Pages:
Author

Topic: [ANN]: cpuminer-opt v3.8.8.1, open source optimized multi-algo CPU miner - page 45. (Read 444040 times)

newbie
Activity: 15
Merit: 0
is CPU mining zcoin (lyra2z) still profitable? I have i7-7800X 3.5 GHz giving me about 950-971 kH/s where my 1080Ti gives me 3300 kH/s. I could not find easy resources to see if CPU is profitable
legendary
Activity: 1470
Merit: 1114
Is there a possibility to make a SHA version with AES-AVX/AES-AVX2/4WAY to improve the tests?

There should be no problem adding 4way to the Ryzen build. 4way only needs AVX2 and all Ryzens
have it. Itcould become an issue in the future if some low-budget CPUs are built with SHA but not
AVX2.

Edit: At this time the only algo that can use both 4way and SHA is skein.
jr. member
Activity: 196
Merit: 1

cpuminer-opt-3.7.7-sha win

https://ufile.io/mkuq4


Thanks for that. Do you have a howto guide? I need to file it for when I finally upgrade my build environment

With your permission I will add your link to the OP.

I'm cross compiling cpuminer-opt for my own usage for some time (since I posted the instructions here)
You can link to my github
https://github.com/yuzi-co/miners/releases/download/1/cpuminer-opt-3.7.7-4ward.7z

It includes all the windows versions you release (compiled with openssl 1.1) + sha + all necessary dlls

One thing I couldn't resolve is that neoscrypt always fails to run (same in My9bot's version)


Thank you!

I did some benchmarks (Ryzen 7 1700 at 3.7GHz) with your package

SHA256T - SHA 16 THREADS
Code:
[2017-12-18 14:20:40] 16 miner threads started, using 'sha256t' algorithm.
[2017-12-18 14:20:40] Total: 1572.86 kH, 22.62 MH/s
[2017-12-18 14:20:45] Total: 278.74 MH, 58.72 MH/s
[2017-12-18 14:20:50] Total: 243.37 MH, 58.17 MH/s
[2017-12-18 14:20:55] Total: 276.47 MH, 58.41 MH/s
[2017-12-18 14:21:00] Total: 284.05 MH, 58.46 MH/s
[2017-12-18 14:21:05] Total: 279.11 MH, 59.02 MH/s
[2017-12-18 14:21:10] Total: 266.39 MH, 59.40 MH/s
[2017-12-18 14:21:15] Total: 292.94 MH, 59.38 MH/s

SHA256T - SHA 8 THREADS
Code:
[2017-12-18 14:23:25] 8 miner threads started, using 'sha256t' algorithm.
[2017-12-18 14:23:25] Total: 1048.58 kH, 20.97 MH/s
[2017-12-18 14:23:30] Total: 176.50 MH, 41.74 MH/s
[2017-12-18 14:23:35] Total: 188.22 MH, 41.66 MH/s
[2017-12-18 14:23:40] Total: 187.26 MH, 41.78 MH/s
[2017-12-18 14:23:45] Total: 203.56 MH, 41.76 MH/s
[2017-12-18 14:23:50] Total: 208.79 MH, 41.86 MH/s


SKEIN - SHA 16 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 SHA.
Start mining with SSE2 SHA.

[2017-12-18 14:24:40] 16 miner threads started, using 'skein' algorithm.
[2017-12-18 14:24:41] Total: 524.29 kH, 1483.29 kH/s
[2017-12-18 14:24:45] Total: 61.07 MH, 23.24 MH/s
[2017-12-18 14:24:50] Total: 111.79 MH, 23.29 MH/s
[2017-12-18 14:24:55] Total: 116.39 MH, 23.27 MH/s
[2017-12-18 14:25:00] Total: 115.93 MH, 23.33 MH/s
[2017-12-18 14:25:05] Total: 115.96 MH, 23.22 MH/s

SKEIN - SHA 8 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 SHA.
Start mining with SSE2 SHA.

[2017-12-18 14:25:38] 8 miner threads started, using 'skein' algorithm.
[2017-12-18 14:25:38] Total: 1572.86 kH, 6930.61 kH/s
[2017-12-18 14:25:43] Total: 79.42 MH, 18.37 MH/s
[2017-12-18 14:25:48] Total: 82.68 MH, 18.31 MH/s
[2017-12-18 14:25:53] Total: 89.29 MH, 18.31 MH/s
[2017-12-18 14:25:58] Total: 82.48 MH, 18.30 MH/s

M7M - SHA 16 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AES AVX AVX2 SHA.
Start mining with AES AVX2 SHA.

[2017-12-18 14:26:51] 16 miner threads started, using 'm7m' algorithm.
[2017-12-18 14:26:58] Total: 1048.58 kH, 153.21 kH/s
[2017-12-18 14:27:03] Total: 1475.41 kH, 298.13 kH/s
[2017-12-18 14:27:08] Total: 1462.74 kH, 299.35 kH/s
[2017-12-18 14:27:13] Total: 1473.17 kH, 293.62 kH/s
[2017-12-18 14:27:19] Total: 1470.94 kH, 292.78 kH/s
[2017-12-18 14:27:22] Total: 1446.18 kH, 287.93 kH/s
[2017-12-18 14:27:23] Total: 1314.45 kH, 293.83 kH/s

M7M - SHA 8 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AES AVX AVX2 SHA.
Start mining with AES AVX2 SHA.

[2017-12-18 14:27:44] 8 miner threads started, using 'm7m' algorithm.
[2017-12-18 14:27:48] Total: 393.22 kH, 87.37 kH/s
[2017-12-18 14:27:49] Total: 742.73 kH, 226.83 kH/s
[2017-12-18 14:27:54] Total: 453.60 kH, 229.37 kH/s
[2017-12-18 14:27:59] Total: 1144.11 kH, 228.59 kH/s
[2017-12-18 14:28:04] Total: 1099.87 kH, 224.53 kH/s
[2017-12-18 14:28:09] Total: 1091.55 kH, 222.76 kH/s

LBRY - SHA 16 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AVX AVX2 SHA.
Start mining with AVX2 SHA.

[2017-12-18 14:29:03] 16 miner threads started, using 'lbry' algorithm.
[2017-12-18 14:29:03] Total: 393.22 kH, 1994.04 kH/s
[2017-12-18 14:29:08] Total: 28.31 MH, 10.36 MH/s
[2017-12-18 14:29:13] Total: 46.71 MH, 10.34 MH/s
[2017-12-18 14:29:18] Total: 49.09 MH, 10.32 MH/s
[2017-12-18 14:29:23] Total: 51.59 MH, 10.33 MH/s
[2017-12-18 14:29:28] Total: 51.65 MH, 10.29 MH/s
[2017-12-18 14:29:33] Total: 50.62 MH, 10.32 MH/s

LBRY - SHA 8 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AVX AVX2 SHA.
Start mining with AVX2 SHA.

[2017-12-18 14:29:54] 8 miner threads started, using 'lbry' algorithm.
[2017-12-18 14:29:54] Total: 655.36 kH, 4567.30 kH/s
[2017-12-18 14:29:59] Total: 18.29 MH, 7300.38 kH/s
[2017-12-18 14:30:04] Total: 36.52 MH, 7297.99 kH/s
[2017-12-18 14:30:09] Total: 36.49 MH, 7293.69 kH/s
[2017-12-18 14:30:14] Total: 36.46 MH, 7234.66 kH/s
[2017-12-18 14:30:19] Total: 36.17 MH, 7211.15 kH/s

MYR-GR - SHA 16 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AES AVX AVX2 SHA.
Start mining with AES AVX2 SHA.

[2017-12-18 14:31:08] 16 miner threads started, using 'myr-gr' algorithm.
[2017-12-18 14:31:08] Total: 3145.73 kH, 8517.45 kH/s
[2017-12-18 14:31:13] Total: 22.12 MH, 11.17 MH/s
[2017-12-18 14:31:18] Total: 47.95 MH, 11.11 MH/s
[2017-12-18 14:31:23] Total: 44.93 MH, 11.17 MH/s
[2017-12-18 14:31:26] Total: 44.12 MH, 11.06 MH/s

MYR-GR - SHA 8 THREADS
Code:
         **********  cpuminer-opt 3.7.7  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700 Eight-Core Processor          .
SW built on Dec 17 2017 with GCC 6.2.1.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AES AVX AVX2 SHA.
Start mining with AES AVX2 SHA.

[2017-12-18 14:32:04] 8 miner threads started, using 'myr-gr' algorithm.
[2017-12-18 14:32:04] Total: 1107.11 kH, 5887.11 kH/s
[2017-12-18 14:32:09] Total: 31.06 MH, 9538.90 kH/s
[2017-12-18 14:32:14] Total: 34.63 MH, 9553.20 kH/s
[2017-12-18 14:32:19] Total: 37.34 MH, 9566.10 kH/s
[2017-12-18 14:32:24] Total: 31.56 MH, 9568.59 kH/s

Is there a possibility to make a SHA version with AES-AVX/AES-AVX2/4WAY to improve the tests?
member
Activity: 473
Merit: 18
I found the cross-compile instructions

https://bitcointalksearch.org/topic/m.23580581

since then i compiled it all to a script:
Code:
#!/bin/bash
OPENSSL_VER='1.1.0g'
CURL_VER='7.56.1'
GMP_VER='6.1.2'
DIR=$(pwd)
CORES=$(grep -c ^processor /proc/cpuinfo)

sudo apt-get install build-essential libz-mingw-w64-dev libz-mingw-w64 git autoconf -y

[[ -f openssl-${OPENSSL_VER}.tar.gz ]] || wget https://www.openssl.org/source/openssl-${OPENSSL_VER}.tar.gz
[[ -d openssl-${OPENSSL_VER} ]] || tar xvfz openssl-${OPENSSL_VER}.tar.gz
cd openssl-${OPENSSL_VER}
./Configure mingw64 shared --cross-compile-prefix=x86_64-w64-mingw32- || exit 1
make -j ${CORES} || exit 1

cd ${DIR}
[[ -f curl-${CURL_VER}.tar.bz2 ]] || wget https://curl.haxx.se/download/curl-${CURL_VER}.tar.bz2
[[ -d curl-${CURL_VER} ]] || tar xvfj curl-${CURL_VER}.tar.bz2
cd curl-${CURL_VER}
./configure --with-winssl --with-winidn --host=x86_64-w64-mingw32 || exit 1
make -j ${CORES} || exit 1

cd ${DIR}
[[ -f gmp-${GMP_VER}.tar.bz2 ]] || wget https://gmplib.org/download/gmp/gmp-${GMP_VER}.tar.bz2
[[ -d gmp-${GMP_VER} ]] || tar xvfj gmp-${GMP_VER}.tar.bz2
cd gmp-${GMP_VER}
./configure --host=x86_64-w64-mingw32 || exit 1
make -j ${CORES} || exit 1

cd ${DIR}
[[ -d cpuminer-opt ]] || git clone https://github.com/JayDDee/cpuminer-opt.git
cd cpuminer-opt
ln -s ../gmp-${GMP_VER}/gmp.h

cat > winbuild-cross.sh << EOF
#!/bin/bash

export LDFLAGS="-L${DIR}/curl-${CURL_VER}/lib/.libs -L${DIR}/gmp-${GMP_VER}/.libs -L${DIR}/openssl-${OPENSSL_VER}"
F="--with-curl=${DIR}/curl-${CURL_VER} --with-crypto=${DIR}/openssl-${OPENSSL_VER} --host=x86_64-w64-mingw32"

sed -i 's/"-lpthread"/"-lpthreadGC2"/g' configure.ac


mkdir release
cp /usr/x86_64-w64-mingw32/lib/zlib1.dll release/
cp /usr/x86_64-w64-mingw32/lib/libwinpthread-1.dll release/
cp /usr/lib/gcc/x86_64-w64-mingw32/6.2-win32/libstdc++-6.dll release/
cp /usr/lib/gcc/x86_64-w64-mingw32/6.2-win32/libgcc_s_seh-1.dll release/
cp ${DIR}/openssl-${OPENSSL_VER}/libcrypto-1_1-x64.dll release/
cp ${DIR}/curl-${CURL_VER}/lib/.libs/libcurl-4.dll release/


make distclean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=core-avx2 -Wall -DFOUR_WAY" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-4way.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=core-avx2 -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-aes-avx2.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=znver1 -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-aes-sha.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=corei7-avx -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-aes-avx.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -maes -msse4.2 -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-aes-sse42.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=corei7 -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-sse42.exe

make clean || echo clean
rm -f config.status
./autogen.sh || echo done
CFLAGS="-O3 -march=core2 -Wall" ./configure \$F
make -j ${CORES}
strip -s cpuminer.exe
mv cpuminer.exe release/cpuminer-sse2.exe

EOF

chmod a+x winbuild-cross.sh
./winbuild-cross.sh
legendary
Activity: 1470
Merit: 1114
I have i7-7800X CPU @ 3.50GHz

Trying to see which CPU miner is best for lyra2z... anyone know...? I tried 4way and AES-avx2. they seem to get the same KH and KH/s.
For aes-sse42, it seems i get lower.

From personal test, it seems like aes-avx2 is way to go. can anyone confirm? also, what would be safe CPU temperature for CPU mining?

AVX2 is the way to go, 4 way does nothing for lyra2. You can also try reducing the number of threads, it might improve cache
performance. The cooler the CPU the better. Make sure the fan is at 100%, it's cheaper to replace a worn out fan than a burned out CPU.

I see. Do I have to modify my bat file to run less threads? sorry this is my first time using cpuminer so wanted to see what I need to try to see if I can get the maximum performance.

The option needs to be on the cpuminer command line or in the config file if you use one. If you run cpuminer from a bat
file then yes, you need to add the option to the comand line in the bat file.
legendary
Activity: 1470
Merit: 1114

cpuminer-opt-3.7.7-sha win

https://ufile.io/mkuq4


Thanks for that. Do you have a howto guide? I need to file it for when I finally upgrade my build environment

With your permission I will add your link to the OP.

I'm cross compiling cpuminer-opt for my own usage for some time (since I posted the instructions here)
You can link to my github
https://github.com/yuzi-co/miners/releases/download/1/cpuminer-opt-3.7.7-4ward.7z

It includes all the windows versions you release (compiled with openssl 1.1) + sha + all necessary dlls

One thing I couldn't resolve is that neoscrypt always fails to run (same in My9bot's version)


I found the cross-compile instructions

https://bitcointalksearch.org/topic/m.23580581
newbie
Activity: 15
Merit: 0
I have i7-7800X CPU @ 3.50GHz

Trying to see which CPU miner is best for lyra2z... anyone know...? I tried 4way and AES-avx2. they seem to get the same KH and KH/s.
For aes-sse42, it seems i get lower.

From personal test, it seems like aes-avx2 is way to go. can anyone confirm? also, what would be safe CPU temperature for CPU mining?

AVX2 is the way to go, 4 way does nothing for lyra2. You can also try reducing the number of threads, it might improve cache
performance. The cooler the CPU the better. Make sure the fan is at 100%, it's cheaper to replace a worn out fan than a burned out CPU.

I see. Do I have to modify my bat file to run less threads? sorry this is my first time using cpuminer so wanted to see what I need to try to see if I can get the maximum performance.
full member
Activity: 239
Merit: 100

cpuminer-opt-3.7.7-sha win

https://ufile.io/mkuq4


Thanks for that. Do you have a howto guide? I need to file it for when I finally upgrade my build environment

With your permission I will add your link to the OP.

Oh please let this happen!  Been waiting for SHA optimization for Windows since it was available for Linux.

While I understand the link is to the SHA enabled Win EXE, I'm a little weary about using it because it's not from the dev themselves though.

dont use it its ok for me
member
Activity: 388
Merit: 13
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.

I confirm major performance increase with Ryzen in scrypt adding --cpu-affinity 0x5555

From 22 H/s to 29 H/s with 8 threads and from 28 to 35 H/s with 15 threads!
member
Activity: 473
Merit: 18

cpuminer-opt-3.7.7-sha win

https://ufile.io/mkuq4


Thanks for that. Do you have a howto guide? I need to file it for when I finally upgrade my build environment

With your permission I will add your link to the OP.

I'm cross compiling cpuminer-opt for my own usage for some time (since I posted the instructions here)
You can link to my github
https://github.com/yuzi-co/miners/releases/download/1/cpuminer-opt-3.7.7-4ward.7z

It includes all the windows versions you release (compiled with openssl 1.1) + sha + all necessary dlls

One thing I couldn't resolve is that neoscrypt always fails to run (same in My9bot's version)
full member
Activity: 187
Merit: 100
Cryptocurrency enthusiast
The cooler the CPU the better. Make sure the fan is at 100%, it's cheaper to replace a worn out fan than a burned out CPU.

Hi guys. For using AVX/AVX2 in mining safely on latest Intel CPUs with full thread/core count, use Z170/Z270/Z370 chipset with "avx offset" bios setting. So, e.g. your 4GHz cpu will be running at 3.6-3.8GHz while mining with avx/avx2/4-way on all cores/threads.

To (partly) avoid speed degradation, lock turbo boost speed (if you do not want to overclock, of course) in bios - set max performance plan, disable c-states, disable speed-step. This is supported on non-overclocking chipsets as well (e.g. H270, B250).

Simple benchs for i5-7600 (non-k) for lyra2z330 AVX2 with -t 2:

1) Stock settings (base clock 3.5GHz, goes to 4.0GHz for 2 busy cores) - ~812-824H/s
2) Stock settings + speed-step OFF (base clock 3.6GHz, turbo boost on, goes to 4.0GHz for 2 busy cores) - ~829-836H/s
3) Performance bios profile + speed-step OFF + c-states OFF (base clock 3.9GHz @ all cores, turbo boost on, but it does not go to 4.0GHz) - ~844-861H/s

I may note that this CPU goes to 98C (2C to Tj Tongue) under lyra* with -t 4 and avx2/4-way under 150W TDP tower cooler, so avx offset helps to keep it @80-81C at 3.5-3.6GHz.

So, this just shows that turbo boost reacts slowly, and should be locked for a mining PC. I've tested for cryptonote as well, but the diff is much less - ~0.5% (as there's perfectly stable cores utilization there - no jumping load at all).
legendary
Activity: 1470
Merit: 1114
I have i7-7800X CPU @ 3.50GHz

Trying to see which CPU miner is best for lyra2z... anyone know...? I tried 4way and AES-avx2. they seem to get the same KH and KH/s.
For aes-sse42, it seems i get lower.

From personal test, it seems like aes-avx2 is way to go. can anyone confirm? also, what would be safe CPU temperature for CPU mining?

AVX2 is the way to go, 4 way does nothing for lyra2. You can also try reducing the number of threads, it might improve cache
performance. The cooler the CPU the better. Make sure the fan is at 100%, it's cheaper to replace a worn out fan than a burned out CPU.
hero member
Activity: 677
Merit: 500

cpuminer-opt-3.7.7-sha win

https://ufile.io/mkuq4


Thanks for that. Do you have a howto guide? I need to file it for when I finally upgrade my build environment

With your permission I will add your link to the OP.

Oh please let this happen!  Been waiting for SHA optimization for Windows since it was available for Linux.

While I understand the link is to the SHA enabled Win EXE, I'm a little weary about using it because it's not from the dev themselves though.
newbie
Activity: 15
Merit: 0
I have i7-7800X CPU @ 3.50GHz

Trying to see which CPU miner is best for lyra2z... anyone know...? I tried 4way and AES-avx2. they seem to get the same KH and KH/s.
For aes-sse42, it seems i get lower.

From personal test, it seems like aes-avx2 is way to go. can anyone confirm? also, what would be safe CPU temperature for CPU mining?

legendary
Activity: 1470
Merit: 1114
The results for nist5 on Ryzen are baffling. I get much better performance with i7-6700K
using 4way and all threads. It's even more baffling that the tribus results on Ryzen are
consistent with Intel. They share a lot of code.

4way 8 threads: 2720
4way 4 threads: 2205
1way 8 threads: 2055
1way 4 threads: 1850
jr. member
Activity: 196
Merit: 1
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.

In fact. Joblo, is there an updated algo list that receive boost from SHA hardware acceleration? I found a little list some pages before:

Quote
sha256t, lbry, skein, myr-groestl, m7m.

Are there more algos?

I converted all of them at the time and I don't recall any new algos that can use it.

What about nist5? Can you try that again? I'd like to understand what's going on there.
I get good performance on my Intel.

I reduced overclock (to keep everything cold). New results with NIST5:

NIST5 4way 8 threads with --cpu-affinity 0x5555
Code:
[2017-12-17 20:33:17] nist5 block 14635, diff 14699.054
[2017-12-17 20:33:27] CPU #6: 2097.15 kH, 209.42 kH/s
[2017-12-17 20:33:27] CPU #2: 2097.15 kH, 207.48 kH/s
[2017-12-17 20:33:27] CPU #5: 2097.15 kH, 205.79 kH/s
[2017-12-17 20:33:27] CPU #7: 2097.15 kH, 205.61 kH/s
[2017-12-17 20:33:27] CPU #4: 2097.15 kH, 204.96 kH/s
[2017-12-17 20:33:27] CPU #1: 2097.15 kH, 204.46 kH/s
[2017-12-17 20:33:27] CPU #0: 2097.15 kH, 204.01 kH/s
[2017-12-17 20:33:27] CPU #3: 2097.15 kH, 199.72 kH/s

NIST5 16 threads
Code:
[2017-12-17 20:47:55] nist5 block 14649, diff 22837.326
[2017-12-17 20:47:55] CPU #2: 667.71 kH, 121.83 kH/s
[2017-12-17 20:47:55] CPU #3: 672.92 kH, 122.76 kH/s
[2017-12-17 20:47:55] CPU #0: 454.52 kH, 83.08 kH/s
[2017-12-17 20:47:55] CPU #1: 653.82 kH, 119.54 kH/s
[2017-12-17 20:47:55] CPU #14: 647.14 kH, 118.14 kH/s
[2017-12-17 20:47:55] CPU #7: 657.04 kH, 119.95 kH/s
[2017-12-17 20:47:55] CPU #6: 635.59 kH, 116.06 kH/s
[2017-12-17 20:47:55] CPU #11: 681.85 kH, 124.55 kH/s
[2017-12-17 20:47:55] CPU #5: 682.78 kH, 124.85 kH/s
[2017-12-17 20:47:55] CPU #4: 570.23 kH, 104.24 kH/s
[2017-12-17 20:47:55] CPU #12: 565.09 kH, 103.26 kH/s
[2017-12-17 20:47:55] CPU #10: 681.59 kH, 124.53 kH/s
[2017-12-17 20:47:55] CPU #8: 617.07 kH, 112.53 kH/s
[2017-12-17 20:47:55] CPU #9: 684.27 kH, 124.89 kH/s
[2017-12-17 20:47:55] CPU #15: 669.73 kH, 122.23 kH/s
[2017-12-17 20:47:55] CPU #13: 642.05 kH, 117.12 kH/s

NIST5 AES-AVX2 8 threads with --cpu-affinity 0x5555
Code:
[2017-12-17 20:59:30] nist5 block 14655, diff 22762.164
[2017-12-17 20:59:36] CPU #4: 2097.15 kH, 369.94 kH/s
[2017-12-17 20:59:36] CPU #5: 2097.15 kH, 365.38 kH/s
[2017-12-17 20:59:36] CPU #7: 2097.15 kH, 365.42 kH/s
[2017-12-17 20:59:36] CPU #6: 2097.15 kH, 365.26 kH/s
[2017-12-17 20:59:36] CPU #3: 2097.15 kH, 365.13 kH/s
[2017-12-17 20:59:36] CPU #0: 2097.15 kH, 359.60 kH/s
[2017-12-17 20:59:36] CPU #1: 2097.15 kH, 359.14 kH/s
[2017-12-17 20:59:36] CPU #2: 2097.15 kH, 356.60 kH/s

EDIT1:
NIST5 AES-AVX 8 threads with --cpu-affinity 0x5555
Code:
[2017-12-17 21:02:19] nist5 block 14657, diff 22797.808
[2017-12-17 21:02:26] CPU #6: 2097.15 kH, 374.14 kH/s
[2017-12-17 21:02:26] CPU #5: 2097.15 kH, 373.53 kH/s
[2017-12-17 21:02:26] CPU #2: 2097.15 kH, 370.56 kH/s
[2017-12-17 21:02:26] CPU #7: 2097.15 kH, 369.28 kH/s
[2017-12-17 21:02:26] CPU #4: 2097.15 kH, 367.82 kH/s
[2017-12-17 21:02:26] CPU #0: 2097.15 kH, 367.53 kH/s
[2017-12-17 21:02:26] CPU #1: 2097.15 kH, 365.98 kH/s
[2017-12-17 21:02:26] CPU #3: 2097.15 kH, 365.38 kH/s

EDIT2:
Quote from: My9bot
what do you need?

Wich algos are affected by SHA acceleration?
full member
Activity: 239
Merit: 100
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.

In fact. Joblo, is there an updated algo list that receive boost from SHA hardware acceleration? I found a little list some pages before:

Quote
sha256t, lbry, skein, myr-groestl, m7m.

Are there more algos?

I converted all of them at the time and I don't recall any new algos that can use it.

What about nist5? Can you try that again? I'd like to understand what's going on there.
I get good performance on my Intel.

what do you need?
legendary
Activity: 1470
Merit: 1114
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.

In fact. Joblo, is there an updated algo list that receive boost from SHA hardware acceleration? I found a little list some pages before:

Quote
sha256t, lbry, skein, myr-groestl, m7m.

Are there more algos?

I converted all of them at the time and I don't recall any new algos that can use it.

What about nist5? Can you try that again? I'd like to understand what's going on there.
I get good performance on my Intel.
jr. member
Activity: 196
Merit: 1
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.

In fact. Joblo, is there an updated algo list that receive boost from SHA hardware acceleration? I found a little list some pages before:

Quote
sha256t, lbry, skein, myr-groestl, m7m.

Are there more algos?
legendary
Activity: 1470
Merit: 1114
Ya, the default affinity was choosing virtual threads instead physical ones. Damn! 6MH/s!

All Ryzen users should take note. Intel chooses one thread per core before using HT.
Pages:
Jump to: