mfaktc: a CUDA program for Mersenne prefactoring

[KaptainBlaZzed]

Originally Posted by airsquirrels

You must compile for the specific compute version and CUDA version of the card you are using. In this case the 8.0 RC and compute 6.1. Each generation of GPUs requires a separate build

can you please tell me how to do this for windows x64.

[airsquirrels]

Originally Posted by KaptainBlaZzed

can you please tell me how to do this for windows x64.

Unfortunately I do not have a windows environment setup so I don't have build instructions, however I also misspoke - I was thinking of CUDALucas not mfaktc. The same details apply, however with CUDA 8 RC 1 there exists a bug that prevents mfaktc from working properly. We believe it will be fixed in the final CUDA 8 release but we do not know when that will be.

[henryzz]

Originally Posted by airsquirrels

Some of the older cards/CUDA versions supported multiple compute versions and architectures, but Maxwell and Pascal both seem to required specific builds.

Maybe it is a difference with windows. I am pretty certain I have run toolkit v4 stuff on my Maxwell. I have certainly used the same binaries on a 8600GTS and 750Ti

[kladner]

Originally Posted by KaptainBlaZzed

can you please tell me how to do this for windows x64.

I have not seen him post in a little while, but user 'flashjh' aka Jerry is a master of Windows compiles for different versions.

[Gordon]

Originally Posted by kladner

I have not seen him post in a little while, but user 'flashjh' aka Jerry is a master of Windows compiles for different versions.

Until NVIDIA fix their software...

[kladner]

Originally Posted by Gordon

Until NVIDIA fix their software...

True dat.

[tServo]

Originally Posted by airsquirrels

We believe it will be fixed in the final CUDA 8 release but we do not know when that will be.

Latest rumor is end of September, and as we all know, EVERYTHING you read on the internet is true, eh?

[tServo]

Originally Posted by tServo

Latest rumor is end of September, and as we all know, EVERYTHING you read on the internet is true, eh?

Cuda 8 general availability announced today, Sept 28.

[TheJudger]

mfaktc v0.21 (64bit built)
[...]
CUDA version info
  binary compiled for CUDA  8.0
  CUDA runtime version      8.0
  CUDA driver version       8.0

CUDA device info
  name                      GeForce GTX 1080
  compute capability        6.1
[...]

Selftest statistics
  number of tests           26192
  successfull tests         26192

  kernel             | success |   fail
  -------------------+---------+-------
  UNKNOWN kernel     |      0  |      0
  71bit_mul24        |   2586  |      0
  75bit_mul32        |   2682  |      0
  95bit_mul32        |   2867  |      0
  barrett76_mul32    |   1096  |      0
  barrett77_mul32    |   1114  |      0
  barrett79_mul32    |   1153  |      0
  barrett87_mul32    |   1066  |      0
  barrett88_mul32    |   1069  |      0
  barrett92_mul32    |   1084  |      0
  75bit_mul32_gs     |   2420  |      0
  95bit_mul32_gs     |   2597  |      0
  barrett76_mul32_gs |   1079  |      0
  barrett77_mul32_gs |   1096  |      0
  barrett79_mul32_gs |   1130  |      0
  barrett87_mul32_gs |   1044  |      0
  barrett88_mul32_gs |   1047  |      0
  barrett92_mul32_gs |   1062  |      0

selftest PASSED!

Reference (aka Founders Edition) 1080:

Starting trial factoring M66362159 from 2^73 to 2^74 (28.83 GHz-days)
 k_min =  71160531149400
 k_max =  142321062305090
Using GPU kernel "barrett76_mul32_gs"
Date    Time | class   Pct |   time     ETA | GHz-d/day    Sieve     Wait
Sep 29 13:31 |    0   0.1% |  2.345  37m29s |   1106.37    82485    n.a.%
Sep 29 13:31 |    4   0.2% |  2.345  37m27s |   1106.37    82485    n.a.%
Sep 29 13:31 |    9   0.3% |  2.340  37m19s |   1108.73    82485    n.a.%

I guess we've a new performance and performance per watt leader.

Oliver

[tServo]

Originally Posted by TheJudger

[
I guess we've a new performance and performance per watt leader.

Oliver

Also, looks like they fixed their bugs that have been there since CUDA 7.0 & 7.5, yes?

[GP2]

Supposedly CUDA 8 has Visual Studio 2015 support, has anyone tried compiling mfaktc under Windows yet?

Originally Posted by tServo

Also, looks like they fixed their bugs that have been there since CUDA 7.0 & 7.5, yes?

Yes:

  number of tests           26192
  successfull tests         26192