mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > GPU Computing

Reply
 
Thread Tools
Old 2013-01-30, 05:22   #661
flashjh
 
flashjh's Avatar
 
"Jerry"
Nov 2011
Vancouver, WA

1,123 Posts
Default

Quote:
Originally Posted by Koyaanisqatsi View Post
They do 5 ms/it on the current workload for CUDALucas.
I'm getting ~5 ms/it with a 580 now, and that's without tweaking the FFT. That doesn't seem very exciting . I hope the nVidia Titans are not the same as a 580 for CUDA work. Especially since they are going to cost $899 US.

Do you have a K20X or access to one?
flashjh is offline   Reply With Quote
Old 2013-01-30, 07:54   #662
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

72·197 Posts
Default

Quote:
Originally Posted by Koyaanisqatsi View Post
They do 5 ms/it on the current workload for CUDALucas.
Quote:
Originally Posted by flashjh View Post
I'm getting ~5 ms/it with a 580 now, and that's without tweaking the FFT. That doesn't seem very exciting . I hope the nVidia Titans are not the same as a 580 for CUDA work. Especially since they are going to cost $899 US.

Do you have a K20X or access to one?
Wrong tread, this is mfaktO. I suspect some mod moved them here, as the discussion is not connected, too.
LaurV is offline   Reply With Quote
Old 2013-01-30, 23:35   #663
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

27AE16 Posts
Default

Quote:
Originally Posted by kracker View Post
I don't think it will let you even install the drivers without a AMD card... I think
YMMV, but on my Asus AMD board, if I run the chipset driver install with the ATI iGPU disabled in BIOS, the ATI graphics driver is not an available option.

EDIT: All my graphics and crunching run on nVidia. I have, however, been able to run graphics on the ATI iGPU in combination with crunching on a GTX 460. I doubt that the HD49xx (don't remember exactly) iGPU has the cojones to run mfakto. The problem for using it for display, is that I then loose the discreet GPU processing power for Photoshop/Bridge. Hence, I don't use it at all.

Last fiddled with by kladner on 2013-01-30 at 23:45
kladner is offline   Reply With Quote
Old 2013-01-30, 23:49   #664
Bdot
 
Bdot's Avatar
 
Nov 2010
Germany

3·199 Posts
Default

Quote:
Originally Posted by sdbardwick View Post
Results after installing AMD Catalyst (it automatically installed the AMD APP SDK).
Any truncated output is due to crash.
Brilliant, thanks a lot for these tests!

  • The first test (running on the CPU) shows that even without the AMD SDK, mfakto can compile the kernels, but they don't work with the Intel compiler (no factor found).
  • After installing the AMD SDK (as part of Catalyst), you need to force mfakto to use the Intel platform (mfakto w/o -d switch did not find anything)
  • Now the CPU device is missing, instead the HD4000 is available as the only usable device, but it supports less features (no surprise here). The -35 error is CL_INVALID_QUEUE_PROPERTIES - I can remove my request for out-of-order processing on the device, that should get around this error.
  • Compilation of my OpenCL sources does not automatically propagate constants to vectors - I can change that in the code. And it does not know about printf - easy as well.
I think I can provide a version of mfakto soon that solves these issues, I just don't have time right now ... and then we may face the not-working kernels problem again, this time on the GPU. And this one will be harder to tackle.
Bdot is offline   Reply With Quote
Old 2013-01-31, 00:15   #665
Rodrigo
 
Rodrigo's Avatar
 
Jun 2010
Pennsylvania

2·467 Posts
Default

Bdot,

Thank you very much for continuing to follow up on this frustrating issue!

Rodrigo
Rodrigo is offline   Reply With Quote
Old 2013-01-31, 00:21   #666
Bdot
 
Bdot's Avatar
 
Nov 2010
Germany

3·199 Posts
Default

Quote:
Originally Posted by kladner View Post
YMMV, but on my Asus AMD board, if I run the chipset driver install with the ATI iGPU disabled in BIOS, the ATI graphics driver is not an available option.

EDIT: All my graphics and crunching run on nVidia. I have, however, been able to run graphics on the ATI iGPU in combination with crunching on a GTX 460. I doubt that the HD49xx (don't remember exactly) iGPU has the cojones to run mfakto. The problem for using it for display, is that I then loose the discreet GPU processing power for Photoshop/Bridge. Hence, I don't use it at all.
The iGPU of the AMD APUs is capable of running mfakto and delivers ~30GHz-days/day (HD 6550D / A8 3850). You should not try the older iGPUs. Though anything HD4xxx and higher will work, you may end up with 3 or 5 GHz-days/day.
Bdot is offline   Reply With Quote
Old 2013-01-31, 01:12   #667
Bdot
 
Bdot's Avatar
 
Nov 2010
Germany

59710 Posts
Default

Quote:
Originally Posted by Bdot View Post
I just don't have time right now ...
OK, you made me curious, and I can sleep later

I made the changes and created a special hd4000 package. Would you give it a try? I hope I caught all the compile errors.

The mfakto.hd4000.exe binary is the ordinary one, if that succeeds compiling the kernels with the -d 11 switch, then you can try 'mfakto.hd4000-pi.exe - d 11 -st' in order to get detailed performance numbers for the different kernels.
Bdot is offline   Reply With Quote
Old 2013-01-31, 01:43   #668
Rodrigo
 
Rodrigo's Avatar
 
Jun 2010
Pennsylvania

2·467 Posts
Default

Hi Bdot,

I take it that that hd4000 package is intended specifically for sdbardwick to try on his machine, is that right?

Rodrigo
Rodrigo is offline   Reply With Quote
Old 2013-01-31, 02:04   #669
Bdot
 
Bdot's Avatar
 
Nov 2010
Germany

3·199 Posts
Default

Quote:
Originally Posted by Rodrigo View Post
Hi Bdot,

I take it that that hd4000 package is intended specifically for sdbardwick to try on his machine, is that right?

Rodrigo
Yes. And for others who see the HD4000 in the clinfo output and have the ADM SDK as well as the Intel SDK installed.
Bdot is offline   Reply With Quote
Old 2013-01-31, 02:34   #670
sdbardwick
 
sdbardwick's Avatar
 
Aug 2002
North San Diego County

5×137 Posts
Default

Progress...I think.
Hangs on selftest
Code:
C:\hd>mfakto.hd4000 -d 11
mfakto 0.12-Win-HD4000 (64bit build)


Runtime options
  Inifile                   mfakto.ini
WARNING: Cannot read SievePrimesMin from inifile, using default value (5000)
  SievePrimesMin            5000
WARNING: Cannot read SievePrimesMax from inifile, using default value (1000000)
  SievePrimesMax            1000000
WARNING: Cannot read SievePrimes from inifile, using default value (25000)
  SievePrimes               25000
WARNING: Cannot read SievePrimesAdjust from inifile, using default value (0)
  SievePrimesAdjust         0
WARNING: Cannot read NumStreams from inifile, using default value (3)
  NumStreams                3
WARNING: Cannot read GridSize from inifile, using default value (3)
  GridSize                  3
WARNING: Cannot read WorkFile from inifile, using default (worktodo.txt)
  WorkFile                  worktodo.txt
WARNING: Cannot read ResultsFile from inifile, using default (results.txt)
  ResultsFile               results.txt
WARNING: Cannot read Checkpoints from inifile, enabled by default
  Checkpoints               enabled
WARNING: Cannot read CheckpointDelay from inifile, set to 300s by default
  CheckpointDelay           300s
WARNING: Cannot read Stages from inifile, enabled by default
  Stages                    enabled
WARNING: Cannot read StopAfterFactor from inifile, set to 1 by default
  StopAfterFactor           bitlevel
WARNING: Cannot read PrintMode from inifile, set to 0 by default
  PrintMode                 full
  V5UserID                  none
  ComputerID                none
WARNING: Cannot read AllowSleep from inifile, set to 0 by default
  AllowSleep                no
  TimeStampInResults        no
WARNING: Cannot read VectorSize from inifile, set to 4 by default
  VectorSize                4
WARNING: Cannot read GPUType from inifile, using default (AUTO)
  GPUType                   AUTO
WARNING: Cannot read SieveOnGPU from inifile, set to 0 by default
  SieveOnGPU                no
WARNING: Cannot read SmallExp from inifile, set to 0 by default
  SmallExp                  no
WARNING: Cannot read SieveCPUMask from inifile, set to 0 by default
  SieveCPUMask              0
Compiletime options
  SIEVE_SIZE_LIMIT          36kiB
  SIEVE_SIZE                289731bits
  SIEVE_SPLIT               250
  MORE_CLASSES              enabled
Select device - Get device info - Compiling kernels ..........
WARNING: Unknown GPU name, assuming VLIW5 type. Please post the device name "Int
el(R) HD Graphics 4000 (Intel(R) Corporation)" to http://www.mersenneforum.org/s
howthread.php?t=15646 to have it added to mfakto. Set GPUType in mfakto.ini to s
elect a GPU type yourself and avoid this warning.

OpenCL device info
  name                      Intel(R) HD Graphics 4000 (Intel(R) Corporation)
  device (driver) version   OpenCL 1.1  (9.17.10.2932)
  maximum threads per block 512
  maximum threads per grid  134217728
  number of multiprocessors 16 (1280 compute elements)
  clock rate                350MHz

Automatic parameters
  threads per grid          1048576
  optimizing kernels for    VLIW5

running a simple selftest ...
ERROR: selftest failed for M53015323 (mfakto_cl_barrett92)
  no factor found
########## testcase 2/19 (#2598) ##########
Code:
C:\hd>mfakto.hd4000-pi -d 11 -st
mfakto 0.12-Win-HD4000 (64bit build)


Runtime options
  Inifile                   mfakto.ini
WARNING: Cannot read SievePrimesMin from inifile, using default value (5000)
  SievePrimesMin            5000
WARNING: Cannot read SievePrimesMax from inifile, using default value (1000000)
  SievePrimesMax            1000000
WARNING: Cannot read SievePrimes from inifile, using default value (25000)
  SievePrimes               25000
WARNING: Cannot read SievePrimesAdjust from inifile, using default value (0)
  SievePrimesAdjust         0
WARNING: Cannot read NumStreams from inifile, using default value (3)
  NumStreams                3
WARNING: Cannot read GridSize from inifile, using default value (3)
  GridSize                  3
WARNING: Cannot read WorkFile from inifile, using default (worktodo.txt)
  WorkFile                  worktodo.txt
WARNING: Cannot read ResultsFile from inifile, using default (results.txt)
  ResultsFile               results.txt
WARNING: Cannot read Checkpoints from inifile, enabled by default
  Checkpoints               enabled
WARNING: Cannot read CheckpointDelay from inifile, set to 300s by default
  CheckpointDelay           300s
WARNING: Cannot read Stages from inifile, enabled by default
  Stages                    enabled
WARNING: Cannot read StopAfterFactor from inifile, set to 1 by default
  StopAfterFactor           bitlevel
WARNING: Cannot read PrintMode from inifile, set to 0 by default
  PrintMode                 full
  V5UserID                  none
  ComputerID                none
WARNING: Cannot read AllowSleep from inifile, set to 0 by default
  AllowSleep                no
  TimeStampInResults        no
WARNING: Cannot read VectorSize from inifile, set to 4 by default
  VectorSize                4
WARNING: Cannot read GPUType from inifile, using default (AUTO)
  GPUType                   AUTO
WARNING: Cannot read SieveOnGPU from inifile, set to 0 by default
  SieveOnGPU                no
WARNING: Cannot read SmallExp from inifile, set to 0 by default
  SmallExp                  no
WARNING: Cannot read SieveCPUMask from inifile, set to 0 by default
  SieveCPUMask              0
Compiletime options
  SIEVE_SIZE_LIMIT          36kiB
  SIEVE_SIZE                289731bits
  SIEVE_SPLIT               250
  MORE_CLASSES              enabled
  CL_PERFORMANCE_INFO       enabled (DEBUG option)
Select device - Get device info - Compiling kernels ..........
WARNING: Unknown GPU name, assuming VLIW5 type. Please post the device name "Int
el(R) HD Graphics 4000 (Intel(R) Corporation)" to http://www.mersenneforum.org/s
howthread.php?t=15646 to have it added to mfakto. Set GPUType in mfakto.ini to s
elect a GPU type yourself and avoid this warning.

OpenCL device info
  name                      Intel(R) HD Graphics 4000 (Intel(R) Corporation)
  device (driver) version   OpenCL 1.1  (9.17.10.2932)
  maximum threads per block 512
  maximum threads per grid  134217728
  number of multiprocessors 16 (1280 compute elements)
  clock rate                350MHz

Automatic parameters
  threads per grid          1048576
  optimizing kernels for    VLIW5

########## testcase 1/1559 ##########
Starting trial factoring M50804297 from 2^67 to 2^68 (0.59GHz-days)
  k_min = 1599999998520 - k_max = 1900000000000
Using GPU kernel "barrett15_75"
  done |    ETA |     GHz |time/class|    #FCs | avg. rate | SieveP. |CPU idle
1048576 FCs copied in 0.37 ms (11287.15 MB/s), proc'd in 167.74 ms (6.25 M/s)
UPDATED CLINFO (After Catalyst install)

Edit: Additional information for above -st run (let the program run after "hang")
Code:
########## testcase 1/1559 ##########
Starting trial factoring M50804297 from 2^67 to 2^68 (0.59GHz-days)
  k_min = 1599999998520 - k_max = 1900000000000
Using GPU kernel "barrett15_75"
  done |    ETA |     GHz |time/class|    #FCs | avg. rate | SieveP. |CPU idle
1048576 FCs copied in 0.37 ms (11287.15 MB/s), proc'd in 167.74 ms (6.25 M/s)
Error -5: Copying h_ktab(clEnqueueWriteBuffer)
ERROR from tf_class.
Error exit as selftest failed
Attached Files
File Type: txt updtCL.txt (13.9 KB, 135 views)

Last fiddled with by sdbardwick on 2013-01-31 at 02:50
sdbardwick is offline   Reply With Quote
Old 2013-01-31, 02:38   #671
kracker
 
kracker's Avatar
 
"Mr. Meeseeks"
Jan 2012
California, USA

23·271 Posts
Default

Quote:
Originally Posted by Bdot View Post
The iGPU of the AMD APUs is capable of running mfakto and delivers ~30GHz-days/day (HD 6550D / A8 3850).
A bit different here.
Running two instances on integrated (6550D) to max it.
Attached Thumbnails
Click image for larger version

Name:	6550D.jpg
Views:	158
Size:	157.0 KB
ID:	9215  
kracker is offline   Reply With Quote
Reply



Similar Threads
Thread Thread Starter Forum Replies Last Post
gpuOwL: an OpenCL program for Mersenne primality testing preda GpuOwl 2718 2021-07-06 18:30
mfaktc: a CUDA program for Mersenne prefactoring TheJudger GPU Computing 3497 2021-06-05 12:27
LL with OpenCL msft GPU Computing 433 2019-06-23 21:11
OpenCL for FPGAs TObject GPU Computing 2 2013-10-12 21:09
Program to TF Mersenne numbers with more than 1 sextillion digits? Stargate38 Factoring 24 2011-11-03 00:34

All times are UTC. The time now is 03:15.


Mon Aug 2 03:15:56 UTC 2021 up 9 days, 21:44, 0 users, load averages: 1.11, 1.28, 1.35

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.