Hi Razankar, welcome to the forum and +1 for your first post.
I'm late to the party because I've been visiting my mum every day in hospital for the last 2 weeks and haven't been on the forum.
Your card should handle 4 at a time which will be more efficient than only running 2 at once. So if you wish to try it later, copy and paste the below into Notepad:
<app_info>
<app>
<name>poemcl</name>
<user_friendly_name>POEM++ OpenCL</user_friendly_name>
</app>
<file_info>
<name>poemcl_1.3_windows_intelx86__opencl_ati_100</name>
<executable/>
</file_info>
<app_version>
<app_name>poemcl</app_name>
<version_num>103</version_num>
<plan_class>opencl_ati_100</plan_class>
<avg_ncpus>0.5</avg_ncpus>
<max_ncpus>2</max_ncpus>
<flops>2.1e10</flops>
<coproc>
<type>ATI</type>
<count>0.25</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>poemcl_1.3_windows_intelx86__opencl_ati_100</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
Save as app_info.xml. Make sure to click drop down box Save As type: and select All files (*) so that .txt is not appended to file name. Place this app_info.xml file in C:\ProgramData\BOINC\projects\boinc.fzk.de_poem folder.
Restart BOINC, you should see something like "Poem@Home | 30/06/2012 6:55:18 PM | Found app_info.xml; using anonymous platform" in BOINC Manager Event Log.
Download some fresh WUs and go to town. Good luck and good crunching.
This app_info.xml uses 2 CPU cores to support 4 GPU tasks.
Because your AMD Phenom CPU has 6 physical cores this should work OK. Once you have it working satisfactorily you may wish to experiment with allocating 3 CPU cores to support GPU processing and/or running 5 tasks at a time. The aim is to reduce average task time, which is achieved by maximising GPU load percentage and thus increasing efficiency. What you finally decide on depends on the performance of your hardware and your own preference.
The use of this app_info.xml will prevent your NVIDIA card downloading and processing POEM GPU tasks. It is better not to use it and the associated CPU resources for POEM GPU when you have an ATI card available in that box which is more efficient for POEM. You could perhaps use it for another less CPU intensive GPU project if you wished.
Unlike some other GPU projects, POEM OpenCL GPU uses a lot of CPU resources so the performance/efficiency directly relates to the quantity of CPU resources allocated. If you allocate no CPU resources the POEM GPU application may be extremely sluggish and in some cases may freeze. If you allocate all CPU resources, POEM GPU performance will be maximised. The compromise you select will depend on the level of support you wish to give to CPU projects. Because of the reliance on CPU and the large amount of data transferred, performance is also affected by CPU speed and system memory speed/latency.
POEM GPU application on ATI/AMD cards does this research up to 100-130 times faster than a CPU core, that's oodles of research work processed so it is worthwhile to give it a whirl now and again when electricity bill allows. As a side benefit the credit rate when processing multiple concurrent tasks, although not as high as some code cracking GPU applications, is quite generous for a biology type project.
I am currently on one of my POEM "binges", running 5 tasks at once on my HD 5870. This processes lots of POEM protein research every day, which grants about 500K credits. It requires 5 of my 8 hyperthreaded CPU cores though, so only 3 CPU cores are doing CPU projects. I could do a bit more POEM work by allocating all 8 cores to support POEM GPU but I still like to support some old and new CPU projects, even if it's only with 3 cores.
Author
Topic: My cc_file (Read 15758 times)