Hi guys, I need some help about setting the affinity of the folding processes. Firstly, let me explain my rig - it's a C2D E6750 (which will be upgraded to a C2Q in the near future) with two GTX260's in it (and a 3rd will be added as well). What I WAS doing was had an SMP (2) client running and the two GPU clients. Obviously each GPU client needs about half a core, so really, all three of the processes were competing with each other to get CPU time. So I've tried to improve the situation slightly. And I'm hoping this will nudge the PPD upwards a bit My idea is to run a single core dedicated to the GPU (done, and succesful using a system variable to force the two cleints onto core 0). Side note, from what I can see looking at task manager, it appears that an nvidia GPU really only needs about 30% of a core... Anyways, I digress, so with one core dedicated to feeding the GPU's, I want the other core to run a thread (uniprocessor). Again, setting this up in FAHControl is no problem. My only concern is how do I make sure that the client always assigns the uniprocessor cleint to run on core 1? I've gone into task manager and set the affinity to 1, but when I restart, that setting will be lost won't it? Does this make sense what I am trying to do? The idea is also that once I've got the quad in there, I need to set it so that the SMP thread will run on cores 1-3 with core 0 dedicated to feeding the GPU's. Thanks in advance guys!
Thanks for that - it works a treat. I would have thought that there would have been an affinity flag in there somewhere but alas, problem is solved.
