astropulse GPU


log in

Advanced search

Message boards : Number crunching : astropulse GPU

Previous · 1 · 2 · 3
Author Message
Profile Vipin Palazhi
Avatar
Send message
Joined: 29 Feb 08
Posts: 247
Credit: 103,099,458
RAC: 36,151
India
Message 1333158 - Posted: 31 Jan 2013, 13:37:16 UTC - in response to Message 1332920.


Not really, My experience of running AP on my 9800GTX+ is that it's just not worth it,

Claggy


Thanks for the info Claggy.

JarrettH
Send message
Joined: 14 Nov 02
Posts: 68
Credit: 12,552,179
RAC: 2,592
Canada
Message 1333234 - Posted: 31 Jan 2013, 18:20:49 UTC

How do you install the Astropulse GPU app? Simple instructions would be enormously helpful. Thanks :D
____________

Profile Mike
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 23410
Credit: 31,885,474
RAC: 23,462
Germany
Message 1333241 - Posted: 31 Jan 2013, 18:44:34 UTC

Just download the new AP package from there.

http://setiathome.berkeley.edu/forum_thread.php?id=70723

Shut down Boinc.
Copy all files to you project directory.
Open **NV_r1761.aistub and copy appinfo section and paste it into your appinfo.xml.
Restart boinc.

____________

JarrettH
Send message
Joined: 14 Nov 02
Posts: 68
Credit: 12,552,179
RAC: 2,592
Canada
Message 1333249 - Posted: 31 Jan 2013, 19:26:14 UTC
Last modified: 31 Jan 2013, 19:47:08 UTC

Thank you!

And how do you figure out the number of compute units your GPU has?
Will it just start running my Astropulse units on the GPU from now on? I had Astropulse work prior. Do I have to remove the old Astropulse V6 CPU references?

High end cards (more than 12 compute units)
-unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -hp
Mid range cards (less than 12 compute units)
-unroll 10 -ffa_block 6144 -ffa_block_fetch 1536 -hp
entry level GPU (less than 6 compute units)
-unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 -hp

Would my 550 Ti be more entry or mid-range?

<app>
<name>astropulse_v6</name>
</app>
<file_info>
<name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</name>
<executable/>
</file_info>
<file_info>
<name>libfftw3f-3.dll</name>
<executable/>
</file_info>
<file_info>
<name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</name>
</file_info>
<app_version>
<app_name>astropulse_v6</app_name>
<version_num>604</version_num>
<platform>windows_intelx86</platform>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.2</max_ncpus>
<plan_class>cuda_opencl_100</plan_class>
<cmdline></cmdline>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
<open_name>ap_cmdline.txt</open_name>
</file_ref>
</app_version>
<app_version>
<app_name>astropulse_v6</app_name>
<version_num>604</version_num>
<platform>windows_intelx86</platform>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.2</max_ncpus>
<plan_class>opencl_nvidia_100</plan_class>
<cmdline>-unroll 10 -ffa_block 6144 -ffa_block_fetch 1536 -hp</cmdline>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
<open_name>ap_cmdline.txt</open_name>
</file_ref>
</app_version>

____________

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4048
Credit: 32,693,609
RAC: 413
United Kingdom
Message 1333262 - Posted: 31 Jan 2013, 19:44:18 UTC - in response to Message 1333249.
Last modified: 31 Jan 2013, 19:46:19 UTC

Thank you!

And how do you figure out the number of compute units your GPU has?
Will it just start running my Astropulse units on the GPU from now on? I had Astropulse work prior. Do I have to remove the old Astropulse V6 CPU references?

High end cards (more than 12 compute units)
-unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -hp
Mid range cards (less than 12 compute units)
-unroll 10 -ffa_block 6144 -ffa_block_fetch 1536 -hp
entry level GPU (less than 6 compute units)
-unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 -hp

I have a 550 Ti


Read the stderr.txt of a completed task, it'll say how many compute units your GPU has.

Rather than putting cmdline parameters in the app_info, put them on the ap_cmdline_win_x86_SSE2_OpenCL_NV.txt file instead, means you can change them, any they will take on the next app start, rather than the next Boinc restart,
at the moment you put the cmdline parameters in only the 2nd <app_version> section, and will so end end being unused:

<app>
<name>astropulse_v6</name>
</app>
<file_info>
<name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</name>
<executable/>
</file_info>
<file_info>
<name>libfftw3f-3.dll</name>
<executable/>
</file_info>
<file_info>
<name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</name>
</file_info>
<app_version>
<app_name>astropulse_v6</app_name>
<version_num>604</version_num>
<platform>windows_intelx86</platform>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.2</max_ncpus>
<plan_class>cuda_opencl_100</plan_class>
<cmdline></cmdline>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
<open_name>ap_cmdline.txt</open_name>
</file_ref>
</app_version>
<app_version>
<app_name>astropulse_v6</app_name>
<version_num>604</version_num>
<platform>windows_intelx86</platform>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.2</max_ncpus>
<plan_class>opencl_nvidia_100</plan_class>
<cmdline>-unroll 10 -ffa_block 6144 -ffa_block_fetch 1536 -hp</cmdline>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>AP6_win_x86_SSE2_OpenCL_NV_r1761.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
<open_name>ap_cmdline.txt</open_name>
</file_ref>
</app_version>


Claggy

JarrettH
Send message
Joined: 14 Nov 02
Posts: 68
Credit: 12,552,179
RAC: 2,592
Canada
Message 1333268 - Posted: 31 Jan 2013, 19:52:36 UTC
Last modified: 31 Jan 2013, 19:55:04 UTC

Ohhh I see. Do I need to add the cmdline tags around it when placed in ap_cmdline_win_x86_SSE2_OpenCL_NV?

I'll check for that file, but here's a workunit: http://setiathome.berkeley.edu/result.php?resultid=2815964854

multiProcs 4 ?
____________

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4048
Credit: 32,693,609
RAC: 413
United Kingdom
Message 1333276 - Posted: 31 Jan 2013, 20:09:45 UTC - in response to Message 1333268.
Last modified: 31 Jan 2013, 20:10:12 UTC

Ohhh I see. Do I need to add the cmdline tags around it when placed in ap_cmdline_win_x86_SSE2_OpenCL_NV?

I'll check for that file, but here's a workunit: http://setiathome.berkeley.edu/result.php?resultid=2815964854

multiProcs 4 ?

No you don't need any tags, just -unroll 10 -ffa_block 6144 -ffa_block_fetch 1536 -hp will do.

Double check once you complete an OpenCL task, it should be 4 too.

Claggy

Previous · 1 · 2 · 3

Message boards : Number crunching : astropulse GPU

Copyright © 2014 University of California