Folding Forum

7im wrote:Keep SMP slot at default priority of idle. Change GPU slot priority to "low"... slot option = "core-priority" with setting = "low"

I just noticed there's a problem with that; it has probably existed since the introduction of FahCoreWrapper.exe. The wrapper priority gets adjusted but the actual core remains at default priority. Maybe that could be fixed with ticket #800? If so, maybe that ticket deserves a priority upgrade, since the "core-priority" setting actually appears to be broken.

In the meantime, there are 3rd party utilities like Process Lasso which allow you to adjust process priorities and affinities automatically. I like Lasso because you can increase the polling interval to 10s and shut down the GUI, only leaving the engine (ProcessGovernor.exe) running in the background. The engine itself requires very little CPU time once you've configured everything to your liking.

Napoleon wrote:

7im wrote:Keep SMP slot at default priority of idle. Change GPU slot priority to "low"... slot option = "core-priority" with setting = "low"

I just noticed there's a problem with that; it has probably existed since the introduction of FahCoreWrapper.exe. The wrapper priority gets adjusted but the actual core remains at default priority. Maybe that could be fixed with ticket #800? If so, maybe that ticket deserves a priority upgrade, since the "core-priority" setting actually appears to be broken.

I wouldn't want to try to co-opt the #800 enhancement for something that's really a bug either in the wrapper or in the FahCore(s) itself. (That would also take care of differing priories.) I think we need some more detailed research to see which. I've always wondered: Does a "low" setting actually change the priority of FahCore_1x when running in V6? Taskmgr doesn't provide enough information.

Then, too, the priority settings for a core running on NVidia and running on ATI would necessarily be set quite differently -- and GPUs on add-on cards would probably be quite different than integrated GPUs on the mb.

Sorry if this seems like a bump, but I think the edit I added to my earlier "benchmark" post is rather significant: viewtopic.php?f=80&t=22445&p=224555#p224555.