
-    {
-      // empty pad block
-     rnd512( state, _mm_setzero_si128(), 
-                       _mm_set_epi8( 0,0,0,0, 0,0,0,0, 0,0,0,0, 0x80,0,0,0 ) );
-    }
+       // empty pad block
+       rnd512( state, m128_zero, m128_const_64( 0, 0x80000000 ) );

    finalization512( state, (uint32*) output );
    if ( state->hashbitlen > 512 )
@@ -363,6 +358,56 @@ HashReturn update_and_final_luffa( hashState_luffa *state, BitSequence* output,
    return SUCCESS;
 }

+
+int luffa_full( hashState_luffa *state, BitSequence* output, int hashbitlen,
+              const BitSequence* data, size_t inlen )
+{
+// Optimized for integrals of 16 bytes, good for 64 and 80 byte len
+    int i;
+    state->hashbitlen = hashbitlen;
+    /* set the lower 32 bits to '1' */
+    MASK= _mm_set_epi32(0x00000000, 0x00000000, 0x00000000, 0xffffffff);
+    /* set all bits to '1' */
+    ALLONE = _mm_set_epi32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff);
+    /* set the 32-bit round constant values to the 128-bit data field */
+    for ( i=0; i<32; i++ )
+        CNS128[i] = _mm_load_si128( (__m128i*)&CNS_INIT[i*4] );
+    for ( i=0; i<10; i++ )
+    state->chainv[i] = _mm_load_si128( (__m128i*)&IV[i*4] );
+    memset(state->buffer, 0, sizeof state->buffer );
+
+    // update
+
+    int blocks = (int)( inlen / 32 );
+    state->rembytes = inlen % 32;
+
+    // full blocks
+    for ( i = 0; i < blocks; i++ )
+    {
+       rnd512( state, mm128_bswap_32( casti_m128i( data, 1 ) ),
+                      mm128_bswap_32( casti_m128i( data, 0 ) ) );
+       data += MSG_BLOCK_BYTE_LEN;
+    }
+
+    // final
+
+    // 16 byte partial block exists for 80 byte len
+    if ( state->rembytes  )
+       // padding of partial block
+       rnd512( state, m128_const_64( 0, 0x80000000 ),
+                      mm128_bswap_32( cast_m128i( data ) ) );
+    else
+       // empty pad block
+       rnd512( state, m128_zero, m128_const_64( 0, 0x80000000 ) );
+
+    finalization512( state, (uint32*) output );
+    if ( state->hashbitlen > 512 )
+        finalization512( state, (uint32*)( output+128 ) );
+
+    return SUCCESS;
+}
+
+
 /***************************************************/
 /* Round function         */
 /* state: hash context    */
--- a/algo/luffa/luffa_for_sse2.h
+++ b/algo/luffa/luffa_for_sse2.h
@@ -1,3 +1,6 @@
+#if !defined(LUFFA_FOR_SSE2_H__)
+#define LUFFA_FOR_SSE2_H__ 1
+
 /*
 * luffa_for_sse2.h
 * Version 2.0 (Sep 15th 2009)
@@ -48,8 +51,6 @@
 typedef struct {
    uint32 buffer[8] __attribute((aligned(32)));
    __m128i chainv[10] __attribute((aligned(32)));   /* Chaining values */
-//    uint64 bitlen[2]; /* Message length in bits */
-//    uint32 rembitlen; /* Length of buffer data to be hashed */
    int hashbitlen;
    int rembytes;
 } hashState_luffa;
@@ -65,5 +66,6 @@ HashReturn final_luffa( hashState_luffa *state, BitSequence *hashval );
 HashReturn update_and_final_luffa( hashState_luffa *state, BitSequence* output,
                                   const BitSequence* data, size_t inlen );

-
-
+int luffa_full( hashState_luffa *state, BitSequence* output, int hashbitlen,
+                                   const BitSequence* data, size_t inlen );
+#endif   // LUFFA_FOR_SSE2_H___
--- a/algo/lyra2/allium-4way.c
+++ b/algo/lyra2/allium-4way.c
@@ -115,9 +115,8 @@ void allium_16way_hash( void *state, const void *input )
   intrlv_4x128( vhashA, hash0, hash1, hash2, hash3, 256 );
   intrlv_4x128( vhashB, hash4, hash5, hash6, hash7, 256 );

-   cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
+   cube_4way_full( &ctx.cube, vhashA, 256, vhashA, 32 );
+   cube_4way_full( &ctx.cube, vhashB, 256, vhashB, 32 );

   dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
   dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
@@ -125,10 +124,8 @@ void allium_16way_hash( void *state, const void *input )
   intrlv_4x128( vhashA, hash8, hash9, hash10, hash11, 256 );
   intrlv_4x128( vhashB, hash12, hash13, hash14, hash15, 256 );

-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
+   cube_4way_full( &ctx.cube, vhashA, 256, vhashA, 32 );
+   cube_4way_full( &ctx.cube, vhashB, 256, vhashB, 32 );

   dintrlv_4x128( hash8, hash9, hash10, hash11, vhashA, 256 );
   dintrlv_4x128( hash12, hash13, hash14, hash15, vhashB, 256 );
@@ -169,7 +166,6 @@ void allium_16way_hash( void *state, const void *input )
   skein256_8way_update( &ctx.skein, vhashB, 32 );
   skein256_8way_close( &ctx.skein, vhashB );

-
   dintrlv_8x64( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
                 vhashA, 256 );
   dintrlv_8x64( hash8, hash9, hash10, hash11, hash12, hash13, hash14, hash15,
@@ -179,77 +175,43 @@ void allium_16way_hash( void *state, const void *input )

   intrlv_4x128( vhash, hash0, hash1, hash2, hash3, 256 );

-   groestl256_4way_update_close( &ctx.groestl, vhash, vhash, 256 );
+   groestl256_4way_full( &ctx.groestl, vhash, vhash, 256 );

   dintrlv_4x128( state, state+32, state+64, state+96, vhash, 256 );
   intrlv_4x128( vhash, hash4, hash5, hash6, hash7, 256 );

-   groestl256_4way_init( &ctx.groestl, 32 );
-   groestl256_4way_update_close( &ctx.groestl, vhash, vhash, 256 );
+   groestl256_4way_full( &ctx.groestl, vhash, vhash, 256 );
   
   dintrlv_4x128( state+128, state+160, state+192, state+224, vhash, 256 );
   intrlv_4x128( vhash, hash8, hash9, hash10, hash11, 256 );

-   groestl256_4way_init( &ctx.groestl, 32 );
-   groestl256_4way_update_close( &ctx.groestl, vhash, vhash, 256 );
+   groestl256_4way_full( &ctx.groestl, vhash, vhash, 256 );

   dintrlv_4x128( state+256, state+288, state+320, state+352, vhash, 256 );
   intrlv_4x128( vhash, hash12, hash13, hash14, hash15, 256 );

-   groestl256_4way_init( &ctx.groestl, 32 );
-   groestl256_4way_update_close( &ctx.groestl, vhash, vhash, 256 );
+   groestl256_4way_full( &ctx.groestl, vhash, vhash, 256 );
 
   dintrlv_4x128( state+384, state+416, state+448, state+480, vhash, 256 );
   
 #else

-   update_and_final_groestl256( &ctx.groestl, state, hash0, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+32, hash1, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+64, hash2, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+96, hash3, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+128, hash4, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+160, hash5, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+192, hash6, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+224, hash7, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+256, hash8, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+288, hash9, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+320, hash10, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+352, hash11, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+384, hash12, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+416, hash13, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+448, hash14, 256 );
-   memcpy( &ctx.groestl, &allium_16way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+480, hash15, 256 );
-
+   groestl256_full( &ctx.groestl, state,     hash0,  256 );
+   groestl256_full( &ctx.groestl, state+32,  hash1,  256 );
+   groestl256_full( &ctx.groestl, state+64,  hash2,  256 );
+   groestl256_full( &ctx.groestl, state+96,  hash3,  256 );
+   groestl256_full( &ctx.groestl, state+128, hash4,  256 );
+   groestl256_full( &ctx.groestl, state+160, hash5,  256 );
+   groestl256_full( &ctx.groestl, state+192, hash6,  256 );
+   groestl256_full( &ctx.groestl, state+224, hash7,  256 );
+   groestl256_full( &ctx.groestl, state+256, hash8,  256 );
+   groestl256_full( &ctx.groestl, state+288, hash9,  256 );
+   groestl256_full( &ctx.groestl, state+320, hash10, 256 );
+   groestl256_full( &ctx.groestl, state+352, hash11, 256 );
+   groestl256_full( &ctx.groestl, state+384, hash12, 256 );
+   groestl256_full( &ctx.groestl, state+416, hash13, 256 );
+   groestl256_full( &ctx.groestl, state+448, hash14, 256 );
+   groestl256_full( &ctx.groestl, state+480, hash15, 256 );
 #endif
 }

@@ -263,37 +225,32 @@ int scanhash_allium_16way( struct work *work, uint32_t max_nonce,
   const uint32_t first_nonce = pdata[19];
   uint32_t n = first_nonce;
   const uint32_t last_nonce = max_nonce - 16;
-   const uint32_t Htarg = ptarget[7];
   __m512i  *noncev = (__m512i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )
-      ( (uint32_t*)ptarget )[7] = 0x0000ff;
+   if ( bench ) ( (uint32_t*)ptarget )[7] = 0x0000ff;

   mm512_bswap32_intrlv80_16x32( vdata, pdata );
+   *noncev = _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 );
+
   blake256_16way_init( &allium_16way_ctx.blake );
   blake256_16way_update( &allium_16way_ctx.blake, 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 ) );
-
     allium_16way_hash( hash, vdata );
-     pdata[19] = n;

-     for ( int lane = 0; lane < 16; lane++ ) if ( (hash+(lane<<3))[7] <= Htarg )
+     for ( int lane = 0; lane < 16; lane++ ) 
+     if ( unlikely( valid_hash( hash+(lane<<3), ptarget ) && !bench ) )
     {
-        if ( fulltest( hash+(lane<<3), ptarget ) && !opt_benchmark )
-        {
-           pdata[19] = n + lane;
-           submit_lane_solution( work, hash+(lane<<3), mythr, lane );
-         }
+         pdata[19] = bswap_32( n + lane );
+         submit_solution( work, hash+(lane<<3), mythr );
     }
+     *noncev = _mm512_add_epi32( *noncev, m512_const1_32( 16 ) );
     n += 16;
-   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
-
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart) );
+   pdata[19] = n;
   *hashes_done = n - first_nonce;
   return 0;
 }
@@ -320,18 +277,18 @@ bool init_allium_8way_ctx()
   return true;
 }

-void allium_8way_hash( void *state, const void *input )
+void allium_8way_hash( void *hash, const void *input )
 {
-   uint32_t vhashA[8*8] __attribute__ ((aligned (64)));
-   uint32_t vhashB[8*8] __attribute__ ((aligned (64)));
-   uint32_t hash0[8] __attribute__ ((aligned (32)));
-   uint32_t hash1[8] __attribute__ ((aligned (32)));
-   uint32_t hash2[8] __attribute__ ((aligned (32)));
-   uint32_t hash3[8] __attribute__ ((aligned (32)));
-   uint32_t hash4[8] __attribute__ ((aligned (64)));
-   uint32_t hash5[8] __attribute__ ((aligned (32)));
-   uint32_t hash6[8] __attribute__ ((aligned (32)));
-   uint32_t hash7[8] __attribute__ ((aligned (32)));
+   uint64_t vhashA[4*8] __attribute__ ((aligned (64)));
+   uint64_t vhashB[4*8] __attribute__ ((aligned (64)));
+   uint64_t *hash0 = (uint64_t*)hash;
+   uint64_t *hash1 = (uint64_t*)hash+ 4;
+   uint64_t *hash2 = (uint64_t*)hash+ 8;
+   uint64_t *hash3 = (uint64_t*)hash+12;
+   uint64_t *hash4 = (uint64_t*)hash+16;
+   uint64_t *hash5 = (uint64_t*)hash+20;
+   uint64_t *hash6 = (uint64_t*)hash+24;
+   uint64_t *hash7 = (uint64_t*)hash+28;
   allium_8way_ctx_holder ctx __attribute__ ((aligned (64))); 

   memcpy( &ctx, &allium_8way_ctx, sizeof(allium_8way_ctx) );
@@ -398,69 +355,52 @@ void allium_8way_hash( void *state, const void *input )
   dintrlv_4x64( hash0, hash1, hash2, hash3, vhashA, 256 );
   dintrlv_4x64( hash4, hash5, hash6, hash7, vhashB, 256 );

-   update_and_final_groestl256( &ctx.groestl, state, hash0, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+32, hash1, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+64, hash2, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+96, hash3, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+128, hash4, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+160, hash5, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+192, hash6, 256 );
-   memcpy( &ctx.groestl, &allium_8way_ctx.groestl,
-           sizeof(hashState_groestl256) );
-   update_and_final_groestl256( &ctx.groestl, state+224, hash7, 256 );
+   groestl256_full( &ctx.groestl, hash0, hash0, 256 );
+   groestl256_full( &ctx.groestl, hash1, hash1, 256 );
+   groestl256_full( &ctx.groestl, hash2, hash2, 256 );
+   groestl256_full( &ctx.groestl, hash3, hash3, 256 );
+   groestl256_full( &ctx.groestl, hash4, hash4, 256 );
+   groestl256_full( &ctx.groestl, hash5, hash5, 256 );
+   groestl256_full( &ctx.groestl, hash6, hash6, 256 );
+   groestl256_full( &ctx.groestl, hash7, hash7, 256 );
 }

 int scanhash_allium_8way( struct work *work, uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8*8] __attribute__ ((aligned (64)));
+   uint64_t hash[4*8] __attribute__ ((aligned (64)));
   uint32_t vdata[20*8] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
-   uint32_t *ptarget = work->target;
+   uint64_t *ptarget = (uint64_t*)work->target;
   const uint32_t first_nonce = pdata[19];
   const uint32_t last_nonce = max_nonce - 8;
   uint32_t n = first_nonce;
-   const uint32_t Htarg = ptarget[7];
   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  
-
-   if ( opt_benchmark )
-      ( (uint32_t*)ptarget )[7] = 0x0000ff;
+   const int thr_id = mythr->id;  
+   const bool bench = opt_benchmark;

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
+   *noncev = _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n );
+
   blake256_8way_init( &allium_8way_ctx.blake );
   blake256_8way_update( &allium_8way_ctx.blake, vdata, 64 );

   do {
-     *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
-                                                 n+3, n+2, n+1, n ) );
-
     allium_8way_hash( hash, vdata );
-     pdata[19] = n;

-     for ( int lane = 0; lane < 8; lane++ ) if ( (hash+(lane<<3))[7] <= Htarg )
+     for ( int lane = 0; lane < 8; lane++ )
     {
-        if ( fulltest( hash+(lane<<3), ptarget ) && !opt_benchmark )
+        const uint64_t *lane_hash = hash + (lane<<2);
+        if ( unlikely( valid_hash( lane_hash, ptarget ) && !bench ) )
        {
-           pdata[19] = n + lane;
-           submit_lane_solution( work, hash+(lane<<3), mythr, lane );
-         }
+           pdata[19] = bswap_32( n + lane );
+           submit_solution( work, lane_hash, mythr );
+        }
     }
     n += 8;
-   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
-
+     *noncev = _mm256_add_epi32( *noncev, m256_const1_32( 8 ) );
+   } while ( likely( (n <= last_nonce) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
   *hashes_done = n - first_nonce;
   return 0;
 }
--- a/algo/lyra2/allium.c
+++ b/algo/lyra2/allium.c
@@ -1,4 +1,7 @@
 #include "lyra2-gate.h"
+
+#if !( defined(ALLIUM_16WAY) || defined(ALLIUM_8WAY) || defined(ALLIUM_4WAY) )
+
 #include <memory.h>
 #include "algo/blake/sph_blake.h"
 #include "algo/keccak/sph_keccak.h"
@@ -73,37 +76,35 @@ int scanhash_allium( struct work *work, uint32_t max_nonce,
                     uint64_t *hashes_done, struct thr_info *mythr )
 {
    uint32_t _ALIGN(128) hash[8];
-    uint32_t _ALIGN(128) endiandata[20];
+    uint32_t _ALIGN(128) edata[20];
    uint32_t *pdata = work->data;
    uint32_t *ptarget = work->target;
-
-    const uint32_t Htarg = ptarget[7];
    const uint32_t first_nonce = pdata[19];
    uint32_t nonce = first_nonce;
-    int thr_id = mythr->id;  // thr_id arg is deprecated
+    const int thr_id = mythr->id; 

    if ( opt_benchmark )
        ptarget[7] = 0x3ffff;

    for ( int i = 0; i < 19; i++ )
-        be32enc( &endiandata[i], pdata[i] );
+        edata[i] = bswap_32( pdata[i] );

    sph_blake256_init( &allium_ctx.blake );
-    sph_blake256( &allium_ctx.blake, endiandata, 64 );
+    sph_blake256( &allium_ctx.blake, edata, 64 );

    do {
-        be32enc( &endiandata[19], nonce );
-        allium_hash( hash, endiandata );
-        if ( hash[7] <= Htarg )
-        if ( fulltest( hash, ptarget ) && !opt_benchmark )
+        edata[19] = nonce;
+        allium_hash( hash, edata );
+        if ( valid_hash( hash, ptarget ) && !opt_benchmark )
        {
-            pdata[19] = nonce;
+            pdata[19] = bswap_32( nonce );
            submit_solution( work, hash, mythr );
        }
        nonce++;
    } while ( nonce < max_nonce && !work_restart[thr_id].restart );
    pdata[19] = nonce;
-    *hashes_done = pdata[19] - first_nonce + 1;
+    *hashes_done = pdata[19] - first_nonce;
    return 0;
 }

+#endif
--- a/algo/lyra2/lyra2-gate.c
+++ b/algo/lyra2/lyra2-gate.c
@@ -94,12 +94,12 @@ bool lyra2rev2_thread_init()
   const int64_t ROW_LEN_BYTES = ROW_LEN_INT64 * 8;

   int size = (int64_t)ROW_LEN_BYTES * 4; // nRows;
-#if defined (LYRA2REV2_8WAY)
+#if defined (LYRA2REV2_16WAY)
   l2v2_wholeMatrix = _mm_malloc( 2 * size, 64 );   // 2 way
-   init_lyra2rev2_8way_ctx();;
-#elif defined (LYRA2REV2_4WAY)
+   init_lyra2rev2_16way_ctx();;
+#elif defined (LYRA2REV2_8WAY)
   l2v2_wholeMatrix = _mm_malloc( size, 64 );
-   init_lyra2rev2_4way_ctx();;
+   init_lyra2rev2_8way_ctx();;
 #else
   l2v2_wholeMatrix = _mm_malloc( size, 64 );
   init_lyra2rev2_ctx();
@@ -109,17 +109,17 @@ bool lyra2rev2_thread_init()

 bool register_lyra2rev2_algo( algo_gate_t* gate )
 {
-#if defined (LYRA2REV2_8WAY)
+#if defined (LYRA2REV2_16WAY)
+  gate->scanhash  = (void*)&scanhash_lyra2rev2_16way;
+  gate->hash      = (void*)&lyra2rev2_16way_hash;
+#elif defined (LYRA2REV2_8WAY)
  gate->scanhash  = (void*)&scanhash_lyra2rev2_8way;
  gate->hash      = (void*)&lyra2rev2_8way_hash;
-#elif defined (LYRA2REV2_4WAY)
-  gate->scanhash  = (void*)&scanhash_lyra2rev2_4way;
-  gate->hash      = (void*)&lyra2rev2_4way_hash;
 #else
  gate->scanhash  = (void*)&scanhash_lyra2rev2;
  gate->hash      = (void*)&lyra2rev2_hash;
 #endif
-  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
  gate->miner_thread_init = (void*)&lyra2rev2_thread_init;
  opt_target_factor = 256.0;
  return true;
@@ -194,7 +194,7 @@ bool register_allium_algo( algo_gate_t* gate )

 /////////////////////////////////////////

-bool phi2_has_roots;
+bool phi2_has_roots = false;
 bool phi2_use_roots = false;

 int phi2_get_work_data_size() { return phi2_use_roots ? 144 : 128; }
@@ -215,26 +215,23 @@ void phi2_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
   size_t t;

   algo_gate.gen_merkle_root( merkle_tree, sctx );
-   // Increment extranonce2
-   for ( t = 0; t < sctx->xnonce2_size && !( ++sctx->job.xnonce2[t] ); t++ );
-   // Assemble block header
   algo_gate.build_block_header( g_work, le32dec( sctx->job.version ),
                  (uint32_t*) sctx->job.prevhash, (uint32_t*) merkle_tree,
-                  le32dec( sctx->job.ntime ), le32dec(sctx->job.nbits) );
+                  le32dec( sctx->job.ntime ), le32dec(sctx->job.nbits), NULL );
   for ( t = 0; t < 16; t++ )
      g_work->data[ 20+t ] = ((uint32_t*)sctx->job.extra)[t];
 }

-
 bool register_phi2_algo( algo_gate_t* gate )
 {
-//   init_phi2_ctx();
-   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
+   gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
   gate->get_work_data_size = (void*)&phi2_get_work_data_size;
   gate->decode_extra_data  = (void*)&phi2_decode_extra_data;
   gate->build_extraheader  = (void*)&phi2_build_extraheader;
   opt_target_factor = 256.0;
-#if defined(PHI2_4WAY)
+#if defined(PHI2_8WAY)
+   gate->scanhash           = (void*)&scanhash_phi2_8way;
+#elif defined(PHI2_4WAY)
   gate->scanhash           = (void*)&scanhash_phi2_4way;
 #else
   init_phi2_ctx();
--- a/algo/lyra2/lyra2-gate.h
+++ b/algo/lyra2/lyra2-gate.h
@@ -51,28 +51,29 @@ bool init_lyra2rev3_ctx();
 //////////////////////////////////

 #if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
-  #define LYRA2REV2_8WAY 1
+  #define LYRA2REV2_16WAY 1
 #elif defined(__AVX2__)
-  #define LYRA2REV2_4WAY 1
+  #define LYRA2REV2_8WAY 1
 #endif

 extern __thread uint64_t* l2v2_wholeMatrix;

 bool register_lyra2rev2_algo( algo_gate_t* gate );

-#if defined(LYRA2REV2_8WAY)
+#if defined(LYRA2REV2_16WAY)
+
+void lyra2rev2_16way_hash( void *state, const void *input );
+int scanhash_lyra2rev2_16way( struct work *work, uint32_t max_nonce,
+                             uint64_t *hashes_done, struct thr_info *mythr );
+bool init_lyra2rev2_16way_ctx();
+
+#elif defined(LYRA2REV2_8WAY)

 void lyra2rev2_8way_hash( void *state, const void *input );
 int scanhash_lyra2rev2_8way( struct work *work, uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr );
 bool init_lyra2rev2_8way_ctx();

-#elif defined(LYRA2REV2_4WAY)
-
-void lyra2rev2_4way_hash( void *state, const void *input );
-int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
-                             uint64_t *hashes_done, struct thr_info *mythr );
-bool init_lyra2rev2_4way_ctx();

 #else

@@ -185,19 +186,26 @@ bool init_allium_ctx();

 /////////////////////////////////////////

-#if defined(__AVX2__) && defined(__AES__)
-//  #define PHI2_4WAY
+#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+  #define PHI2_8WAY 1
+#elif defined(__AVX2__) && defined(__AES__)
+  #define PHI2_4WAY 1
 #endif

-bool phi2_has_roots;
+extern bool phi2_has_roots;

 bool register_phi2_algo( algo_gate_t* gate );
-#if defined(PHI2_4WAY)
+#if defined(PHI2_8WAY)
+
+void phi2_8way_hash( void *state, const void *input );
+int scanhash_phi2_8way( struct work *work, uint32_t max_nonce,
+                     uint64_t *hashes_done, struct thr_info *mythr );
+
+#elif defined(PHI2_4WAY)

 void phi2_hash_4way( void *state, const void *input );
 int scanhash_phi2_4way( struct work *work, uint32_t max_nonce,
                     uint64_t *hashes_done, struct thr_info *mythr );
-//void init_phi2_ctx();

 #else

--- a/algo/lyra2/lyra2h-4way.c
+++ b/algo/lyra2/lyra2h-4way.c
@@ -76,7 +76,7 @@ int scanhash_lyra2h_4way( struct work *work, uint32_t max_nonce,
           && !opt_benchmark )
      {
          pdata[19] = n+i;         
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+          submit_solution( work, hash+(i<<3), mythr );
      }
      n += 4;
   } while (  (n < max_nonce-4) && !work_restart[thr_id].restart);
--- a/algo/lyra2/lyra2h.c
+++ b/algo/lyra2/lyra2h.c
@@ -1,4 +1,7 @@
 #include "lyra2-gate.h"
+
+#if !( defined(LYRA2H_8WAY) || defined(LYRA2H_4WAY) )
+
 #include <memory.h>
 #include <mm_malloc.h>
 #include "lyra2.h"
@@ -71,3 +74,4 @@ int scanhash_lyra2h( struct work *work, uint32_t max_nonce,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
+#endif
--- a/algo/lyra2/lyra2rev2-4way.c
+++ b/algo/lyra2/lyra2rev2-4way.c
@@ -7,23 +7,227 @@
 #include "algo/cubehash/cubehash_sse2.h" 
 #include "algo/cubehash/cube-hash-2way.h"

-#if defined (LYRA2REV2_8WAY)
+
+#if defined (LYRA2REV2_16WAY)
+
+typedef struct {
+   blake256_16way_context    blake;
+   keccak256_8way_context    keccak;
+   cubehashParam             cube;
+   skein256_8way_context     skein;
+   bmw256_16way_context      bmw;
+} lyra2v2_16way_ctx_holder __attribute__ ((aligned (64)));
+
+static lyra2v2_16way_ctx_holder l2v2_16way_ctx;
+
+bool init_lyra2rev2_16way_ctx()
+{
+   keccak256_8way_init( &l2v2_16way_ctx.keccak );
+   cubehashInit( &l2v2_16way_ctx.cube, 256, 16, 32 );
+   skein256_8way_init( &l2v2_16way_ctx.skein );
+   bmw256_16way_init( &l2v2_16way_ctx.bmw );
+   return true;
+}
+
+void lyra2rev2_16way_hash( void *state, const void *input )
+{
+   uint32_t vhash[8*16] __attribute__ ((aligned (128)));
+   uint32_t hash0[8] __attribute__ ((aligned (64)));
+   uint32_t hash1[8] __attribute__ ((aligned (64)));
+   uint32_t hash2[8] __attribute__ ((aligned (64)));
+   uint32_t hash3[8] __attribute__ ((aligned (64)));
+   uint32_t hash4[8] __attribute__ ((aligned (64)));
+   uint32_t hash5[8] __attribute__ ((aligned (64)));
+   uint32_t hash6[8] __attribute__ ((aligned (64)));
+   uint32_t hash7[8] __attribute__ ((aligned (64)));
+   uint32_t hash8[8] __attribute__ ((aligned (64)));
+   uint32_t hash9[8] __attribute__ ((aligned (64)));
+   uint32_t hash10[8] __attribute__ ((aligned (64)));
+   uint32_t hash11[8] __attribute__ ((aligned (64)));
+   uint32_t hash12[8] __attribute__ ((aligned (64)));
+   uint32_t hash13[8] __attribute__ ((aligned (64)));
+   uint32_t hash14[8] __attribute__ ((aligned (64)));
+   uint32_t hash15[8] __attribute__ ((aligned (64)));
+   lyra2v2_16way_ctx_holder ctx __attribute__ ((aligned (64)));
+   memcpy( &ctx, &l2v2_16way_ctx, sizeof(l2v2_16way_ctx) );
+
+   blake256_16way_update( &ctx.blake, input + (64<<4), 16 );
+   blake256_16way_close( &ctx.blake, vhash );
+
+   dintrlv_16x32( hash0,  hash1,  hash2,  hash3,
+                  hash4,  hash5,  hash6,  hash7,
+                  hash8,  hash9,  hash10, hash11,
+                  hash12, hash13, hash14, hash15, vhash, 256 );
+
+   intrlv_8x64( vhash, hash0, hash1, hash2, hash3,
+                       hash4, hash5, hash6, hash7, 256 );
+
+   keccak256_8way_update( &ctx.keccak, vhash, 32 );
+   keccak256_8way_close( &ctx.keccak, vhash );
+
+   dintrlv_8x64( hash0, hash1, hash2, hash3,
+                 hash4, hash5, hash6, hash7, vhash, 256 );
+   intrlv_8x64( vhash, hash8,  hash9,  hash10, hash11,
+                       hash12, hash13, hash14, hash15, 256 );
+
+   keccak256_8way_init( &ctx.keccak );
+   keccak256_8way_update( &ctx.keccak, vhash, 32 );
+   keccak256_8way_close( &ctx.keccak, vhash );
+
+   dintrlv_8x64( hash8,  hash9,  hash10,  hash11,
+                 hash12, hash13, hash14, hash5, vhash, 256 );
+
+   cubehash_full( &ctx.cube, (byte*) hash0,  256, (const byte*) hash0,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash1,  256, (const byte*) hash1,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash2,  256, (const byte*) hash2,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash3,  256, (const byte*) hash3,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash4,  256, (const byte*) hash4,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash5,  256, (const byte*) hash5,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash6,  256, (const byte*) hash6,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash7,  256, (const byte*) hash7,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash8,  256, (const byte*) hash8,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash9,  256, (const byte*) hash9,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash10, 256, (const byte*) hash10, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash11, 256, (const byte*) hash11, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash12, 256, (const byte*) hash12, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash13, 256, (const byte*) hash13, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash14, 256, (const byte*) hash14, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash15, 256, (const byte*) hash15, 32 );
+
+
+   intrlv_2x256( vhash, hash0, hash1, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash0, hash1, vhash, 256 );
+   intrlv_2x256( vhash, hash2, hash3, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash2, hash3, vhash, 256 );
+   intrlv_2x256( vhash, hash4, hash5, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash4, hash5, vhash, 256 );
+   intrlv_2x256( vhash, hash6, hash7, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash6, hash7, vhash, 256 );
+   intrlv_2x256( vhash, hash8, hash9, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash8, hash9, vhash, 256 );
+   intrlv_2x256( vhash, hash10, hash11, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash10, hash11, vhash, 256 );
+   intrlv_2x256( vhash, hash12, hash13, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash12, hash13, vhash, 256 );
+   intrlv_2x256( vhash, hash14, hash15, 256 );
+   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
+   dintrlv_2x256( hash14, hash15, vhash, 256 );
+
+   intrlv_8x64( vhash, hash0, hash1, hash2, hash3,
+                       hash4, hash5, hash6, hash7, 256 );
+   skein256_8way_update( &ctx.skein, vhash, 32 );
+   skein256_8way_close( &ctx.skein, vhash );
+
+   dintrlv_8x64( hash0, hash1, hash2, hash3,
+                 hash4, hash5, hash6, hash7, vhash, 256 );
+   intrlv_8x64( vhash, hash8,  hash9,  hash10, hash11, hash12,
+                       hash13, hash14, hash15, 256 );
+
+   skein256_8way_init( &ctx.skein );
+   skein256_8way_update( &ctx.skein, vhash, 32 );
+   skein256_8way_close( &ctx.skein, vhash );
+
+   dintrlv_8x64( hash8,  hash9,  hash10, hash11,
+                 hash12, hash13, hash14, hash15, vhash, 256 );
+
+   
+   cubehash_full( &ctx.cube, (byte*) hash0,  256, (const byte*) hash0, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash1,  256, (const byte*) hash1, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash2,  256, (const byte*) hash2, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash3,  256, (const byte*) hash3, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash4,  256, (const byte*) hash4, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash5,  256, (const byte*) hash5, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash6,  256, (const byte*) hash6, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash7,  256, (const byte*) hash7, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash8,  256, (const byte*) hash8,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash9,  256, (const byte*) hash9,  32 );
+   cubehash_full( &ctx.cube, (byte*) hash10, 256, (const byte*) hash10, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash11, 256, (const byte*) hash11, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash12, 256, (const byte*) hash12, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash13, 256, (const byte*) hash13, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash14, 256, (const byte*) hash14, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash15, 256, (const byte*) hash15, 32 );
+
+   intrlv_16x32( vhash, hash0,  hash1,  hash2,  hash3,
+                        hash4,  hash5,  hash6,  hash7,
+                        hash8,  hash9,  hash10, hash11,
+                        hash12, hash13, hash14, hash15, 256 );
+
+   bmw256_16way_update( &ctx.bmw, vhash, 32 );
+   bmw256_16way_close( &ctx.bmw, state );
+}
+
+int scanhash_lyra2rev2_16way( struct work *work, const uint32_t max_nonce,
+                             uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t hash[8*16] __attribute__ ((aligned (128)));
+   uint32_t vdata[20*16] __attribute__ ((aligned (64)));
+   uint32_t *hashd7 = &hash[7*16];
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 16;
+   uint32_t n = first_nonce;
+   const uint32_t targ32 = ptarget[7];
+   __m512i  *noncev = (__m512i*)vdata + 19;
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+
+   if ( bench )  ptarget[7] = 0x0000ff;
+
+   mm512_bswap32_intrlv80_16x32( vdata, pdata );
+   *noncev = _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 );
+   blake256_16way_init( &l2v2_16way_ctx.blake );
+   blake256_16way_update( &l2v2_16way_ctx.blake, vdata, 64 );
+
+   do
+   {
+      lyra2rev2_16way_hash( hash, vdata );
+
+      for ( int lane = 0; lane < 16; lane++ )
+      if ( unlikely( hashd7[lane] <= targ32 ) )
+      {
+         extr_lane_16x32( lane_hash, hash, lane, 256 );
+         if ( likely( valid_hash( lane_hash, ptarget ) && !bench ) )
+         {
+             pdata[19] = bswap_32( n + lane );
+             submit_solution( work, lane_hash, mythr );
+         }
+      }
+      *noncev = _mm512_add_epi32( *noncev, m512_const1_32( 16 ) );
+      n += 16;
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
+#elif defined (LYRA2REV2_8WAY)

 typedef struct {
   blake256_8way_context     blake;
-   keccak256_8way_context    keccak;
-   cube_4way_context          cube;
-   skein256_8way_context     skein;
-   bmw256_8way_context          bmw;
+   keccak256_4way_context    keccak;
+   cubehashParam             cube;
+   skein256_4way_context     skein;
+   bmw256_8way_context       bmw;
 } lyra2v2_8way_ctx_holder __attribute__ ((aligned (64)));

 static lyra2v2_8way_ctx_holder l2v2_8way_ctx;

 bool init_lyra2rev2_8way_ctx()
 {
-   keccak256_8way_init( &l2v2_8way_ctx.keccak );
-   cube_4way_init( &l2v2_8way_ctx.cube, 256, 16, 32 );
-   skein256_8way_init( &l2v2_8way_ctx.skein );
+   keccak256_4way_init( &l2v2_8way_ctx.keccak );
+   cubehashInit( &l2v2_8way_ctx.cube, 256, 16, 32 );
+   skein256_4way_init( &l2v2_8way_ctx.skein );
   bmw256_8way_init( &l2v2_8way_ctx.bmw );
   return true;
 }
@@ -31,8 +235,6 @@ bool init_lyra2rev2_8way_ctx()
 void lyra2rev2_8way_hash( void *state, const void *input )
 {
   uint32_t vhash[8*8] __attribute__ ((aligned (128)));
-   uint32_t vhashA[8*8] __attribute__ ((aligned (64)));
-   uint32_t vhashB[8*8] __attribute__ ((aligned (64)));
   uint32_t hash0[8] __attribute__ ((aligned (64)));
   uint32_t hash1[8] __attribute__ ((aligned (64)));
   uint32_t hash2[8] __attribute__ ((aligned (64)));
@@ -47,103 +249,113 @@ void lyra2rev2_8way_hash( void *state, const void *input )
   blake256_8way_update( &ctx.blake, input + (64<<3), 16 );
   blake256_8way_close( &ctx.blake, vhash );

-   rintrlv_8x32_8x64( vhashA, vhash, 256 );
+   dintrlv_8x32( hash0, hash1, hash2, hash3,
+                 hash4, hash5, hash6, hash7, vhash, 256 );

-   keccak256_8way_update( &ctx.keccak, vhashA, 32 );
-   keccak256_8way_close( &ctx.keccak, vhash );
+   intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 256 );
+   keccak256_4way_update( &ctx.keccak, vhash, 32 );
+   keccak256_4way_close( &ctx.keccak, vhash );
+   dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 256 );
+   intrlv_4x64( vhash, hash4, hash5, hash6, hash7, 256 );
+   keccak256_4way_init( &ctx.keccak );
+   keccak256_4way_update( &ctx.keccak, vhash, 32 );
+   keccak256_4way_close( &ctx.keccak, vhash );
+   dintrlv_4x64( hash4, hash5, hash6, hash7, vhash, 256 );

-   rintrlv_8x64_4x128( vhashA, vhashB, vhash, 256 );
+   cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );

-   cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );

-   dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
-   dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
-
-   intrlv_2x256( vhash, hash0, hash1, 256 );
-   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
-   dintrlv_2x256( hash0, hash1, vhash, 256 );
-   intrlv_2x256( vhash, hash2, hash3, 256 );
-   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
-   dintrlv_2x256( hash2, hash3, vhash, 256 );
-   intrlv_2x256( vhash, hash4, hash5, 256 );
-   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
-   dintrlv_2x256( hash4, hash5, vhash, 256 );
-   intrlv_2x256( vhash, hash6, hash7, 256 );
-   LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
-   dintrlv_2x256( hash6, hash7, vhash, 256 );
-
-   intrlv_8x64( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
-                hash7, 256 );
-
-   skein256_8way_update( &ctx.skein, vhash, 32 );
-   skein256_8way_close( &ctx.skein, vhash );
-
-   rintrlv_8x64_4x128( vhashA, vhashB, vhash, 256 );
-
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
+   LYRA2REV2( l2v2_wholeMatrix, hash0, 32, hash0, 32, hash0, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash1, 32, hash1, 32, hash1, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash2, 32, hash2, 32, hash2, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash3, 32, hash3, 32, hash3, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash4, 32, hash4, 32, hash4, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash5, 32, hash5, 32, hash5, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash6, 32, hash6, 32, hash6, 32, 1, 4, 4 );
+   LYRA2REV2( l2v2_wholeMatrix, hash7, 32, hash7, 32, hash7, 32, 1, 4, 4 );
   
-   dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
-   dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
+   intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 256 );
+   skein256_4way_update( &ctx.skein, vhash, 32 );
+   skein256_4way_close( &ctx.skein, vhash );
+   dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 256 );
+   intrlv_4x64( vhash, hash4, hash5, hash6, hash7, 256 );
+   skein256_4way_init( &ctx.skein );
+   skein256_4way_update( &ctx.skein, vhash, 32 );
+   skein256_4way_close( &ctx.skein, vhash );
+   dintrlv_4x64( hash4, hash5, hash6, hash7, vhash, 256 );

-   intrlv_8x32( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6, 
-                hash7, 256 );
+   cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );
+
+   intrlv_8x32( vhash, hash0, hash1, hash2, hash3,
+                       hash4, hash5, hash6, hash7, 256 );

   bmw256_8way_update( &ctx.bmw, vhash, 32 );
   bmw256_8way_close( &ctx.bmw, state );
 }

-int scanhash_lyra2rev2_8way( struct work *work, uint32_t max_nonce,
+int scanhash_lyra2rev2_8way( struct work *work, const 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 *hash7 = &(hash[7<<3]);
-   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
+   uint32_t *hashd7 = &hash[7*8];
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
   const uint32_t first_nonce = pdata[19];
   const uint32_t last_nonce = max_nonce - 8;
   uint32_t n = first_nonce;
-   const uint32_t Htarg = ptarget[7];
-   __m256i *noncev = (__m256i*)vdata + 19;   // aligned
-   int thr_id = mythr->id; 
+   const uint32_t targ32 = ptarget[7];
+   __m256i  *noncev = (__m256i*)vdata + 19;
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )
-      ( (uint32_t*)ptarget )[7] = 0x0000ff;
+   if ( bench )  ptarget[7] = 0x0000ff;

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
-
+   *noncev = _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n );
   blake256_8way_init( &l2v2_8way_ctx.blake );
   blake256_8way_update( &l2v2_8way_ctx.blake, vdata, 64 );

   do
   {
-      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
-                                                  n+3, n+2, n+1, n ) );
-
      lyra2rev2_8way_hash( hash, vdata );
-      pdata[19] = n;

-      for ( int lane = 0; lane < 8; lane++ ) if ( hash7[lane] <= Htarg )
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hashd7[lane] <= targ32 ) )
      {
         extr_lane_8x32( lane_hash, hash, lane, 256 );
-         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+         if ( likely( valid_hash( lane_hash, ptarget ) && !bench ) )
         {
-            pdata[19] = n + lane;
-            submit_lane_solution( work, lane_hash, mythr, lane );
+             pdata[19] = bswap_32( n + lane );
+             submit_solution( work, lane_hash, mythr );
         }
      }
+      *noncev = _mm256_add_epi32( *noncev, m256_const1_32( 8 ) );
      n += 8;
-   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
   *hashes_done = n - first_nonce;
   return 0;
 }

+#endif
+
+/*
 #elif defined (LYRA2REV2_4WAY)

 typedef struct {
@@ -226,15 +438,16 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
 {
   uint32_t hash[8*4] __attribute__ ((aligned (64)));
   uint32_t vdata[20*4] __attribute__ ((aligned (64)));
-   uint32_t *hash7 = &(hash[7<<2]);
+   uint32_t *hashd7 = &(hash[7<<2]);
   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 4;
   uint32_t n = first_nonce;
-   const uint32_t Htarg = ptarget[7];
-   __m128i *noncev = (__m128i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const uint32_t targ32 = ptarget[7];
+   __m128i *noncev = (__m128i*)vdata + 19;  
+   int thr_id = mythr->id; 

   if ( opt_benchmark )
      ( (uint32_t*)ptarget )[7] = 0x0000ff;
@@ -249,21 +462,22 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );

      lyra2rev2_4way_hash( hash, vdata );
-      pdata[19] = n;

-      for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
+      for ( int lane = 0; lane < 4; lane++ ) if ( hashd7[lane] <= targ32 )
      {
         extr_lane_4x32( lane_hash, hash, lane, 256 );
-         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+         if ( valid_hash( lane_hash, ptarget ) && !opt_benchmark )
         {
            pdata[19] = n + lane;         
-            submit_lane_solution( work, lane_hash, mythr, lane );
+            submit_solution( work, lane_hash, mythr );
         }
      }
      n += 4;
-   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
-   *hashes_done = n - first_nonce + 1;
+   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
   return 0;
 }

 #endif
+*/
--- a/algo/lyra2/lyra2rev2.c
+++ b/algo/lyra2/lyra2rev2.c
@@ -1,4 +1,7 @@
 #include "lyra2-gate.h"
+
+#if !( defined(LYRA2REV2_16WAY) || defined(LYRA2REV2_8WAY) || defined(LYRA2REV2_4WAY) )
+
 #include <memory.h>
 #include "algo/blake/sph_blake.h"
 #include "algo/cubehash/sph_cubehash.h"
@@ -96,7 +99,7 @@ int scanhash_lyra2rev2( struct work *work,
 		lyra2rev2_hash(hash, endiandata);

 		if (hash[7] <= Htarg )
-      if( fulltest( hash, ptarget ) && !opt_benchmark )
+      if( valid_hash( hash, ptarget ) && !opt_benchmark )
      {
 			pdata[19] = nonce;
         submit_solution( work, hash, mythr );
@@ -107,4 +110,4 @@ int scanhash_lyra2rev2( struct work *work,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
-
+#endif
--- a/algo/lyra2/lyra2rev3-4way.c
+++ b/algo/lyra2/lyra2rev3-4way.c
@@ -79,19 +79,16 @@ void lyra2rev3_16way_hash( void *state, const void *input )
   dintrlv_2x256( hash14, hash15, vhash, 256 );

   intrlv_4x128( vhash, hash0, hash1, hash2, hash3, 256 );
-   cube_4way_update_close( &ctx.cube, vhash, vhash, 32 );
+   cube_4way_full( &ctx.cube, vhash, 256, vhash, 32 );
   dintrlv_4x128( hash0, hash1, hash2, hash3, vhash, 256 );
   intrlv_4x128( vhash, hash4, hash5, hash6, hash7, 256 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhash, vhash, 32 );
+   cube_4way_full( &ctx.cube, vhash, 256, vhash, 32 );
   dintrlv_4x128( hash4, hash5, hash6, hash7, vhash, 256 );
   intrlv_4x128( vhash, hash8, hash9, hash10, hash11, 256 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhash, vhash, 32 );
+   cube_4way_full( &ctx.cube, vhash, 256, vhash, 32 );
   dintrlv_4x128( hash8, hash9, hash10, hash11, vhash, 256 );
   intrlv_4x128( vhash, hash12, hash13, hash14, hash15, 256 );
-   cube_4way_init( &ctx.cube, 256, 16, 32 );
-   cube_4way_update_close( &ctx.cube, vhash, vhash, 32 );
+   cube_4way_full( &ctx.cube, vhash, 256, vhash, 32 );
   dintrlv_4x128( hash12, hash13, hash14, hash15, vhash, 256 );

   intrlv_2x256( vhash, hash0, hash1, 256 );
@@ -133,15 +130,15 @@ int scanhash_lyra2rev3_16way( struct work *work, const uint32_t max_nonce,
 {
   uint32_t hash[8*16] __attribute__ ((aligned (128)));
   uint32_t vdata[20*16] __attribute__ ((aligned (64)));
-   uint32_t *hash7 = &hash[7<<4];
+   uint32_t *hashd7 = &hash[7*16];
   uint32_t lane_hash[8] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
   const uint32_t *ptarget = work->target;
   const uint32_t first_nonce = pdata[19];
   uint32_t n = first_nonce;
   const uint32_t last_nonce = max_nonce - 16;
-   const uint32_t Htarg = ptarget[7];
-   __m512i  *noncev = (__m512i*)vdata + 19;   // aligned
+   const uint32_t targ32 = ptarget[7];
+   __m512i  *noncev = (__m512i*)vdata + 19;
   const int thr_id = mythr->id;

   if ( opt_benchmark )  ( (uint32_t*)ptarget )[7] = 0x0000ff;
@@ -162,17 +159,18 @@ int scanhash_lyra2rev3_16way( struct work *work, const uint32_t max_nonce,
      pdata[19] = n;

      for ( int lane = 0; lane < 16; lane++ )
-      if ( unlikely( hash7[lane] <= Htarg ) )
+      if ( unlikely( hashd7[lane] <= targ32 ) )
      {
         extr_lane_16x32( lane_hash, hash, lane, 256 );
-         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
+         if ( likely( valid_hash( lane_hash, ptarget ) && !opt_benchmark ) )
         {
             pdata[19] = n + lane;
-             submit_lane_solution( work, lane_hash, mythr, lane );
+             submit_solution( work, lane_hash, mythr );
         }
      }
      n += 16;
   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
   *hashes_done = n - first_nonce;
   return 0;
 }
@@ -197,7 +195,7 @@ bool init_lyra2rev3_8way_ctx()

 void lyra2rev3_8way_hash( void *state, const void *input )
 {
-   uint32_t vhash[8*8] __attribute__ ((aligned (64)));
+   uint32_t vhash[8*8] __attribute__ ((aligned (128)));
   uint32_t hash0[8] __attribute__ ((aligned (64)));
   uint32_t hash1[8] __attribute__ ((aligned (32)));
   uint32_t hash2[8] __attribute__ ((aligned (32)));
@@ -224,21 +222,14 @@ void lyra2rev3_8way_hash( void *state, const void *input )
   LYRA2REV3( l2v3_wholeMatrix, hash6, 32, hash6, 32, hash6, 32, 1, 4, 4 );
   LYRA2REV3( l2v3_wholeMatrix, hash7, 32, hash7, 32, hash7, 32, 1, 4, 4 );

-   cubehashUpdateDigest( &ctx.cube, (byte*) hash0, (const byte*) hash0, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash1, (const byte*) hash1, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash2, (const byte*) hash2, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash3, (const byte*) hash3, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash4, (const byte*) hash4, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash5, (const byte*) hash5, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash6, (const byte*) hash6, 32 );
-   cubehashInit( &ctx.cube, 256, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*) hash7, (const byte*) hash7, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
+   cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );

   LYRA2REV3( l2v3_wholeMatrix, hash0, 32, hash0, 32, hash0, 32, 1, 4, 4 );
   LYRA2REV3( l2v3_wholeMatrix, hash1, 32, hash1, 32, hash1, 32, 1, 4, 4 );
@@ -260,46 +251,47 @@ void lyra2rev3_8way_hash( void *state, const void *input )
 int scanhash_lyra2rev3_8way( struct work *work, const uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8*8] __attribute__ ((aligned (64)));
+   uint32_t hash[8*8] __attribute__ ((aligned (128)));
   uint32_t vdata[20*8] __attribute__ ((aligned (64)));
-   uint32_t *hash7 = &hash[7<<3];
+   uint32_t *hashd7 = &hash[7*8];
   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
   uint32_t *pdata = work->data;
-   const uint32_t *ptarget = work->target;
+   uint32_t *ptarget = work->target;
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 8;
   uint32_t n = first_nonce;
-   const uint32_t Htarg = ptarget[7];
-   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
+   const uint32_t targ32 = ptarget[7];
+   __m256i  *noncev = (__m256i*)vdata + 19;  
   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )  ( (uint32_t*)ptarget )[7] = 0x0000ff;
+   if ( bench )  ptarget[7] = 0x0000ff;

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
-
+   *noncev = _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n );
   blake256_8way_init( &l2v3_8way_ctx.blake );
   blake256_8way_update( &l2v3_8way_ctx.blake, vdata, 64 );

   do
   {
-      *noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
-                                                  n+3, n+2, n+1, n ) );
-
      lyra2rev3_8way_hash( hash, vdata );
      pdata[19] = n;

      for ( int lane = 0; lane < 8; lane++ )
-      if ( unlikely( hash7[lane] <= Htarg ) )
+      if ( unlikely( hashd7[lane] <= targ32 ) )
      {
         extr_lane_8x32( lane_hash, hash, lane, 256 );
-         if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
+         if ( likely( valid_hash( lane_hash, ptarget ) && !bench ) )
         {
-             pdata[19] = n + lane;
-             submit_lane_solution( work, lane_hash, mythr, lane );
+             pdata[19] = bswap_32( n + lane );
+             submit_solution( work, lane_hash, mythr );
         }
      }
+      *noncev = _mm256_add_epi32( *noncev, m256_const1_32( 8 ) );
      n += 8;
-   } while ( likely( (n < max_nonce-8) && !work_restart[thr_id].restart ) );
-   *hashes_done = n - first_nonce + 1;
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
   return 0;
 }

@@ -366,42 +358,41 @@ int scanhash_lyra2rev3_4way( struct work *work, const uint32_t max_nonce,
 {
   uint32_t hash[8*4] __attribute__ ((aligned (64)));
   uint32_t vdata[20*4] __attribute__ ((aligned (64)));
-   uint32_t *hash7 = &(hash[7<<2]);
+   uint32_t *hashd7 = &(hash[7*4]);
   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
   uint32_t *pdata = work->data;
   const uint32_t *ptarget = work->target;
   const uint32_t first_nonce = pdata[19];
   uint32_t n = first_nonce;
-   const uint32_t Htarg = ptarget[7];
-   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
-   const int thr_id = mythr->id;  // thr_id arg is deprecated
+   const uint32_t targ32 = ptarget[7];
+   __m128i  *noncev = (__m128i*)vdata + 19; 
+   const int thr_id = mythr->id;
   
   if ( opt_benchmark )
      ( (uint32_t*)ptarget )[7] = 0x0000ff;

   mm128_bswap32_intrlv80_4x32( vdata, pdata );
+   *noncev = _mm_set_epi32( n+3, n+2, n+1, n );

   blake256_4way_init( &l2v3_4way_ctx.blake );
   blake256_4way_update( &l2v3_4way_ctx.blake, vdata, 64 );

   do
   {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
-
      lyra2rev3_4way_hash( hash, vdata );
-      pdata[19] = n;
-
-      for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
+      for ( int lane = 0; lane < 4; lane++ ) if ( hashd7[lane] <= targ32 )
      {
         extr_lane_4x32( lane_hash, hash, lane, 256 );
-         if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
+         if ( valid_hash( lane_hash, ptarget ) && !opt_benchmark ) 
         {
-              pdata[19] = n + lane;    
-              submit_lane_solution( work, lane_hash, mythr, lane );
+              pdata[19] = bswap_32( n + lane );    
+              submit_solution( work, lane_hash, mythr );
 	      }
      }
+      *noncev = _mm_add_epi32( *noncev, m128_const1_32( 4 ) );
      n += 4;
   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
+   pdata[19] = n;
   *hashes_done = n - first_nonce + 1;
   return 0;
 }
--- a/algo/lyra2/lyra2rev3.c
+++ b/algo/lyra2/lyra2rev3.c
@@ -1,4 +1,7 @@
 #include "lyra2-gate.h"
+
+#if !( defined(LYRA2REV3_16WAY) || defined(LYRA2REV3_8WAY) || defined(LYRA2REV3_4WAY) )
+
 #include <memory.h>
 #include "algo/blake/sph_blake.h"
 #include "algo/cubehash/sph_cubehash.h"
@@ -85,7 +88,7 @@ int scanhash_lyra2rev3( struct work *work,
 	lyra2rev3_hash(hash, endiandata);

      if (hash[7] <= Htarg )
-      if( fulltest( hash, ptarget ) && !opt_benchmark )
+      if( valid_hash( hash, ptarget ) && !opt_benchmark )
      {
          pdata[19] = nonce;
          submit_solution( work, hash, mythr );
@@ -96,4 +99,4 @@ int scanhash_lyra2rev3( struct work *work,
   *hashes_done = pdata[19] - first_nonce + 1;
   return 0;
 }
-
+#endif
--- a/algo/lyra2/lyra2z-4way.c
+++ b/algo/lyra2/lyra2z-4way.c
@@ -97,41 +97,42 @@ void lyra2z_16way_hash( void *state, const void *input )
 int scanhash_lyra2z_16way( struct work *work, uint32_t max_nonce,
                          uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8*16] __attribute__ ((aligned (128)));
+   uint64_t hash[4*16] __attribute__ ((aligned (128)));
   uint32_t vdata[20*16] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
   uint32_t n = first_nonce;
+   const uint32_t last_nonce = max_nonce - 16;
   __m512i  *noncev = (__m512i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )
-      ptarget[7] = 0x0000ff;
+   if ( bench )   ptarget[7] = 0x0000ff;

   mm512_bswap32_intrlv80_16x32( vdata, pdata );
+   *noncev = _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 );
   lyra2z_16way_midstate( vdata );

   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 ) );
      lyra2z_16way_hash( hash, vdata );
-      pdata[19] = n;

-      for ( int i = 0; i < 16; i++ )
-      if ( (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget )
-           && !opt_benchmark )
+      for ( int lane = 0; lane < 16; lane++ )
      {
-          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+        const uint64_t *lane_hash = hash + (lane<<2);
+        if ( unlikely( valid_hash( lane_hash, ptarget ) && !bench ) )
+        {
+           pdata[19] = bswap_32( n + lane );
+           submit_solution( work, lane_hash, mythr );
+        }
      }
+      *noncev = _mm512_add_epi32( *noncev, m512_const1_32( 16 ) );
      n += 16;
-   } while ( (n < max_nonce-16) && !work_restart[thr_id].restart);
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );

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

@@ -195,39 +196,40 @@ void lyra2z_8way_hash( void *state, const void *input )
 int scanhash_lyra2z_8way( struct work *work, uint32_t max_nonce,
                          uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8*8] __attribute__ ((aligned (64)));
+   uint64_t hash[4*8] __attribute__ ((aligned (64)));
   uint32_t vdata[20*8] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 8;
   uint32_t n = first_nonce;
   __m256i  *noncev = (__m256i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )
-      ptarget[7] = 0x0000ff;
+   if ( bench )  ptarget[7] = 0x0000ff;

   mm256_bswap32_intrlv80_8x32( vdata, pdata );
+   *noncev = _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n );
   lyra2z_8way_midstate( vdata );

   do {
-      *noncev = mm256_bswap_32(
-                 _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n ) );
      lyra2z_8way_hash( hash, vdata );
-      pdata[19] = n;

-      for ( int i = 0; i < 8; i++ )
-      if ( (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget )
-           && !opt_benchmark )
+      for ( int lane = 0; lane < 8; lane++ )
      {
-          pdata[19] = n+i;
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+        const uint64_t *lane_hash = hash + (lane<<2);
+        if ( unlikely( valid_hash( lane_hash, ptarget ) && !bench ) )
+        {
+           pdata[19] = bswap_32( n + lane );
+           submit_solution( work, lane_hash, mythr );
+        }
      }
+      *noncev = _mm256_add_epi32( *noncev, m256_const1_32( 8 ) );
      n += 8;
-   } while ( (n < max_nonce-8) && !work_restart[thr_id].restart);
-
-   *hashes_done = n - first_nonce + 1;
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart) );
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
   return 0;
 }

@@ -274,39 +276,40 @@ void lyra2z_4way_hash( void *state, const void *input )
 int scanhash_lyra2z_4way( struct work *work, uint32_t max_nonce,
                          uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8*4] __attribute__ ((aligned (64)));
+   uint64_t hash[4*4] __attribute__ ((aligned (64)));
   uint32_t vdata[20*4] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 4;
   uint32_t n = first_nonce;
   __m128i  *noncev = (__m128i*)vdata + 19;   // aligned
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id; 
+   const bool bench = opt_benchmark;

-   if ( opt_benchmark )
-      ptarget[7] = 0x0000ff;
+   if ( bench )   ptarget[7] = 0x0000ff;

   mm128_bswap32_intrlv80_4x32( vdata, pdata );
+   *noncev = _mm_set_epi32( n+3, n+2, n+1, n );
   lyra2z_4way_midstate( vdata );

   do {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
-
      lyra2z_4way_hash( hash, vdata );
-      pdata[19] = n;
-
-      for ( int i = 0; i < 4; i++ )
-      if ( (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget )
-           && !opt_benchmark )
+      for ( int lane = 0; lane < 4; lane++ )
      {
-          pdata[19] = n+i;         
-          submit_lane_solution( work, hash+(i<<3), mythr, i );
+        const uint64_t *lane_hash = hash + (lane<<2);
+        if ( unlikely( valid_hash( lane_hash, ptarget ) && !bench ) )
+        {
+           pdata[19] = bswap_32( n + lane );
+           submit_solution( work, lane_hash, mythr );
+        }
      }
+      *noncev = _mm_add_epi32( *noncev, m128_const1_32( 4 ) );
      n += 4;
-   } while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );

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

--- a/algo/lyra2/lyra2z.c
+++ b/algo/lyra2/lyra2z.c
@@ -1,6 +1,9 @@
 #include <memory.h>
 #include <mm_malloc.h>
 #include "lyra2-gate.h"
+
+#if !( defined(LYRA2Z_16WAY) || defined(LYRA2Z_8WAY) || defined(LYRA2Z_4WAY) )
+
 #include "lyra2.h"
 #include "algo/blake/sph_blake.h"
 #include "simd-utils.h"
@@ -53,7 +56,7 @@ int scanhash_lyra2z( struct work *work, uint32_t max_nonce,
 	const uint32_t Htarg = ptarget[7];
 	const uint32_t first_nonce = pdata[19];
 	uint32_t nonce = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   int thr_id = mythr->id; 

 	if (opt_benchmark)
 		ptarget[7] = 0x0000ff;
@@ -62,14 +65,13 @@ int scanhash_lyra2z( struct work *work, uint32_t max_nonce,
 		be32enc(&endiandata[i], pdata[i]);
 	}

-        lyra2z_midstate( endiandata );
+   lyra2z_midstate( endiandata );

 	do {
 		be32enc(&endiandata[19], nonce);
                lyra2z_hash( hash, endiandata );

-      if ( hash[7] <= Htarg )
-      if ( fulltest( hash, ptarget ) && !opt_benchmark )
+      if ( valid_hash( hash, ptarget ) && !opt_benchmark )
      {
 			pdata[19] = nonce;
 			submit_solution( work, hash, mythr );
@@ -80,4 +82,4 @@ int scanhash_lyra2z( struct work *work, uint32_t max_nonce,
 	*hashes_done = pdata[19] - first_nonce + 1;
 	return 0;
 }
-
+#endif
--- a/algo/lyra2/lyra2z330.c
+++ b/algo/lyra2/lyra2z330.c
@@ -9,7 +9,7 @@ void lyra2z330_hash(void *state, const void *input, uint32_t height)
 {
 	uint32_t _ALIGN(256) hash[16];

-        LYRA2Z( lyra2z330_wholeMatrix, hash, 32, input, 80, input, 80,
+   LYRA2Z( lyra2z330_wholeMatrix, hash, 32, input, 80, input, 80,
                 2, 330, 256 );

 	memcpy(state, hash, 32);
@@ -18,38 +18,40 @@ void lyra2z330_hash(void *state, const void *input, uint32_t height)
 int scanhash_lyra2z330( struct work *work, uint32_t max_nonce,
                        uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t hash[8] __attribute__ ((aligned (64))); 
-   uint32_t endiandata[20] __attribute__ ((aligned (64)));
+   uint32_t hash[8] __attribute__ ((aligned (128))); 
+   uint32_t edata[20] __attribute__ ((aligned (64)));
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
   uint32_t nonce = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id; 

   if (opt_benchmark)
 	ptarget[7] = 0x0000ff;

-   casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
-   casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
-   casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
-   casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
-   casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
+   casti_m128i( edata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
+   casti_m128i( edata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
+   casti_m128i( edata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
+   casti_m128i( edata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
+   casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
   
   do
   {
-      be32enc( &endiandata[19], nonce );
-      lyra2z330_hash( hash, endiandata, work->height );
-      if ( hash[7] <= Htarg )
-      if ( fulltest( hash, ptarget ) && !opt_benchmark )
+      edata[19] = nonce;
+
+      LYRA2Z( lyra2z330_wholeMatrix, hash, 32, edata, 80, edata, 80,
+                 2, 330, 256 );
+      
+//      lyra2z330_hash( hash, edata, work->height );
+      if ( valid_hash( hash, ptarget ) && !opt_benchmark )
      {
-         pdata[19] = nonce;
+         be32enc( pdata + 19, nonce );
         submit_solution( work, hash, mythr );
      }
      nonce++;
   } while ( nonce < max_nonce && !work_restart[thr_id].restart );
   pdata[19] = nonce;
-   *hashes_done = pdata[19] - first_nonce + 1;
+   *hashes_done = nonce - first_nonce;
   return 0;
 }

@@ -66,7 +68,7 @@ bool lyra2z330_thread_init()

 bool register_lyra2z330_algo( algo_gate_t* gate )
 {
-  gate->optimizations = SSE42_OPT | AVX2_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT;
  gate->miner_thread_init = (void*)&lyra2z330_thread_init;
  gate->scanhash   = (void*)&scanhash_lyra2z330;
  gate->hash       = (void*)&lyra2z330_hash;
--- a/algo/lyra2/phi2-4way.c
+++ b/algo/lyra2/phi2-4way.c
@@ -1,233 +1,501 @@
-/**
- * Phi-2 algo Implementation
- */
-
 #include "lyra2-gate.h"
-
-#if defined(PHI2_4WAY)
-
 #include "algo/skein/skein-hash-4way.h"
 #include "algo/jh/jh-hash-4way.h"
 #include "algo/gost/sph_gost.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#include "algo/echo/aes_ni/hash_api.h"
+#include "lyra2.h"
+#if defined(__VAES__)
+  #include "algo/echo/echo-hash-4way.h"
+#elif defined(__AES__)
+  #include "algo/echo/aes_ni/hash_api.h"
+#endif
+
+#if defined(PHI2_8WAY)
+
+typedef struct {
+     cubehashParam           cube;
+     jh512_8way_context      jh;
+#if  defined(__VAES__)
+     echo_4way_context       echo;
+#else
+     hashState_echo          echo;
+#endif
+     sph_gost512_context     gost;
+     skein512_8way_context   skein;
+} phi2_8way_ctx_holder;
+
+void phi2_8way_hash( void *state, const void *input )
+{
+   unsigned char _ALIGN(128) hash[64*8];
+   unsigned char _ALIGN(128) hashA[64*2];
+   unsigned char _ALIGN(64) hash0[64];
+   unsigned char _ALIGN(64) hash1[64];
+   unsigned char _ALIGN(64) hash2[64];
+   unsigned char _ALIGN(64) hash3[64];
+   unsigned char _ALIGN(64) hash4[64];
+   unsigned char _ALIGN(64) hash5[64];
+   unsigned char _ALIGN(64) hash6[64];
+   unsigned char _ALIGN(64) hash7[64];
+   const int size = phi2_has_roots ? 144 : 80 ;
+   phi2_8way_ctx_holder ctx __attribute__ ((aligned (64)));
+
+   cubehash_full( &ctx.cube, (byte*)hash0, 512,
+                       (const byte*)input,         size );
+   cubehash_full( &ctx.cube, (byte*)hash1, 512,
+                       (const byte*)input +   144, size );
+   cubehash_full( &ctx.cube, (byte*)hash2, 512,
+                       (const byte*)input + 2*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash3, 512,
+                       (const byte*)input + 3*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash4, 512,
+                       (const byte*)input + 4*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash5, 512,
+                       (const byte*)input + 5*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash6, 512,
+                       (const byte*)input + 6*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash7, 512,
+                       (const byte*)input + 7*144, size );
+
+   intrlv_2x256( hashA, hash0, hash1, 512 );
+   LYRA2RE_2WAY( hash,        32, hashA,        32, 1, 8, 8 );
+   LYRA2RE_2WAY( hash + 2*32, 32, hashA + 2*32, 32, 1, 8, 8 );
+   dintrlv_2x256( hash0, hash1, hash, 512 );
+   intrlv_2x256( hashA, hash2, hash3, 512 );
+   LYRA2RE_2WAY( hash,        32, hashA,        32, 1, 8, 8 );
+   LYRA2RE_2WAY( hash + 2*32, 32, hashA + 2*32, 32, 1, 8, 8 );
+   dintrlv_2x256( hash2, hash3, hash, 512 );
+   intrlv_2x256( hashA, hash4, hash5, 512 );
+   LYRA2RE_2WAY( hash,        32, hashA,        32, 1, 8, 8 );
+   LYRA2RE_2WAY( hash + 2*32, 32, hashA + 2*32, 32, 1, 8, 8 );
+   dintrlv_2x256( hash4, hash5, hash, 512 );
+   intrlv_2x256( hashA, hash6, hash7, 512 );
+   LYRA2RE_2WAY( hash,        32, hashA,        32, 1, 8, 8 );
+   LYRA2RE_2WAY( hash + 2*32, 32, hashA + 2*32, 32, 1, 8, 8 );
+   dintrlv_2x256( hash6, hash7, hash, 512 );
+   
+   intrlv_8x64_512( hash, hash0, hash1, hash2, hash3,
+                          hash4, hash5, hash6, hash7 );
+
+   jh512_8way_init( &ctx.jh );
+   jh512_8way_update( &ctx.jh, (const void*)hash, 64 );
+   jh512_8way_close( &ctx.jh, (void*)hash );
+
+   dintrlv_8x64_512( hash0, hash1, hash2, hash3,
+                     hash4, hash5, hash6, hash7, hash );
+
+#if defined (__VAES__)
+
+   unsigned char _ALIGN(64) hashA0[64];
+   unsigned char _ALIGN(64) hashA1[64];
+   unsigned char _ALIGN(64) hashA2[64];
+   unsigned char _ALIGN(64) hashA3[64];
+   unsigned char _ALIGN(64) hashA4[64];
+   unsigned char _ALIGN(64) hashA5[64];
+   unsigned char _ALIGN(64) hashA6[64];
+   unsigned char _ALIGN(64) hashA7[64];
+
+   intrlv_4x128_512( hash, hash0, hash1, hash2, hash3 );
+   echo_4way_full( &ctx.echo, hash, 512, hash, 64 ); 
+   echo_4way_full( &ctx.echo, hash, 512, hash, 64 );
+   dintrlv_4x128_512( hashA0, hashA1, hashA2, hashA3, hash );
+
+   intrlv_4x128_512( hash, hash4, hash5, hash6, hash7 );
+   echo_4way_full( &ctx.echo, hash, 512, hash, 64 );
+   echo_4way_full( &ctx.echo, hash, 512, hash, 64 );    
+   dintrlv_4x128_512( hashA4, hashA5, hashA6, hashA7, hash );
+
+#endif    
+
+   if ( hash0[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash0, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash0 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash0, hashA0, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash0, 512,
+                      (const BitSequence *)hash0, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash0, 512,
+                      (const BitSequence *)hash0, 64 );
+   }
+#endif
+   if ( hash1[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash1, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash1 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash1, hashA1, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash1, 512,
+                      (const BitSequence *)hash1, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash1, 512,
+                      (const BitSequence *)hash1, 64 );
+   }
+#endif
+   if ( hash2[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash2, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash2 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash2, hashA2, 64 );
+#else 
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash2, 512,
+                      (const BitSequence *)hash2, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash2, 512,
+                      (const BitSequence *)hash2, 64 );
+   }
+#endif
+   if ( hash3[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash3, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash3 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash3, hashA3, 64 );
+#else  
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash3, 512,
+                      (const BitSequence *)hash3, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash3, 512,
+                      (const BitSequence *)hash3, 64 );
+   }
+#endif
+   if ( hash4[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash4, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash4 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash4, hashA4, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash4, 512,
+                      (const BitSequence *)hash4, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash4, 512,
+                      (const BitSequence *)hash4, 64 );
+   }
+#endif   
+   if ( hash5[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash5, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash5 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash5, hashA5, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash5, 512,
+                      (const BitSequence *)hash5, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash5, 512,
+                      (const BitSequence *)hash5, 64 );
+   }
+#endif   
+   if ( hash6[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash6, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash6 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash6, hashA6, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash6, 512,
+                      (const BitSequence *)hash6, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash6, 512,
+                      (const BitSequence *)hash6, 64 );
+   }
+#endif   
+   if ( hash7[0] & 1 )
+   {
+      sph_gost512_init( &ctx.gost );
+      sph_gost512( &ctx.gost, (const void*)hash7, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash7 );
+   }
+   else
+#if defined (__VAES__)
+      memcpy( hash7, hashA7, 64 );
+#else
+   {
+      echo_full( &ctx.echo, (BitSequence *)hash7, 512,
+                      (const BitSequence *)hash7, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash7, 512,
+                      (const BitSequence *)hash7, 64 );
+   }
+#endif
+
+   intrlv_8x64_512( hash, hash0, hash1, hash2, hash3,
+                          hash4, hash5, hash6, hash7 );
+
+   skein512_8way_init( &ctx.skein );
+   skein512_8way_update( &ctx.skein, (const void*)hash, 64 );
+   skein512_8way_close( &ctx.skein, (void*)hash );
+
+   for ( int i = 0; i < 4; i++ )
+      casti_m512i( state, i ) = _mm512_xor_si512( casti_m512i( hash, i ),
+                                                  casti_m512i( hash, i+4 ) );
+}
+
+int scanhash_phi2_8way( struct work *work, uint32_t max_nonce,
+                        uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t _ALIGN(128) hash[16*8];
+   uint32_t _ALIGN(128) edata[36*8];
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   uint32_t *hash7 = &(hash[49]);  
+   const uint32_t Htarg = ptarget[7];
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 8;
+   uint32_t n = first_nonce;
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+   if ( bench )      ptarget[7] = 0x00ff;
+
+   phi2_has_roots = false;
+
+   for ( int i = 0; i < 36; i++ )
+   {
+      be32enc( &edata[i], pdata[i] );
+      edata[ i +   36 ] = edata[ i + 2*36 ] = edata[ i + 3*36 ] =
+      edata[ i + 4*36 ] = edata[ i + 5*36 ] = edata[ i + 6*36 ] =
+      edata[ i + 7*36 ] = edata[ i ];
+      if ( i >= 20 && pdata[i] ) phi2_has_roots = true;
+   }
+
+   edata[        19 ] = n;
+   edata[   36 + 19 ] = n+1;
+   edata[ 2*36 + 19 ] = n+2;
+   edata[ 3*36 + 19 ] = n+3;
+   edata[ 4*36 + 19 ] = n+4;
+   edata[ 5*36 + 19 ] = n+5;
+   edata[ 6*36 + 19 ] = n+6;
+   edata[ 7*36 + 19 ] = n+7;
+   
+   do {
+      phi2_8way_hash( hash, edata );
+
+      for ( int lane = 0; lane < 8; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg && !bench ) )
+      {
+         uint64_t _ALIGN(64) lane_hash[8];
+         extr_lane_8x64( lane_hash, hash, lane, 256 );
+         if ( valid_hash( lane_hash, ptarget ) )
+         {
+            be32enc( pdata + 19, n + lane );
+            submit_solution( work, lane_hash, mythr );
+         }
+      }
+      n += 8;
+      edata[        19 ] += 8;
+      edata[   36 + 19 ] += 8;
+      edata[ 2*36 + 19 ] += 8;
+      edata[ 3*36 + 19 ] += 8;
+      edata[ 4*36 + 19 ] += 8;
+      edata[ 5*36 + 19 ] += 8;
+      edata[ 6*36 + 19 ] += 8;
+      edata[ 7*36 + 19 ] += 8;
+   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
+   return 0;
+
+}
+
+#elif defined(PHI2_4WAY)

 typedef struct {
     cubehashParam           cube;
     jh512_4way_context      jh;
+#if  defined(__AES__)
     hashState_echo          echo;
-//     hashState_echo          echo2;
+#else
+     sph_echo512_context     echo;
+#endif
     sph_gost512_context     gost;
     skein512_4way_context   skein;
-} phi2_ctx_holder;
-/*
-phi2_ctx_holder phi2_ctx;
+} phi2_4way_ctx_holder;

-void init_phi2_ctx()
+phi2_4way_ctx_holder phi2_4way_ctx;
+
+void phi2_4way_hash(void *state, const void *input)
 {
-   cubehashInit( &phi2_ctx.cube, 512, 16, 32 );
-   sph_jh512_init(&phi2_ctx.jh);
-   init_echo( &phi2_ctx.echo1, 512 );
-   init_echo( &phi2_ctx.echo2, 512 );
-   sph_gost512_init(&phi2_ctx.gost);
-   sph_skein512_init(&phi2_ctx.skein);
-};
-*/
-void phi2_hash_4way( void *state, const void *input )
-{
-   uint32_t hash[4][16] __attribute__ ((aligned (64)));
-   uint32_t hashA[4][16] __attribute__ ((aligned (64)));
-   uint32_t hashB[4][16] __attribute__ ((aligned (64)));
-   uint32_t vhash[4*16] __attribute__ ((aligned (64)));
+	unsigned char _ALIGN(128) hash[64*4];
+   unsigned char _ALIGN(64) hash0[64];
+   unsigned char _ALIGN(64) hash1[64];
+   unsigned char _ALIGN(64) hash2[64];
+   unsigned char _ALIGN(64) hash3[64];
+   unsigned char _ALIGN(64) hash0A[64];
+   unsigned char _ALIGN(64) hash1A[64];
+   unsigned char _ALIGN(64) hash2A[64];
+   unsigned char _ALIGN(64) hash3A[64];
+   const int size = phi2_has_roots ? 144 : 80 ;
+   phi2_4way_ctx_holder ctx __attribute__ ((aligned (64)));

-//   unsigned char _ALIGN(128) hash[64];
-//	unsigned char _ALIGN(128) hashA[64];
-//	unsigned char _ALIGN(128) hashB[64];
+   cubehash_full( &ctx.cube, (byte*)hash0A, 512,
+                       (const byte*)input,          size );
+   cubehash_full( &ctx.cube, (byte*)hash1A, 512,
+                       (const byte*)input +   144, size );
+   cubehash_full( &ctx.cube, (byte*)hash2A, 512,
+                       (const byte*)input + 2*144, size );
+   cubehash_full( &ctx.cube, (byte*)hash3A, 512,
+                       (const byte*)input + 3*144, size );
+  
+	LYRA2RE( &hash0[ 0], 32, hash0A,    32, hash0A,    32, 1, 8, 8 );
+	LYRA2RE( &hash0[32], 32, hash0A+32, 32, hash0A+32, 32, 1, 8, 8 );
+   LYRA2RE( &hash1[ 0], 32, hash1A,    32, hash1A,    32, 1, 8, 8 );
+   LYRA2RE( &hash1[32], 32, hash1A+32, 32, hash1A+32, 32, 1, 8, 8 );
+   LYRA2RE( &hash2[ 0], 32, hash2A,    32, hash2A,    32, 1, 8, 8 );
+   LYRA2RE( &hash2[32], 32, hash2A+32, 32, hash2A+32, 32, 1, 8, 8 );
+   LYRA2RE( &hash3[ 0], 32, hash3A,    32, hash3A,    32, 1, 8, 8 );
+   LYRA2RE( &hash3[32], 32, hash3A+32, 32, hash3A+32, 32, 1, 8, 8 );

-   phi2_ctx_holder ctx __attribute__ ((aligned (64)));
-//  memcpy( &ctx, &phi2_ctx, sizeof(phi2_ctx) );
-
-   cubehashInit( &ctx.cube, 512, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*)hashB[0], (const byte*)input,
-                        phi2_has_roots ? 144 : 80 );
-   cubehashInit( &ctx.cube, 512, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*)hashB[1], (const byte*)input+144,
-                        phi2_has_roots ? 144 : 80 );
-   cubehashInit( &ctx.cube, 512, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*)hashB[2], (const byte*)input+288,
-                        phi2_has_roots ? 144 : 80 );
-   cubehashInit( &ctx.cube, 512, 16, 32 );
-   cubehashUpdateDigest( &ctx.cube, (byte*)hashB[3], (const byte*)input+432,
-                        phi2_has_roots ? 144 : 80 );
-
-	LYRA2RE( &hashA[0][0], 32, &hashB[0][0], 32, &hashB[0][0], 32, 1, 8, 8 );
-	LYRA2RE( &hashA[0][8], 32, &hashB[0][8], 32, &hashB[0][8], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[1][0], 32, &hashB[1][0], 32, &hashB[1][0], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[1][8], 32, &hashB[1][8], 32, &hashB[1][8], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[2][0], 32, &hashB[2][0], 32, &hashB[2][0], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[2][8], 32, &hashB[2][8], 32, &hashB[2][8], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[3][0], 32, &hashB[3][0], 32, &hashB[3][0], 32, 1, 8, 8 );
-   LYRA2RE( &hashA[3][8], 32, &hashB[3][8], 32, &hashB[3][8], 32, 1, 8, 8 );
-
-   intrlv_4x64( vhash, hashA[0], hashA[1], hashA[2], hashA[3], 512 );
+   intrlv_4x64_512( hash, hash0, hash1, hash2, hash3 );

   jh512_4way_init( &ctx.jh );
-   jh512_4way( &ctx.jh, vhash, 64 );
-   jh512_4way_close( &ctx.jh, vhash );
+   jh512_4way_update( &ctx.jh, (const void*)hash, 64 );
+	jh512_4way_close( &ctx.jh, (void*)hash );

-   dintrlv_4x64( hash[0], hash[1], hash[2], hash[3], vhash, 512 );
+   dintrlv_4x64_512( hash0, hash1, hash2, hash3, hash );

-   if ( hash[0][0] & 1 )
+   if ( hash0[0] & 1 )
  	{
      sph_gost512_init( &ctx.gost );
-      sph_gost512( &ctx.gost, (const void*)hash[0], 64 );
-	   sph_gost512_close( &ctx.gost, (void*)hash[0] );
+      sph_gost512( &ctx.gost, (const void*)hash0, 64 );
+	   sph_gost512_close( &ctx.gost, (void*)hash0 );
 	}
  	else
  	{
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[0],
-                          (const BitSequence *)hash[0], 512 );
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[0],
-                          (const BitSequence *)hash[0], 512 );
+      echo_full( &ctx.echo, (BitSequence *)hash0, 512,
+                      (const BitSequence *)hash0, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash0, 512,
+                      (const BitSequence *)hash0, 64 );
 	}
-
-   if ( hash[1][0] & 1 )
+   if ( hash1[0] & 1 )
   {
      sph_gost512_init( &ctx.gost );
-      sph_gost512( &ctx.gost, (const void*)hash[1], 64 );
-      sph_gost512_close( &ctx.gost, (void*)hash[1] );
+      sph_gost512( &ctx.gost, (const void*)hash1, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash1 );
   }
   else
   {
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[1],
-                          (const BitSequence *)hash[1], 512 );
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[1],
-                          (const BitSequence *)hash[1], 512 );
+      echo_full( &ctx.echo, (BitSequence *)hash1, 512,
+                      (const BitSequence *)hash1, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash1, 512,
+                      (const BitSequence *)hash1, 64 );
   }
-
-   if ( hash[2][0] & 1 )
+   if ( hash2[0] & 1 )
   {
      sph_gost512_init( &ctx.gost );
-      sph_gost512( &ctx.gost, (const void*)hash[2], 64 );
-      sph_gost512_close( &ctx.gost, (void*)hash[2] );
+      sph_gost512( &ctx.gost, (const void*)hash2, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash2 );
   }
   else
   {
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[2],
-                          (const BitSequence *)hash[2], 512 );
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[2],
-                          (const BitSequence *)hash[2], 512 );
+      echo_full( &ctx.echo, (BitSequence *)hash2, 512,
+                      (const BitSequence *)hash2, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash2, 512,
+                      (const BitSequence *)hash2, 64 );
   }
-
-   if ( hash[3][0] & 1 )
+   if ( hash3[0] & 1 )
   {
      sph_gost512_init( &ctx.gost );
-      sph_gost512( &ctx.gost, (const void*)hash[3], 64 );
-      sph_gost512_close( &ctx.gost, (void*)hash[3] );
+      sph_gost512( &ctx.gost, (const void*)hash3, 64 );
+      sph_gost512_close( &ctx.gost, (void*)hash3 );
   }
   else
   {
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[3],
-                          (const BitSequence *)hash[3], 512 );
-      init_echo( &ctx.echo, 512 );
-      update_final_echo ( &ctx.echo, (BitSequence *)hash[3],
-                          (const BitSequence *)hash[3], 512 );
+      echo_full( &ctx.echo, (BitSequence *)hash3, 512,
+                      (const BitSequence *)hash3, 64 );
+      echo_full( &ctx.echo, (BitSequence *)hash3, 512,
+                      (const BitSequence *)hash3, 64 );
   }

-   intrlv_4x64( vhash, hash[0], hash[1], hash[2], hash[3], 512 );
-   
+   intrlv_4x64_512( hash, hash0, hash1, hash2, hash3 );
+
   skein512_4way_init( &ctx.skein );
-	skein512_4way( &ctx.skein, vhash, 64 );
-	skein512_4way_close( &ctx.skein, vhash );
+	skein512_4way_update( &ctx.skein, (const void*)hash, 64 );
+	skein512_4way_close( &ctx.skein, (void*)hash );

-   for (int i=0; i<4; i++)
-   {
-      ( (uint64_t*)vhash    )[i] ^= ( (uint64_t*)vhash    )[i+4];
-      ( (uint64_t*)vhash+ 8 )[i] ^= ( (uint64_t*)vhash+ 8 )[i+4];
-      ( (uint64_t*)vhash+16 )[i] ^= ( (uint64_t*)vhash+16 )[i+4];
-      ( (uint64_t*)vhash+24 )[i] ^= ( (uint64_t*)vhash+24 )[i+4];
-   }
-//   for ( int i = 0; i < 4; i++ )
-//      casti_m256i( vhash, i ) = _mm256_xor_si256( casti_m256i( vhash, i   ),
-//                                                  casti_m256i( vhash, i+4 ) );

-	memcpy( state, vhash, 128 );
+   for ( int i = 0; i < 4; i++ )
+      casti_m256i( state, i ) = _mm256_xor_si256( casti_m256i( hash, i   ),
+                                                  casti_m256i( hash, i+4 ) );
 }

 int scanhash_phi2_4way( struct work *work, uint32_t max_nonce,
 	                     uint64_t *hashes_done, struct thr_info *mythr )
 {
-   uint32_t _ALIGN(128) hash[8];
-   uint32_t _ALIGN(128) edata[36];
-   uint32_t vdata[4][36] __attribute__ ((aligned (64)));
-   uint32_t *hash7 = &(hash[25]);
-   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t _ALIGN(128) hash[16*4];
+   uint32_t _ALIGN(128) edata[36*4];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
+   uint32_t *hash7 = &(hash[25]);   // 3*8+1
   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 4;
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
-
-   if(opt_benchmark){
-   	ptarget[7] = 0x00ff;
-   }
-
-// Data is not interleaved, but hash is.
-// any non-zero data at index 20 or above sets roots true.
-// Split up the operations, bswap first, then set roots.
-
-   phi2_has_roots = false;
-   for ( int i=0; i < 36; i++ )
-   {
-   be32enc(&edata[i], pdata[i]);
-   if (i >= 20 && pdata[i]) phi2_has_roots = true;
-   }
-/*
-   casti_m256i( vdata[0], 0 ) = mm256_bswap_32( casti_m256i( pdata, 0 ) );   
-   casti_m256i( vdata[0], 1 ) = mm256_bswap_32( casti_m256i( pdata, 1 ) );
-   casti_m256i( vdata[0], 2 ) = mm256_bswap_32( casti_m256i( pdata, 2 ) );
-   casti_m256i( vdata[0], 3 ) = mm256_bswap_32( casti_m256i( pdata, 3 ) );
-   casti_m128i( vdata[0], 8 ) = mm128_bswap_32( casti_m128i( pdata, 8 ) );
-   phi2_has_roots = mm128_anybits1( casti_m128i( vdata[0], 5 ) ) ||
-                    mm128_anybits1( casti_m128i( vdata[0], 6 ) ) ||
-                    mm128_anybits1( casti_m128i( vdata[0], 7 ) ) ||
-                    mm128_anybits1( casti_m128i( vdata[0], 8 ) );
-*/   
-
-   memcpy( vdata[0], edata, 144 );
-   memcpy( vdata[1], edata, 144 );
-   memcpy( vdata[2], edata, 144 );
-   memcpy( vdata[3], edata, 144 );
-
-   do {
-      be32enc( &vdata[0][19], n );
-      be32enc( &vdata[1][19], n+1 );
-      be32enc( &vdata[2][19], n+2 );
-      be32enc( &vdata[3][19], n+3 );
-
-      phi2_hash_4way( hash, vdata );
-
-      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_lane_solution( work, lane_hash, mythr, lane );
-          }
-       }
-       n += 4;
-    } while ( ( n < max_nonce - 4 ) && !work_restart[thr_id].restart );
-    *hashes_done = n - first_nonce + 1;
-    return 0;
-}
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+   if ( bench )   	ptarget[7] = 0x00ff;
   
-#endif  // PHI2_4WAY
+   phi2_has_roots = false;
+
+   for ( int i = 0; i < 36; i++ )
+   {
+	   be32enc( &edata[i], pdata[i] );
+      edata[ i+36 ] = edata[ i+72 ] = edata[ i+108 ] = edata[i];
+      if ( i >= 20 && pdata[i] ) phi2_has_roots = true;
+   }
+
+   edata[        19 ] = n;
+   edata[   36 + 19 ] = n+1;
+   edata[ 2*36 + 19 ] = n+2;
+   edata[ 3*36 + 19 ] = n+3;
+   
+   do {
+	   phi2_4way_hash( hash, edata );
+
+      for ( int lane = 0; lane < 4; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg && !bench ) )
+      {
+         uint64_t _ALIGN(64) lane_hash[8]; 
+         extr_lane_4x64( lane_hash, hash, lane, 256 );
+         if ( valid_hash( lane_hash, ptarget ) )
+         {
+            be32enc( pdata + 19, n + lane );
+            submit_solution( work, lane_hash, mythr );
+         }
+      }
+      edata[        19 ] += 4;
+      edata[   36 + 19 ] += 4;
+      edata[ 2*36 + 19 ] += 4;
+      edata[ 3*36 + 19 ] += 4;
+      n +=4;
+   } while ( (n < last_nonce) && !work_restart[thr_id].restart);
+   pdata[19] = n;
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
+#endif
+
--- a/algo/lyra2/phi2.c
+++ b/algo/lyra2/phi2.c
@@ -96,32 +96,29 @@ int scanhash_phi2( struct work *work, uint32_t max_nonce,
 	           uint64_t *hashes_done, struct thr_info *mythr )
 {
   uint32_t _ALIGN(128) hash[8];
-   uint32_t _ALIGN(128) endiandata[36];
+   uint32_t _ALIGN(128) edata[36];
   uint32_t *pdata = work->data;
   uint32_t *ptarget = work->target;
-   const uint32_t Htarg = ptarget[7];
   const uint32_t first_nonce = pdata[19];
   uint32_t n = first_nonce;
-   int thr_id = mythr->id;  // thr_id arg is deprecated
-
-   if(opt_benchmark){
-   	ptarget[7] = 0x00ff;
-   }
+   const int thr_id = mythr->id;
+   const bool bench = opt_benchmark;
+   if( bench )   	ptarget[7] = 0x00ff;

   phi2_has_roots = false;
-   for ( int i=0; i < 36; i++ )
+
+   for ( int i = 0; i < 36; i++ )
   {
-	   be32enc(&endiandata[i], pdata[i]);
+	   be32enc( &edata[i], pdata[i] );
      if ( i >= 20 && pdata[i] ) phi2_has_roots = true;
   }

   do {
-	be32enc( &endiandata[19], n );
-	phi2_hash( hash, endiandata );
-	if ( hash[7] < Htarg )
-   if ( fulltest( hash, ptarget ) && !opt_benchmark )
+	edata[19] = n;
+	phi2_hash( hash, edata );
+   if ( valid_hash( hash, ptarget ) && !opt_benchmark )
  	{
-       pdata[19] = n;
+       be32enc( pdata+19, n );
       submit_solution( work, hash, mythr );
   }
 	n++;
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Jay D Dee	cdd587537e	v3.14.3	2020-06-18 17:30:26 -04:00
Jay D Dee	51a1d91abd	v3.14.2	2020-05-30 21:20:44 -04:00
Jay D Dee	13563e2598	v3.14.1	2020-05-21 13:00:29 -04:00
Jay D Dee	9571f85d53	v3.14.0	2020-05-20 13:56:35 -04:00
Jay D Dee	0e69756634	v3.13.2-segwit-test	2020-05-18 18:17:27 -04:00
Jay D Dee	9653bca1e2	v3.13.1.1	2020-05-17 19:21:37 -04:00
Jay D Dee	1c0719e8a4	v3.13.1	2020-05-10 21:34:55 -04:00
Jay D Dee	8b4b4dc613	v3.13.0.1	2020-05-07 17:57:04 -04:00
Jay D Dee	e76feaced8	v3.13.0	2020-05-06 00:53:43 -04:00
Jay D Dee	5e088d00d0	v3.12.8.2	2020-04-24 21:18:56 -04:00
Jay D Dee	972d4d70db	v3.12.8.1	2020-04-17 16:12:45 -04:00
Jay D Dee	e96a6bd699	v3.12.8	2020-04-09 12:56:18 -04:00
Jay D Dee	fb9163185a	v3.12.7	2020-03-20 16:30:12 -04:00
Jay D Dee	6e8b8ed34f	v3.12.6.1	2020-03-07 14:11:06 -05:00
Jay D Dee	c0aadbcc99	v3.12.6	2020-03-05 18:43:20 -05:00
Jay D Dee	3da149418a	v3.12.5	2020-03-01 13:18:17 -05:00
Jay D Dee	720610cce5	v3.12.4.6	2020-02-28 18:20:32 -05:00
Jay D Dee	cedcf4d070	v3.12.4.5	2020-02-28 02:42:22 -05:00
Jay D Dee	81b50c3c71	v3.12.4.4	2020-02-25 14:07:32 -05:00
Jay D Dee	0e1e88f53e	v3.12.4.3	2020-02-24 21:35:19 -05:00
Jay D Dee	45c77a5c81	v3.12.4.2	2020-02-23 15:31:06 -05:00
Jay D Dee	dbce7e0721	v3.12.4.1	2020-02-22 18:06:39 -05:00
Jay D Dee	6d66051de6	v3.12.4	2020-02-21 16:34:53 -05:00
Jay D Dee	b93be8816a	v3.12.3.1	2020-02-18 12:05:47 -05:00
Jay D Dee	19b0ac6d5c	v3.12.3	2020-02-13 04:25:33 -05:00
Jay D Dee	3da2b958cf	v3.12.2	2020-02-09 13:30:40 -05:00
Jay D Dee	dc2f8d81d3	v3.12.1	2020-02-07 20:18:20 -05:00
Jay D Dee	fc97ef174a	v3.12.0.1	2020-02-06 22:50:20 -05:00
Jay D Dee	13523a12f9	v3.12.0	2020-02-05 22:50:58 -05:00
Jay D Dee	1b76cee239	v3.11.9	2020-02-04 01:31:59 -05:00
Jay D Dee	0681ca996d	v3.11.8	2020-01-30 03:47:11 -05:00
Jay D Dee	88f81fda0b	v3.11.7	2020-01-26 04:33:39 -05:00
Jay D Dee	103e6ad36c	v3.11.6	2020-01-23 00:11:08 -05:00