OpenCL Applications
log in

Advanced search

Message boards : News : OpenCL Applications

Author Message
Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 14314 - Posted: 26 Jun 2012, 18:56:45 UTC

OpenCL versions of the application have been released for 32-bit and 64-bit Windows and for 64-bit Linux.

The validator has also been modified to catch invalid results from AMD HD 7xxx GPUs running the ATI Brook+ applications. Those GPUs should automatically be sent the new OpenCL applications which will validate correctly.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 14319 - Posted: 26 Jun 2012, 20:57:52 UTC

I also fixed a bugh where, when a result returned had 0 max steps and 0 total steps, there is no longer a divide by zero error which would crash the validator.

Old man
Send message
Joined: 24 Aug 09
Posts: 782
Credit: 14,832,359
RAC: 0
Message 14325 - Posted: 27 Jun 2012, 14:56:48 UTC

Is result ok?

Stderr_txt:

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
Collatz Conjecture v3.11 x86_64 for OpenCL
Based on the AMD Brook+ kernels by Gipsel
Parameters I4194304 K256 S1 --device 0
Start 2375910580627515877736
Checking 824633720832 numbers
Device Vendor Advanced Micro Devices, Inc.
Name Tahiti
Board Name AMD Radeon HD 7900 Series
Compute Units 28
Driver Version CAL 1.4.1703 (VM)
OpenCL Version OpenCL 1.1 AMD-APP (898.1)
Workgroup Size Auto
Kernel amd_bitalign
Numbers/Kernel 4194304
Kernels/Reduction 256
Numbers/Reduction 1073741824
Reductions/WU 768
Highest Steps 1829 for 2375910581031892161447
Total Steps 408819347226677
GPU time 777.884 seconds
CPU time 0.514803 seconds
Total time 778.948 seconds
17:45:21 (3904): called boinc_finish

</stderr_txt>
]]>

Umhh and no. It is not valdiated yet. It wait for wingman report his/her result.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 14326 - Posted: 27 Jun 2012, 18:29:33 UTC - in response to Message 14325.

Is result ok?

Stderr_txt:

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
Collatz Conjecture v3.11 x86_64 for OpenCL
Based on the AMD Brook+ kernels by Gipsel
Parameters I4194304 K256 S1 --device 0
Start 2375910580627515877736
Checking 824633720832 numbers
Device Vendor Advanced Micro Devices, Inc.
Name Tahiti
Board Name AMD Radeon HD 7900 Series
Compute Units 28
Driver Version CAL 1.4.1703 (VM)
OpenCL Version OpenCL 1.1 AMD-APP (898.1)
Workgroup Size Auto
Kernel amd_bitalign
Numbers/Kernel 4194304
Kernels/Reduction 256
Numbers/Reduction 1073741824
Reductions/WU 768
Highest Steps 1829 for 2375910581031892161447
Total Steps 408819347226677
GPU time 777.884 seconds
CPU time 0.514803 seconds
Total time 778.948 seconds
17:45:21 (3904): called boinc_finish

</stderr_txt>
]]>

Umhh and no. It is not valdiated yet. It wait for wingman report his/her result.


That looks correct and 2375910581031892161447 has 1829 steps.

[AF>France>Est>Bourgogne]Skwi
Send message
Joined: 15 Apr 11
Posts: 5
Credit: 22,069,565
RAC: 0
Message 14332 - Posted: 28 Jun 2012, 7:49:03 UTC
Last modified: 28 Jun 2012, 8:04:22 UTC

Tasks take between 2 or 3 more time to complete for the same number of points since yesturday.
With my GT440 and Quadro600. Those cards use OpenCL.
My HD3850 don't have this problem, and it don't use OpenCL.
Is that due to the update?

[AF>France>Est>Bourgogne]Skwi
Send message
Joined: 15 Apr 11
Posts: 5
Credit: 22,069,565
RAC: 0
Message 14333 - Posted: 28 Jun 2012, 9:09:51 UTC

With the Quadro600, GPU load is around 99%, and I have heavy lags in windows (7 64bit) with all softwares.
Previous app was around 96% GPU load on this card (without using a app_info file),
and not any lag.

Ashley
Send message
Joined: 22 Jun 12
Posts: 11
Credit: 1,028,851
RAC: 0
Message 14335 - Posted: 28 Jun 2012, 9:54:27 UTC

28/06/2012 3:20:03 PM | | No config file found - using defaults
28/06/2012 3:20:03 PM | | Starting BOINC client version 7.0.25 for windows_x86_64
28/06/2012 3:20:03 PM | | log flags: file_xfer, sched_ops, task
28/06/2012 3:20:03 PM | | Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
28/06/2012 3:20:03 PM | | Data directory: C:\ProgramData\BOINC
28/06/2012 3:20:03 PM | | Running under account Ashley
28/06/2012 3:20:03 PM | | Processor: 6 AuthenticAMD AMD Phenom(tm) II X6 1090T Processor [Family 16 Model 10 Stepping 0]
28/06/2012 3:20:03 PM | | Processor: 512.00 KB cache
28/06/2012 3:20:03 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni cx16 syscall nx lm svm sse4a osvw ibs skinit wdt page1gb rdtscp 3dnowext 3dnow
28/06/2012 3:20:03 PM | | OS: Microsoft Windows 7: Home Premium x64 Edition, (06.01.7600.00)
28/06/2012 3:20:03 PM | | Memory: 8.00 GB physical, 16.00 GB virtual
28/06/2012 3:20:03 PM | | Disk: 465.66 GB total, 65.96 GB free
28/06/2012 3:20:03 PM | | Local time is UTC +10 hours
28/06/2012 3:20:03 PM | | NVIDIA GPU 0: GeForce 9800 GT (driver version 301.42, CUDA version 4.20, compute capability 1.1, 1024MB, 966MB available, 638 GFLOPS peak)
28/06/2012 3:20:03 PM | | ATI GPU 0: Barts (CAL version 1.4.1720, 1024MB, 991MB available, 3302 GFLOPS peak)
28/06/2012 3:20:03 PM | | OpenCL: NVIDIA GPU 0: GeForce 9800 GT (driver version 301.42, device version OpenCL 1.0 CUDA, 1024MB, 966MB available)
28/06/2012 3:20:03 PM | | OpenCL: ATI GPU 0: Barts (driver version CAL 1.4.1720 (VM), device version OpenCL 1.2 AMD-APP (923.1), 1024MB, 991MB available)

Ignore the ati card its working fine.

However the 9800gt is part app changes effect it, the old app took 2 hour and 45 minutes the new app take 4 hours and 27 minutes, what ever you did you made it slower.

The new app

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
Collatz Conjecture v3.11 x86_64 for OpenCL
Based on the AMD Brook+ kernels by Gipsel
Parameters --device 0
Start 2375901240619835304296
Checking 824633720832 numbers
Device Vendor NVIDIA Corporation
Name GeForce 9800 GT
Compute Capability 1.1
Compute Units 14
Driver Version 301.42
OpenCL Version OpenCL 1.0 CUDA
Workgroup Size Auto
Kernel bit shifting
Numbers/Kernel 1048576
Kernels/Reduction 64
Numbers/Reduction 67108864
Reductions/WU 12288
Highest Steps 1723 for 2375901240740861851086
Total Steps 409734322247884
GPU time 15549.1 seconds
CPU time 2.32441 seconds
Total time 15551 seconds
11:39:47 (5600): called boinc_finish

</stderr_txt>
]]>

one of the old app.

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
Running Collatz Conjecture (3x+1) CUDA GPU application v2.01
based on version 1.2 by Gipsel
instructed by BOINC client to use device 0
Reading input file ... done.
Checking 824633720832 numbers starting with 2375898455419443194216
No checkpoint data found.
needed 1754 steps for 2375898455979230887529
420352558836524 total executed steps for 824633720832 numbers

WU completed.
GPU time: 10204.6 seconds
Elapsed time: 10327.1
called boinc_finish

</stderr_txt>
]]>

Ashley

Old man
Send message
Joined: 24 Aug 09
Posts: 782
Credit: 14,832,359
RAC: 0
Message 14339 - Posted: 28 Jun 2012, 16:45:04 UTC

Nimi collatz_2375830067204944800104_824633720832_2
Työpaketti 52026930
Luotu 27 Jun 2012 14:29:26 UTC
Lähetetty 27 Jun 2012 14:32:18 UTC
Vastaanotettu 28 Jun 2012 16:19:43 UTC
Tila palvelimella Valmis
Tulos Onnistui
Tila ohjelmassa Valmis
Exit status 0 (0x0)
Tietokoneen tunniste 91338
Raportoinnin takaraja 4 Jul 2012 14:32:18 UTC
Laskenta-aika 797.62
Suoritinaika 0.67
Vahvistuksen tila Vahvistetut
Pisteet 3,043.11
Sovellusversio collatz
Anonyymi alusta (ATIn GPU)
Stderr output

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
Collatz Conjecture v3.11 x86_64 for OpenCL
Based on the AMD Brook+ kernels by Gipsel
Parameters I4194304 K256 S1 --device 0
Start 2375830067204944800104
Checking 824633720832 numbers
Device Vendor Advanced Micro Devices, Inc.
Name Tahiti
Board Name AMD Radeon HD 7900 Series
Compute Units 28
Driver Version CAL 1.4.1703 (VM)
OpenCL Version OpenCL 1.1 AMD-APP (898.1)
Workgroup Size Auto
Kernel amd_bitalign
Numbers/Kernel 4194304
Kernels/Reduction 256
Numbers/Reduction 1073741824
Reductions/WU 768
Highest Steps 1896 for 2375830067885245568225
Total Steps 412671067836075
GPU time 794.919 seconds
CPU time 0.655204 seconds
Total time 795.429 seconds
19:19:31 (3352): called boinc_finish

</stderr_txt>
]]>


Sorry for finnish lanquage in start. First task validated successfully.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 14340 - Posted: 28 Jun 2012, 18:38:16 UTC - in response to Message 14332.

Tasks take between 2 or 3 more time to complete for the same number of points since yesturday.
With my GT440 and Quadro600. Those cards use OpenCL.
My HD3850 don't have this problem, and it don't use OpenCL.
Is that due to the update?


Isn't OpenCL wonderful? It runs slower and has worse video response!

I'm working on a new CUDA version for CUDA 4.2 which I'm hoping will get the speed back up.

[AF>France>Est>Bourgogne]Skwi
Send message
Joined: 15 Apr 11
Posts: 5
Credit: 22,069,565
RAC: 0
Message 14342 - Posted: 28 Jun 2012, 19:39:00 UTC - in response to Message 14340.

Tasks take between 2 or 3 more time to complete for the same number of points since yesturday.
With my GT440 and Quadro600. Those cards use OpenCL.
My HD3850 don't have this problem, and it don't use OpenCL.
Is that due to the update?


Isn't OpenCL wonderful? It runs slower and has worse video response!

I'm working on a new CUDA version for CUDA 4.2 which I'm hoping will get the speed back up.


Yes, wanderful, realy.

Thanks for geting an eye on it.

Ashley
Send message
Joined: 22 Jun 12
Posts: 11
Credit: 1,028,851
RAC: 0
Message 14344 - Posted: 29 Jun 2012, 3:06:38 UTC - in response to Message 14340.

Tasks take between 2 or 3 more time to complete for the same number of points since yesturday.
With my GT440 and Quadro600. Those cards use OpenCL.
My HD3850 don't have this problem, and it don't use OpenCL.
Is that due to the update?


Isn't OpenCL wonderful? It runs slower and has worse video response!

I'm working on a new CUDA version for CUDA 4.2 which I'm hoping will get the speed back up.


Thanks for the update, I was wondering what you where going to, from what i hear opencl will become the standard when it is optimized for the hardware it will be used on.

In the mean time its good that you are going back to Cuda hopefully it will still run on my 9800gt.

Ashley


Post to thread

Message boards : News : OpenCL Applications


Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.