So I have a new system that I wanted to test out the Intel HD side of things in addition to the nVidia. I figured out I had to enable test applications to get Intel GPU work. I was checking on the results and was wondering why I could find the Intel GPU work, so I had to go to the host and discovered that the app name is:
Binary Radio Pulsar Search (Arecibo) v1.34 () windows_x86_64
There's no reference to it being a GPU client anywhere in there though. I know they are specifically for the Intel GPU because they coincide with the new Intel GPU WUs I got. Right now the app name is the same as the ARM client so it gets a little confusing seeing how things are checking in...
Would that app name be modifiable so it's clear it's a GPU app and not a CPU app?
Thanks,
Yavanius
Copyright © 2024 Einstein@Home. All rights reserved.
Hm. This is all a bit
)
Hm. This is all a bit complicated, which may be why there is no official reply to your question, yet. It is the boinc-client that tells the server about the resources you have, not the name of the binary. And then the project server knows what applications to send to you and what work units for those applications. If your Intel GPU is not seen then it is not unlikely that it was auto-disabled in the BIOS because of the monitor being attached to the NVidia card. There is a lot about it all out there in many forums and there are pseudo-monitors that you can plug to your VGA socket - which would not work with my machine, though give it a try.
Whatever you are doing, the name of the application is not important. All applications are regular binaries that optionally use different extra resources for their acceleration. There are project-supplied app_info.xml files that map binaries to the resources these need and may wait for when the resource (like the GPU) is busy. But, well, I propose to not to mess with those app_info.xml files. That you need to do only when providing self-compiled anonymous binaries.