cpuminer-opt-3.8.0 released. https://github.com/JayDDee/cpuminer-opt/releases/tag/v3.8.04way no longer a seperate feature, included in AVX2. Added x16r algo for Ravencoin, anime algo for Animecoin. More 4way optimizations for X13 and up. Tweaked CPU affinity to better support more than 64 CPUs. Fixed compile problem on some old AMD CPUs. This release is a major milestone for cpuminer-opt. It essentially marks the end of 4way phase 1. 15 of the 17 functions in X17 have some form of optimization. Only fugue and whirlpool still use unoptimized code and 4 way seems impossible on these functions using SIMD. |
|
|
|
I don't own any AMDs si I don't have any personal experience. Based on the specs it should work well with AES and AVX. |
|
|
|
Testing x16r now, can't argue about fees. |
|
|
|
|
If someone really wants me to provide a special binary build I can be bribed. My tip jar has been
empty for a while. I expected some consideration for the hard work implementing 4way,
not to mention my quick response to real issues.
I can build all possible combinations of x86_64 architectures from SSE2 to AVX2 with and without
AES or SHA if sufficiently motivated. I might even put some more effort into figuring out how to
build for the older AMD CPUs.
|
|
|
|
I think there should be an SSE2-AVX EXE for i3 proccesors under windows , I saw it working only in linux,
I need a better reason than "I think". Show me the performance difference from just sse2. I'm trying to reduce then number of builds, not increase them. Old and crippled CPUs do not interest me. I want to focus on AVX2 and AVX512. |
|
|
|
Does this miner not have a web api? I've seen that it binds to port 4048 and when I try to access the api from my browser I get errors every time. localhost:4048 isn't working.
No, just a raw socket. telnet localhost 4048
help
As soon as I connect with telnet and hit any key it disconnects. I can't type help. Don't hit any key just enter a comand. $ telnet 127.0.0.1 4048
Trying 127.0.0.1...
Connected to 127.0.0.1.
Escape character is '^]'.
help
summary
threads
seturl
quit
|Connection closed by foreign host.
|
|
|
|
Does this miner not have a web api? I've seen that it binds to port 4048 and when I try to access the api from my browser I get errors every time. localhost:4048 isn't working.
No, just a raw socket. telnet localhost 4048
help
|
|
|
|
Managed to mine Yenten anyway, using the default miner. Didn't do that from the start since the tutorial I was following was pointing me to this one, cpuminer-opt, which is much more complete but incompatible with my CPU. Off course I went through all the hassle of figuring out how to compile and run the default miner, just to then find out that the pool I chose (absolutely at random) provides you with a neatly pre-packaged, pre-compiled debian executable. Basically, I did every single step in the most useless and time consuming way although I now have a rough knowledge of how to compile stuff on Linux, something which (as you may have noted) has always eluded me. Back to the point, if you're still interested in a fix for this particular AMD CPU... Can you check if other files have similar errors with avxdefs? I don't understand your question completely: are you asking me if I see similar errors within other files during compilation? As far as I can tell, no: avxdefs is the only file giving errors (other stuff only looks like warnings). I did another run, configuring with -march=core2, so that I could give you a complete transcript of the compile report. Here: https://pastebin.com/raw/ckhcAcYDPlease let me know if I didn't understand your question correctly and you're asking me for something else. Thanks for posting the complete compile session. Cryptonight is the only problem so my fix should work. For yenten the default miner is probably equal to cpuminer-opt. |
|
|
|
I'm having a similar error to that of @shvch and ran the same commands to install the required depencies. Total newbie here. I believe this is the relevant portion of errors: In file included from algo/cryptonight/cryptonight-aesni.c:6:0:
./avxdefs.h: In function mm_byteswap_64:
./avxdefs.h:296:3: warning: implicit declaration of function _mm_shuffle_epi8 [-Wimplicit-function-declaration]
./avxdefs.h:296:3: error: incompatible types when returning type int but __m128i was expected
./avxdefs.h: In function mm_byteswap_32:
./avxdefs.h:303:3: error: incompatible types when returning type int but __m128i was expected
./avxdefs.h: In function mm_byteswap_16:
./avxdefs.h:310:3: error: incompatible types when returning type int but __m128i was expected
algo/cryptonight/cryptonight-aesni.c: At top level:
algo/cryptonight/cryptonight-aesni.c:107:33: warning: ctx defined but not used [-Wunused-variable]
mv -f algo/cryptonight/.deps/cpuminer-cryptonight-common.Tpo algo/cryptonight/.deps/cpuminer-cryptonight-common.Po
gcc -std=gnu99 -DHAVE_CONFIG_H -I. -Iyes/include -fno-strict-aliasing -I. -Iyes/include -Wno-pointer-sign -Wno-pointer-to-int-cast -O3 -march=native -Wall -Iyes/include -MT algo/cubehash/cpuminer-sph_cubehash.o -MD -MP -MF algo/cubehash/.deps/cpuminer-sph_cubehash.Tpo -c -o algo/cubehash/cpuminer-sph_cubehash.o `test -f 'algo/cubehash/sph_cubehash.c' || echo './'`algo/cubehash/sph_cubehash.c
make[2]: *** [algo/cryptonight/cpuminer-cryptonight-aesni.o] Error 1
make[2]: *** Waiting for unfinished jobs....
mv -f algo/cryptonight/.deps/cpuminer-cryptonight.Tpo algo/cryptonight/.deps/cpuminer-cryptonight.Po
mv -f algo/bmw/.deps/cpuminer-sph_bmw.Tpo algo/bmw/.deps/cpuminer-sph_bmw.Po
mv -f algo/cubehash/.deps/cpuminer-sph_cubehash.Tpo algo/cubehash/.deps/cpuminer-sph_cubehash.Po
make[2]: Leaving directory `/root/cpuminer-opt'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/cpuminer-opt'
make: *** [all] Error 2
strip: 'cpuminer': No such file
Your CPU May fall into that AMD black hole where SIMD support was inconsistent. I recently optimized bytswap to use shuffle which is available in SSSE3. Your CPU's specs have SSSE3 but apparently not support for _mm_shuffle_epi8. Have you tried compiling with -march=core2? I'll try to come up with a fix for next release but it's hard to get motivated for old CPUs, especially old AMD CPUs. I may have a simple solution as long as cryptonight is the only culprit. Can you check if other files have similar errors with avxdefs? |
|
|
|
I'm having a similar error to that of @shvch and ran the same commands to install the required depencies. Total newbie here. I believe this is the relevant portion of errors: In file included from algo/cryptonight/cryptonight-aesni.c:6:0:
./avxdefs.h: In function mm_byteswap_64:
./avxdefs.h:296:3: warning: implicit declaration of function _mm_shuffle_epi8 [-Wimplicit-function-declaration]
./avxdefs.h:296:3: error: incompatible types when returning type int but __m128i was expected
./avxdefs.h: In function mm_byteswap_32:
./avxdefs.h:303:3: error: incompatible types when returning type int but __m128i was expected
./avxdefs.h: In function mm_byteswap_16:
./avxdefs.h:310:3: error: incompatible types when returning type int but __m128i was expected
algo/cryptonight/cryptonight-aesni.c: At top level:
algo/cryptonight/cryptonight-aesni.c:107:33: warning: ctx defined but not used [-Wunused-variable]
mv -f algo/cryptonight/.deps/cpuminer-cryptonight-common.Tpo algo/cryptonight/.deps/cpuminer-cryptonight-common.Po
gcc -std=gnu99 -DHAVE_CONFIG_H -I. -Iyes/include -fno-strict-aliasing -I. -Iyes/include -Wno-pointer-sign -Wno-pointer-to-int-cast -O3 -march=native -Wall -Iyes/include -MT algo/cubehash/cpuminer-sph_cubehash.o -MD -MP -MF algo/cubehash/.deps/cpuminer-sph_cubehash.Tpo -c -o algo/cubehash/cpuminer-sph_cubehash.o `test -f 'algo/cubehash/sph_cubehash.c' || echo './'`algo/cubehash/sph_cubehash.c
make[2]: *** [algo/cryptonight/cpuminer-cryptonight-aesni.o] Error 1
make[2]: *** Waiting for unfinished jobs....
mv -f algo/cryptonight/.deps/cpuminer-cryptonight.Tpo algo/cryptonight/.deps/cpuminer-cryptonight.Po
mv -f algo/bmw/.deps/cpuminer-sph_bmw.Tpo algo/bmw/.deps/cpuminer-sph_bmw.Po
mv -f algo/cubehash/.deps/cpuminer-sph_cubehash.Tpo algo/cubehash/.deps/cpuminer-sph_cubehash.Po
make[2]: Leaving directory `/root/cpuminer-opt'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/cpuminer-opt'
make: *** [all] Error 2
strip: 'cpuminer': No such file
I'm running this on an AMD Opteron 4284 which, according to "cat /proc/cpuinfo" has "SSE2" enabled under "flags". The installed dependencies are copypasted from the install manual: apt-get install build-essential libssl-dev libcurl4-openssl-dev libjansson-dev libgmp-dev automake I have tried to install both with ./build.sh and ./autogen.sh CFLAGS="-O3 -march=native -Wall" CXXFLAGS="$CFLAGS -std=gnu++11"
./configure --with-curl
make the former gives the error posted above while the latter gives the same error when I run the "make" comand. I'd gladly try an older version if anyone can instruct me how to do so, though I'd prefer to understand what am I doing wrong with this. Many thanks in advance! EDIT: apparently, someone encountered a similar error here: https://bitcointalk.org/index.php?topic=1317918.msg14697284#msg14697284 to which it was suggested to him to run gcc -march=native -Q --help=target|grep march. The result of that command is "amdfam10" so I tried to run ./configure by replacing "-march=native" both with "-march=amdfam10" and "-march=btver1" (copied from the linked message). Unfortunately, this changed nothing. EDIT2: Tried the legacy version, 3.5.9.1. Very similar error about "__m128i" being incompatible with "int". Also, something about SSSE3 not being enabled. Your CPU May fall into that AMD black hole where SIMD support was inconsistent. I recently optimized bytswap to use shuffle which is available in SSSE3. Your CPU's specs have SSSE3 but apparently not support for _mm_shuffle_epi8. Have you tried compiling with -march=core2? I'll try to come up with a fix for next release but it's hard to get motivated for old CPUs, especially old AMD CPUs. |
|
|
|
hey braves! It is possible insert a command in .bat, that run miner.exe, that run threads -t variable? Example: 30 min w/ 4 threads, them 60 min w/ 2 threads,... Thank you! Great hashes  cpuminer --help is your friend. --time-limit |
|
|
|
BTW with previous versions there were no such problems and build was successful.
This won't be easy, especially if I have to ask for the same info repeatedly. If it worked before why did you reinstall all the packages? Why are you compiling with sudo? I need to see the FIRST errors and what CPU you have. Try building the previous version again the exact same way. Try compiling from the source tarball instead of cloning. |
|
|
|
After cloning the rep and runnig ./build.sh in ubuntu I am getting errors make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/shvchk/cpuminer-opt'
Makefile:818: recipe for target 'all' failed
make: *** [all] Error 2
strip: 'cpuminer': No such file
what am I doing wrong ? I assume you're a first timer. It always helps to mention that up front. Have you read the instructions and ensure all dependencies are installed? Find the first few errors, read them, try to understand what it means. If you can't figure it out post it here. It also helps if you mention your CPU. |
|
|
|
|
It's not the "weird screen" that will worry employees, it's the mysterious process using so much CPU.
You're trying to run a program often identified as malware and trying to do it surrepticiously.
No, that won't raise any suspicion.
If you're really the owner you way overspent on your PCs if you can run a CPU miner without affecting
the primary task. Not the move of a smart business owner.
You will be found out by the employees, your deception will be exposed, and your integrity as a boss/owner damaged.
Or you're lying and just an employee trying to scam your boss.
Either way you're a lyer, but I say do it, and get caught. People like you need to be exposed.
|
|
|
|
Small question: is it possible to autodetect the cpu features on startup and just select the "best" - most appropriate - version?
As the miner itself detects the features already this should be possible in theory but maybe not that easy to implement with the "loading" of the correct binary/code
Not worth the effort to think about it. Any run time checking would increase bloat (a growing issue (pun intended)) and slow things down. Optimizing means doing as much as possible at compile time to reduce the decisions made at run time. It's the classic generalization vs specialization dilema. |
|
|
|
|
V3.7.10 will likely be the last release with 4way as a distinct feature. It will now be included
in AVX2 when avaiable.
As a result I will no longer compile seperate 4way Windows binaries.
In addition I will no longer produce sse42 or avx-sha builds. Nehalem CPUs will use the sse2 build
while Ryzen will use avx2-sha.
I will also refine (shorten) the file names. Any other listed feature will include SSE2. AVX2 will include AVX,
AES and SSE2.
All features will be listed at program startup.
|
|
|
|
the link is downloading 3.6.10 Silly mistake, fixed. |
|
|
|
I keep getting the following error:
./cpuminer -a x11 --benchmark
********** cpuminer-opt 3.7.9 ***********
A CPU miner with multi algo support and optimized for CPUs
with AES_NI and AVX2 and SHA extensions.
BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
CPU: Intel(R) Core(TM) i3-2100 CPU @ 3.10GHz.
SW built on Jan 16 2018 with GCC 7.2.0.
CPU features: SSE2 AVX.
SW features: SSE2 AVX.
Algo features: SSE2 AES AVX AVX2 4WAY.
A CPU with AES and AVX2 is required!
Thank you in advance for the effort and work dedicated to this software.
I had incorrectly assumed a CPU with AVX also has AES. The i3-2100 doesn't have AES. I'll fix that in the next release (very soon). In the meantime you can use an older version or compile with only SSE42 Thanks for reporting. |
|
|
|
|
I getting fed up with people who don't know how to use their system then come here whining without
providing any useful information. I will no longer ask. If the initial post has insufficient information I
will simply ignore it.
Dual CPU systems and their operating systems are much more complicated that regular desktops.
Learn how to use them properly before complaining here.
|
|
|
|
|
Show Posts
