I want to be able to run two GPU WUs on the same GPU but want them to be from different projects, e.g. GPUgrid and Milkyway. The parameter might look like this for the GPUgrid project app_config:

<app d=0>
<name>acemdlong</name>
<gpu_versions>
<cpu_usage>1.0</cpu_usage>
<gpu_usage>0.5</gpu_usage>
</gpu_versions>
</app>
<app d=1>
<name>acemdlong</name>
<gpu_versions>
<cpu_usage>1.0</cpu_usage>
<gpu_usage>0.5</gpu_usage>
</gpu_versions>
</app>

Then also have similar in the app_config for Milkyway:

<app d=0>
<name>milkyway</name>
<gpu_versions>
<cpu_usage>1.0</cpu_usage>
<gpu_usage>0.5</gpu_usage>
</gpu_versions>
</app>
<app d=1>
<name>milkyway</name>
<gpu_versions>
<cpu_usage>1.0</cpu_usage>
<gpu_usage>0.5</gpu_usage>
</gpu_versions>
</app>

It's interesting that

<exclude_gpu>
   <device_num>N</device_num>
   <app>appname</app>
</exclude_gpu>

is available.

---

that already works by default but it can be messy (inefficieint)

with MW set to 0.19 and Einstein set to 0.33 I ended up after while with one MV taking 0nly 1/5 of gpu and one Einstein taking only 1/3 of the same one and wasted gpu time.

Normally 3 Einstein on one gpu or 5 MW but if both enabled then things get haphazard quickly.

that already works by default but it can be messy (inefficieint)

with MW set to 0.19 and Einstein set to 0.33 I ended up after while with one MV taking 0nly 1/5 of gpu and one Einstein taking only 1/3 of the same one and wasted gpu time.

Normally 3 Einstein on one gpu or 5 MW but if both enabled then things get haphazard quickly.

That doesn't control which task runs on which GPU, which is the desire.