Is there any way to allow BOINC to request GPU/CUDA work from only one project - that is, stop it asking for GPU work from projects I know do not have CUDA WUs?
I am running 6.6.36 on Windows (various flavours).
TIA

---

Where's the problem? When there ar no CUDA applications, you'll get no CUDA tasks from that project and BOINC will reduce the frequency of requests.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

Thank you for your response. I'm not in a position to comment directly on what you say happens. What I notice is that BOINC is requesting GPU tasks from all four projects I have running on the one machine I have a CUDA device, and doing it pretty frequently. I only want to run SETI on my GPU, and I do not want BOINC wasting time asking for CUDA WUs from projects that don't have them to send. And CP messages (below) seem to suggest that my settings are wrong.

20/06/2009 12:36:23	climateprediction.net	Sending scheduler request: To fetch work.
20/06/2009 12:36:23	climateprediction.net	Requesting new tasks for GPU
20/06/2009 12:36:34	climateprediction.net	Scheduler request completed: got 0 new tasks
20/06/2009 12:36:34	climateprediction.net	Message from server: No work sent
20/06/2009 12:36:34	climateprediction.net	Message from server: No work available for the applications you have selected.  Please check your settings on the web site.

---

climate i think does not do cuda at the moment

That is the point. CP does not do CUDA, therefore I would like to be able to tell BOINC to stop asking CP (and other projects that don't do CUDA) for CUDA work units.

---

That is the point. CP does not do CUDA, therefore I would like to be able to tell BOINC to stop asking CP (and other projects that don't do CUDA) for CUDA work units.

Does it really matter? Is it a measurable overhead for either your processing or your comms bandwidth?

On the other hand, many years ago we were told that contacting a project scheduler (and hence accessing the BOINC database) was a costly operation in server resource terms. Now we seem to be happy throwing these redundant RPCs all over the place (even if only twice/host/project/day). I still worry that this may be a contributing factor in the spate of server problems we've seen recently, but we'd need input from a BOINC project server administrator to evaluate that.

Thanks for your input. I have noticed that many of BOINC's calls to servers are not successful (for both GPU and CPU WUs). I have even seen reference sites inaccessible and then I get the (incorrect) message that there is something awry with my internet connection when there clearly is not. Like you, I cannot see how these redundant calls are helping matters and it would seem to be relatively simple to say, "don't make calls for something I don't want you to do anyway". As this does not seem to be an option, maybe it is not so simple.

---

It is currently necessary for the Boinc manager to ask or it woundn't know when Cuda work becomes available at a project. The developers are aware and looking for a better mechanism for this process.