|Keep in mind I was running on 8 cores and now I am getting more PPD on 7 cores.|
|Suppose your hardware has M cores (or M virtual cores if you have HypwerThreading). To process SMP, FahCore_a3 requires the use of "-smp N" where N can be equal to or less than the number of cores M, that your OS reports that you have. (The default value of N is the maximum number, M.)|
There have been several reports stating that setting N less than M might perform better than N=M.
Originally Posted by Grandpa_01
Seems we need a little clarification here -smp 8 will net you quite a bit more PPD when running a dedicated folding machine that is running only the smp client. If you are going to be running a GPU or use the computer allot and this includes browsing the Internet then -smp 7 will net you more PPD with the smp client. The smp client does not do very well when you use your computer for other tasks than folding and this goes for bigadv and regular smp folding.
Originally Posted by $ilent
my pc is on 24/7, I leave it folding all nigth and during day while im at work, soon as im home i leave it folding during internet stuff, but if i game for few hrs i close the gpu and cpu client. So its folding 100% 8 cores for least 20 hours a day.
edit: even with smp 7, cpu usage still showing 100% on all 8 core.
Originally Posted by Alazar
Not sure what you are trying to clarify, Grandpa_01. The forums discuss this as overall performance difference, even for dedicated rigs. The very nature of the SMP client itself is why a 100% cpu usage isn't the optimal configuration.
I am trying to find the article that Punchy talked about getting significantly different results based upon how many cores he was folding with in the 24-core machines he was playing with. I recall something odd like -smp 19 was vastly superior in performance than -smp 24. Can't find the discussion though.