PSST, come over to Xs and PM me your email and I'll send the file to you.

If this shouldn't be here, let me know and I can pull it.

This is the code that I found elsewhere that would apply.

<app_info>
<app>
<name>hcc1</name>
<user_friendly_name>Help Conquer Cancer</user_friendly_name>
</app>
<file_info>
<name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</name>
<executable/>
</file_info>
<file_info>
<name>hcckernel.cl.6.56</name>
<executable/>
</file_info>
<app_version>
<app_name>hcc1</app_name>
<version_num>656</version_num>
<platform>windows_intelx86</platform>
<plan_class>nvidia_hcc1</plan_class>
<avg_ncpus>1</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>hcckernel.cl.6.56</file_name>
<open_name>hcckernel.cl</open_name>
</file_ref>
</app_version>
</app_info>


What you would want to look at is the part:

<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>

If you change it from 1 to 0.5, you will get 2 units per GPU. Make it 0.33, and you get 3 units per GPU. 0.25 would get 4 units per GPU.

I would only suggest doing this if you are seeing consistent usage below 90%. Setting it to 2 units per GPU will likely be more than enough to get any card up to 99-100% usage. Also remember, the more GPU units you run, to more CPU cores will be used, taking away from your normal CPU crunching.

You need to change the application names to the appropriate ATI ones. I'll check and see if I can find something.


EDIT: Found it. BTW, these came from the "lag issue" thread. Kudos to nanoprobe. (I just removed the
<cmdline>--gpu-target-frequency 30 --gpu-polling-mode 8 --process-priority 1</cmdline> part)

<app_info>
<app>
<name>hcc1</name>
<user_friendly_name>Help Conquer Cancer</user_friendly_name>
</app>
<file_info>
<name>wcg_hcc1_img_6.56_windows_intelx86__ati_hcc1</name>
<executable/>
</file_info>
<file_info>
<name>hcckernel.cl.6.56</name>
<executable/>
</file_info>
<app_version>
<app_name>hcc1</app_name>
<version_num>656</version_num>
<platform>windows_intelx86</platform>
<plan_class>ati_hcc1</plan_class>
<avg_ncpus>1.0</avg_ncpus>
<max_ncpus>1.0</max_ncpus>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>wcg_hcc1_img_6.56_windows_intelx86__ati_hcc1</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>hcckernel.cl.6.56</file_name>
<open_name>hcckernel.cl</open_name>
</file_ref>
</app_version>
</app_info>

IT seems that the GPU taking a CPU core would affect any and all other projects running on the CPU. You would still be able to run other projects on whatever cores, if any, that are left.


-------
EDIT:

I just did a quick test. I ran 8 units on my 2 590s (4 GPUs). While the usage of the GPU went from an average of 90% to 96%, the CPU usage at all times during crunching these units went from 4 full cores to 8 full cores.

The app_info.xml just sets some parameters for the specified application to use, it does not tell BOINC to only do the work listed in the file.


I also edited my post above while you were posting.