• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

CPU speed and CPU load for GPU folding

I

ICE_9

2[H]4U
Joined
Feb 28, 2005
Messages
3,459
I briefly searched the forum for some info and I need more specifics.

Does your single GPU client take a single full CPU core?
If not, what operating system and setup are you running?
If a quad core, do you run SMP on top?

Seperate note:
Does you see a significant change in PPD if you lower your CPU speed?

Thanks!
 
I"ve been doing a bunch of research on this. Thanks to the info provided to me by Xil and Tiger, it turns out that, under XP, there's an idle loop encoded into the CPU portion of the GPU client while the GPU waits for the CPU's input. Supposedly this loop is just waiting for information, and not really the execution of instructions.

It appears that one can have as many as 4-5 different GPU clients off of a single CPU, under XP, with no apparent PPD loss.

And yes, you can use a VM to load all 4 cores, and have a GPU client in windows on top of that. One of the cores will be reduced to around 80-90% effectiveness though.
 
I'm running 4 GPU cores off a single E2180, so 2 per core is fine, even if it is low end proc.

 
I'll be running 4 cores off a single core AM2 Processor... I'll get back to you OP :p.
 
I was running a x2 3800 with one GPU core. No way I could run the SMP client anymore since the GPU client took a big hit. I tried running a single client on the free core, but I can't figure out how to lock affinity for the CPU client to the same core. After it loads a new unit, both cores go at it and my GPU client takes a big dump in production.
 
I run gpu + smp on a e6300@3.15 in xp with no loss of points from what I can tell. There may be some, but the point difference between the smp client and the normal client more than makes up for it,
 
ok. So, in my little test I have found that their appears to be no difference in frame times when I run my E6600 and the single GPU core for my 8800GS. I have a program that will auto engage the Speedstep on the processor and drop the multiplier down to 1.6 GHz. This appears to net me a drop of 14W from my system. It isn't much, but it is something.

Program:
http://www.diefer.de/speedswitchxp/
 
You must log in or register to reply here.
Back
Top