I use two GPUs for several BOINC tasks, but have only been able to receive MOO! tasks that require two GPUs. Is there a way to get single GPU MOO! tasks when BOINC is set to use two or more GPUs?
Thanks for any suggestions.

I use two GPUs for several BOINC tasks, but have only been able to receive MOO! tasks that require two GPUs. Is there a way to get single GPU MOO! tasks when BOINC is set to use two or more GPUs?
Thanks for any suggestions.

Are you saying that only one gpu is being used here or that both gpu's are being used but on only one task?

If only one gpu is being used you need a cc_config.xml file like this:

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

You are using Windows so open NOTEPAD and copy and paste the above lines into it, then save the file as a txt type file, but with the file name of "cc_config.xml" without the quotes of course. Save it in the main Boinc directory of c:\program data/boinc. Do NOT use a word processing program to create the file, it will put some extra stuff in there and Boinc won't know how to handle it and it won't work, ONLY use NOTEPAD!!!

If you are saying that both gpu's are crunching the same unit, then I have no clue other than to exclude one of the gpu's from Moo your cc_config.xml file would look like this:

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
<exclude_gpu>
<url>http://moowrap.net/</url>
<device_num>1</device_num>
</exclude_gpu>
</options>
</cc_config>

That would exclude one of the gpu's from running Moo gpu units. Lots of people do this if they want to crunch 2 different projects gpu units, one project on each gpu. For instance to exclude Einstein from one of your gpu's, so you could run one gpu here at Moo and one gpu at Einstein your cc_config.xml file would look like this:

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
<exclude_gpu>
<url>http://moowrap.net/</url>
<device_num>1</device_num>
</exclude_gpu>
<exclude_gpu>
<url>http://einstein.phys.uwm.edu/</url>
<device_num>0</device_num>
</exclude_gpu>
</options>
</cc_config>

That tells Boinc to use all of your gpu's and then excludes gpu 0 from running at Einstein and gpu 1 from running Moo units, since both of your gpu's are the same it doesn't make any difference which gpu you exclude where.

Hi Mikey and thanks for the post.

I changed my cc_config file to the following to include your suggestion:

<?xml version="1.0" encoding="UTF-8" ?>
<cc_config>
<log_flags />
<options>
<fetch_on_update>1</fetch_on_update>
<use_all_gpus>1</use_all_gpus>
<client_download_url>http://boinc.berkeley.edu/download.php</client_download_url>
<client_version_check_url>http://boinc.berkeley.edu/download.php?xml=1</client_version_check_url>
<network_test_url>http://www.google.com/</network_test_url>
<http_transfer_timeout>300</http_transfer_timeout>
<http_transfer_timeout_bps>10</http_transfer_timeout_bps>
<max_event_log_lines>2000</max_event_log_lines>
<max_file_xfers>8</max_file_xfers>
<max_file_xfers_per_project>8</max_file_xfers_per_project>
<ncpus>14</ncpus>
<rec_half_life_days>10</rec_half_life_days>
<save_stats_days>30</save_stats_days>
<exclude_gpu>
<url>http://einstein.phys.uwm.edu</url>
<device_num>1</device_num>
<type>NVIDIA</type>
<app>einsteinbinary_BRP4G</app>
</exclude_gpu>
<exclude_gpu>
<url>http://moowrap.net/</url>
<device_num>1</device_num>
</exclude_gpu>
<exclude_gpu>
<url>http://einstein.phys.uwm.edu</url>
<device_num>1</device_num>
<type>NVIDIA</type>
<app>einsteinbinary_BRP6</app>
</exclude_gpu>
</options>
</cc_config>

I then Read the config files, suspended all GPU projects, resumed MOO, and immediately got 3 tasks. All 3 said they require 2 Nvidia GPUs, and none of them started computing even though both GPUs were free.

Removing the MOO GPU exclusion and Reading the config files immediately started one MOO task on 2 GPUs.

Also, previously I had created an app_config for MOO that limited MOO to one concurrent project, but that didn't change the 2 GPUs per task:

<app_config>
<app>
<name>dnetc</name>
<max_concurrent>1</max_concurrent>
</app>
</app_config>


So it seems that MOO tasks always use 2 GPUs, which I can manage manually if necessary.

Art

I did notice a problem here:

<exclude_gpu>
<url>http://einstein.phys.uwm.edu</url>
<device_num>1</device_num>
<type>NVIDIA</type>
<app>einsteinbinary_BRP4G</app>
</exclude_gpu>
<exclude_gpu>
<url>http://moowrap.net/</url>
<device_num>1</device_num>
</exclude_gpu>
<exclude_gpu>
<url>http://einstein.phys.uwm.edu</url>
<device_num>1</device_num>
<type>NVIDIA</type>
<app>einsteinbinary_BRP6</app>
</exclude_gpu>

If you notice all the gpu's you are excluding are:
<device_num>1</device_num>

That can NOT be as you have both an AMD and an Nvidia gpu listed. Also I don't think you need the AMD or Nvidia line if you are using the 'device_num' line. BUT if you ARE using the 'device_num" line it needs to be correct, Boinc always labels the 1st gpu it sees as device zero and the 2nd gpu device one, the 3rd gpu device 2 etc. I think you could take out the 'device_num' line as long as you continue to use the AMD and Nvidia labels, but ONLY if you only have one of each.

Go into the 'Event Log' in the Boinc Manager and every time Boinc starts up it lists, by type, AMD, Nvidia or even Intel gpu, every gpu it sees, it ALSO lists the device number on the same line. For instance on my laptop it says:

4/3/2016 5:51:35 AM | | OpenCL: Intel GPU 0: Intel(R) HD Graphics 4000 (driver version 10.18.10.4276, device version OpenCL 1.2, 1298MB, 1298MB available, 141 GFLOPS peak)

You can see it says "GPU 0", and yes that's a zero. I did have to stop and restart Boinc as the log is only so long and mine didn't show that as it's been running for over 3 weeks.