[LOCKED] cpuminer-opt v3.12.3, open source optimized multi-algo CPU miner

[My9bot]

That's weird. The compiler doesn't see the NO_AES_NI macro.

You should be able to comment everything out until I figure this out.

thank you now compiled successfully

[1714141815]

1714141815

[1714141815]

1714141815

[1714141815]

1714141815

[1714141815]

1714141815

[1714141815]

1714141815

[joblo]

That's weird. The compiler doesn't see the NO_AES_NI macro.

You should be able to comment everything out until I figure this out.

thank you now compiled successfully

Was that from the post I deleted? It didn't work for me, I must have messed it up.

[My9bot]

That's weird. The compiler doesn't see the NO_AES_NI macro.

You should be able to comment everything out until I figure this out.

thank you now compiled successfully

Was that from the post I deleted? It didn't work for me, I must have messed it up.

no compiled successfully after i commented everything out

[integrale]

i would like to report

Xubuntu 16.04 LTS  cpuminer-opt 3.5.10  GCC 5.4.1   Amd Athlon x2 240 sse2

Xevan
hmq1725
yescrypt  
timetravel

yet tested  still stack-smashing error

Cryptonight works

[joblo]

So, no real progress on either issue. Thanks for testing.

At least now there is a workaround for the compile error.

Windows binaries will work using Wine to possibly get around the stack smashing.

Other than that I'm kinda stuck on both issues, time for a break.

[joblo]

i would like to report

Xubuntu 16.04 LTS  cpuminer-opt 3.5.10  GCC 5.4.1   Amd Athlon x2 240 sse2

Xevan
hmq1725
yescrypt  
timetravel

yet tested  still stack-smashing error

Cryptonight works

You get stack smashing on all these? I didn't realize that, I thought it was only hmq1725.
Yescrypt has nothing in common with the others.

I thought you had an i5, it's very confusing.

[integrale]

yes you need urgently a break , enjoy it

btw as i was tested another machine   i expected this  running on Ubuntu 14.04 with gcc 4.8.4

         **********  cpuminer-opt 3.5.10  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
     Forked from TPruvot's cpuminer-multi with credits
     to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
     Wolf0, Jeff Garzik and Optiminer.

CPU:         Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
CPU features: SSE2 AES AVX
SW built on Feb 26 2017 with GCC 4.8.4
SW features: SSE2
Algo features: SSE2 AES AVX AVX2
Start mining with SSE2

[2017-02-27 00:04:00] Starting Stratum on stratum+tcp://yiimp.ccminer.org:3555
[2017-02-27 00:04:00] 8 miner threads started, using 'timetravel' algorithm.
[2017-02-27 00:04:00] Stratum difficulty set to 0.125
[2017-02-27 00:04:00] timetravel block 399599, diff 6.775
[2017-02-27 00:04:01] CPU #1: 6219 H, 41.02 kH/s
[2017-02-27 00:04:01] Accepted 1/1 (100%), 6219 H, 41.02 kH/s, 55C
[2017-02-27 00:04:02] CPU #0: 65.54 kH, 59.95 kH/s
[2017-02-27 00:04:02] CPU #2: 65.54 kH, 60.04 kH/s
[2017-02-27 00:04:02] CPU #4: 65.54 kH, 60.06 kH/s
[2017-02-27 00:04:02] CPU #3: 65.54 kH, 59.77 kH/s
[2017-02-27 00:04:02] CPU #7: 65.54 kH, 60.40 kH/s
[2017-02-27 00:04:02] CPU #6: 65.54 kH, 59.07 kH/s
[2017-02-27 00:04:02] CPU #5: 65.54 kH, 58.85 kH/s
[2017-02-27 00:04:05] CPU #2: 207.93 kH, 63.55 kH/s
[2017-02-27 00:04:05] Accepted 2/2 (100%), 607.37 kH, 462.68 kH/s, 62C
[2017-02-27 00:04:10] CPU #7: 514.55 kH, 63.60 kH/s
[2017-02-27 00:04:10] Accepted 3/3 (100%), 1056.38 kH, 465.88 kH/s, 63C
[2017-02-27 00:04:12] CPU #1: 689.11 kH, 63.51 kH/s
[2017-02-27 00:04:12] Accepted 4/4 (100%), 1739.27 kH, 488.37 kH/s, 64C
[2017-02-27 00:04:12] CPU #2: 465.57 kH, 63.49 kH/s
[2017-02-27 00:04:12] Accepted 5/5 (100%), 1996.90 kH, 488.30 kH/s, 64C
[2017-02-27 00:04:13] CPU #2: 17.04 kH, 62.81 kH/s

even compiled  as followed         CFLAGS="-O3 -march=corei7-avx -Wall" CXXFLAGS="$CFLAGS -std=gnu++11" ./configure --with-curl (native wont works)

[joblo]

yes you need urgently a break , enjoy it

btw as i was tested another machine   i expected this  running on Ubuntu 14.04 with gcc 4.8.4

         **********  cpuminer-opt 3.5.10  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
     Forked from TPruvot's cpuminer-multi with credits
     to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
     Wolf0, Jeff Garzik and Optiminer.

CPU:         Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
CPU features: SSE2 AES AVX
SW built on Feb 26 2017 with GCC 4.8.4
SW features: SSE2
Algo features: SSE2 AES AVX AVX2
Start mining with SSE2



even compiled  as followed         CFLAGS="-O3 -march=corei7-avx -Wall" CXXFLAGS="$CFLAGS -std=gnu++11" ./configure --with-curl (native wont works)

Are you sure you compiled with corei7-avx? It says it was compiled without AES and AVX2.

[integrale]

thats the command line i putted in , because compiling with built in function (as native gave me an error)

[integrale]

ac_cs_config="'--with-curl' 'CFLAGS=-O3 -march=corei7-avx -Wall' 'CXXFLAGS=-O3 -march=corei7-avx -Wall -std=gnu++11'"
ac_cs_version="\
cpuminer-opt config.status 3.5.10
configured by ./configure, generated by GNU Autoconf 2.69,
  with options \"$ac_cs_config\"

[joblo]

thats the command line i putted in , because compiling with built in function (as native gave me an error)

Yet another mystery. Can you post the errors you get when you use ./build.sh. The newer compilers
might recongnize the brand name. You could also try -march-sandybridge.

I have no idea why it would not compile for AES and AVX when using -march=corei7-avx.

[integrale]

the mystery looks like this

./autogen.sh  > ./configure > make      ERROR but work with the cmd    CFLAGS="-O3 -march=corei7-avx -Wall" CXXFLAGS="$CFLAGS -std=gnu++11" ./configure --with-curl but it exclude the AVX

./build.sh > ./configure.sh > make  works correctly  with all supported extensions


looks that i choose the wrong way to compile it correctly ?

[joblo]

That's better. You seem to have a variety of CPUs. If you're willing you could help define
the problem better.

Please always try ./build.sh first. Only try other options if it fails.

So I don't get confused we should stick with one CPU at a time. Here is the info I have for
your i7-2600 system, please correct any errors.

CPU: i7-2600
OS: Ubuntu 14.04
GCC: 4.8.4
compile: ./build.sh 

If you have problems please report. If everything works, try one of the xeons. Please provide the info above
for each CPU with the test results.

The old AMD will be last.

 

[joblo]

I've built a VM based on Ubuntu 16.04 with gcc 5.4. It crashes on hmq1725 with statck protector
enabled but works with it disabled. CPU is i7-6700K. AVX2 isn't available, probably a VM issue.

I can take over the troubleshooting now.

Edit: I found the bug in hmq1725. An array was defined as size 20 when it should have been 32 (0x20).
I deny reponsibility for the bug as it was a cut and paste and worked until the compiler got smarter.

until the next release comes out:

Edit /algo/hmq1275.c line 310, change array size to 32.

Code:

old:  uint32_t endiandata[20] __attribute__((aligned(64)));

new:  uint32_t endiandata[32] __attribute__((aligned(64)));

[Epsylon3]

Code:

valgrind --leak-check=full -v --log-file=report.log ./cpuminer ....

that will use secure memory allocations and logs overflows etc. This error is happening if you write too far in an array... and indeed doesnt crash all the time..

told you..

[joblo]

Code:

valgrind --leak-check=full -v --log-file=report.log ./cpuminer ....

that will use secure memory allocations and logs overflows etc. This error is happening if you write too far in an array... and indeed doesnt crash all the time..

told you..

I saw nothing in the valgrind output, What did I miss?

[joblo]

I've also found the problem with compiling hodl with gcc 6.3. There was a macro named min
defined in miner.h. Don't know why compiler version matters.

[Epsylon3]

I saw nothing in the valgrind output, What did I miss?

A crash i guess :p on crash it create a report

[oldDIN]

I've also found the problem with compiling hodl with gcc 6.3. There was a macro named min
defined in miner.h. Don't know why compiler version matters.

Crumbling on "min" and "max"
What does it affect?

Code:

#ifndef min
#define min(a,b) (a>b ? (b) :(a))
#endif
#ifndef max
#define max(a,b) (a<b ? (b) : (a))
#endif

If you rename?

I risked
Replace with

Code:

#ifndef minmin
#define minmin(a,b) (a>b ? (b) :(a))
#endif
#ifndef maxmax
#define maxmax(a,b) (a<b ? (b) : (a))
#endif

On W7-64 GCC 6.3 FX-6300 i3-2200 Compilation successful - aes-avx, sse42.
Before v9 v10 compile collapsed

Code:

./configure CFLAGS="-Ofast -march=corei7-avx -Wall" CXXFLAGS="$CFLAGS -std=gnu++11" --with-crypto --with-curl
./configure CFLAGS="-Ofast -march=corei7 -Wall" CXXFLAGS="$CFLAGS -std=gnu++11" --with-crypto --with-curl

Test AMD deviations from the original are not revealed for 10 algorithms.

[nizzuu]

Tried to use (precompiled from gdrive) windows binary v3.5.10 for x11 algo (Dash):

cpuminer-aes-avx2 -a x11 -t 3 -o stratum+tcp://dash.suprnova.cc:9995 -u aaa -p zzz

This try gives a crash.

The cryptolight and cryptonote algos do work, e.g.:
cpuminer-aes-avx2 -a cryptolight -t 4 -o stratum+tcp://aeon.pool.minergate.com ...
cpuminer-aes-avx2 -a cryptonight -t 3 -o stratum+tcp://xmr.pool.minergate.com ...

Win7 x64 SP1, 16Gb RAM, i5-4570 Haswell

Some details from dump file:

(d10.13d8): Access violation - code c0000005 (first/second chance not available)
ntdll!NtWaitForMultipleObjects+0xa:
00000000`779bc0ea c3              ret

Thanks for the possible solutions.