Důrazně se doporučuje používat nejnovější vývojové verze BOINC (nestabilní

Zatím doporučují pouze velmi zkušeným uživatelům.
Kód: Vybrat vše
<app_info>
<app>
<name>poemcl</name>
<user_friendly_name>POEM++ OpenCL</user_friendly_name>
</app>
<file_info>
<name>poemcl_0.1_windows_intelx86__opencl_ati_100</name>
<executable/>
</file_info>
<app_version>
<app_name>poemcl</app_name>
<version_num>1</version_num>
<plan_class>opencl_ati_100</plan_class>
<avg_ncpus>0.5</avg_ncpus>
<max_ncpus>0.5</max_ncpus>
<flops>xxxxxx</flops>
<coproc>
<type>ATI</type>
<count>0.25</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>poemcl_0.1_windows_intelx86__opencl_ati_100</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
Information to run more then one gpu app per gpu in windows 7 (other windows may work)
Make a file called app_info.xml in the program \ProgramData\BOINC\projects\boinc.fzk.de_poem
What decides how many tasks to run PER GPU is the "<count>0.2</count>"
take 1, divided by the number of tasks you want to run PER GPU.
A "count" of 1 will run 1 task / gpu
A "count" of 0.5 will run 2 tasks / gpu
A "count" of 0.33 will run 3 tasks / gpu
A "count" of 0.25 will run 4 tasks / gpu
A "count" of 0.2 will run 5 tasks / gpu
A "count" of 0.125 will run 8 tasks / gpu
A "count" of 0.1 will run 10 tasks / gpu
If you have 2 gpu's, and a "count" of .33, you will be running 6 tasks at once, 3 on each gpu.
Set "no new tasks" in Boinc manager an complete and return ALL WORK before making, or deleteing the app_info file. You can edit it as much as you want, but creating or deleting the file will purge all work you have on the system, and the server will think you still have it until it times out. After ANY change to the app_info file you have to stop and restart BOINC, A reboot will do this also.
use the parts between the **************** but not the ****************.
****************
<app_info>
<app>
<name>poemcl</name>
<user_friendly_name>POEM++ OpenCL</user_friendly_name>
</app>
<file_info>
<name>poemcl_0.1_windows_intelx86__opencl_ati_100</name>
<executable/>
</file_info>
<app_version>
<app_name>poemcl</app_name>
<version_num>1</version_num>
<plan_class>opencl_ati_100</plan_class>
<avg_ncpus>0.75</avg_ncpus>
<max_ncpus>1.0</max_ncpus>
<flops>25000000000</flops>
<coproc>
<type>ATI</type>
<count>0.2</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>poemcl_0.1_windows_intelx86__opencl_ati_100</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
****************
I have 2 ATI 6990 cards, each one has 2 gpu's for a total of 4 gpu's. If I have both gpu's on a single card crunching poem, I can't get above ~65% on each gpu. To limit poem to using 1 gpu on each card I exclude 1 gpu on each card, poem uses whatever is left after excluding. I limit milkyway and collatz to the other 2 gpu's (that poem is not using). This way I get 85~90% on each gpu, and 2 tasks on other projects at the same time.
make a cc_config.xml and put it ProgramData\BOINC
****************
<cc_config>
<options>
<exclude_gpu><url>http://milkyway.cs.rpi.edu/milkyway/</u ... xclude_gpu>
<exclude_gpu><url>http://boinc.thesonntags.com/collatz/</ ... xclude_gpu>
<exclude_gpu><url>http://milkyway.cs.rpi.edu/milkyway/</u ... xclude_gpu>
<exclude_gpu><url>http://boinc.thesonntags.com/collatz/</ ... xclude_gpu>
<exclude_gpu><url>http://boinc.fzk.de/poem/</url><device_ ... xclude_gpu>
<exclude_gpu><url>http://boinc.fzk.de/poem/</url><device_ ... xclude_gpu>
</options>
</cc_config>
****************
Jedu jen na té jedné 4770 a dle tvých rad. Klient 7.0.7nenym píše:Foreste, máš výpomoc? Na HD4770 takový výkon podle mého není dlouhodobě možný, snad prvních pár tasků, než se CreditNew zorientuje.