Sign in with
Sign up | Sign in
Your question
Solved

Prime95, how many Threads to run?

Last response: in Overclocking
Share
January 19, 2010 12:18:43 AM

I was doing some OC stuff and testing with P95 with 4 threads of blended torture test for ~15 minutes.

What do you guys do or recommend for an adaquate test in terms of number or threads to run and type of test and duration?

More about : prime95 threads run

Best solution

a c 125 K Overclocking
January 19, 2010 11:30:42 AM

Depends on your CPU, If you have a quadcore run 4 threads, If you have a Ci7 with hyperthreading run 8 threads. 15 mins is no where near enough time, The minimum should be 8 hours, and I'd recommend 12 hours or if you can leave it going for 24 hours that would be best.

Small FFT's will test your CPU the most and will produce the highest temps. Blend will test memory and CPU.
Share
January 21, 2010 2:23:05 PM

I read this somewhere and have been applying it during my OC activities.

1 - Extreme overlocking - Just run on the primary thread of each cores
2 - If pt1 proves ok after 12 hours, go with the secondary cores
3 - I would suggest going straight to Blended tests as thats where your RAM and CPU are put into play which is basically what happens during your day to day use of your PC.

I'm using an i7 though so aint sure if the same applies for Phenoms. cheers.
m
0
l
Related resources
a c 197 K Overclocking
January 21, 2010 3:08:49 PM

I disagree with #2. I use small fft's for 24 hours. I used to think that 12 hours was enough. But a couple of times, I was testing overnight and discovered next morning that the system I was testing failed at 14 and 18 hours. That system finally passed a 72 hour small fft test run.
m
0
l
January 21, 2010 7:14:19 PM

jsc said:
I disagree with #2. I use small fft's for 24 hours. I used to think that 12 hours was enough. But a couple of times, I was testing overnight and discovered next morning that the system I was testing failed at 14 and 18 hours. That system finally passed a 72 hour small fft test run.



I somewhat agree but must admit not totally. 24 hours is a long time to be honest though I'm guilty of having done around a week of that running :) . I don't see a problem as 12 hours being a baseline at this time to run through the initial test phase. Should it fail later the box probably requires some further tweaking which would normally be a minor change...

my 2 cents.
m
0
l
January 21, 2010 7:16:39 PM

jsc said:
I disagree with #2. I use small fft's for 24 hours. I used to think that 12 hours was enough. But a couple of times, I was testing overnight and discovered next morning that the system I was testing failed at 14 and 18 hours. That system finally passed a 72 hour small fft test run.



I somewhat agree but must admit not totally. 24 hours is a long time to be honest though I'm guilty of having done around a week of that running :) . I don't see a problem as 12 hours being a baseline at this time to run through the initial test phase. Should it fail later the box probably requires some further tweaking which would normally be a minor change...

my 2 cents.
m
0
l
January 27, 2010 11:35:36 PM

Best answer selected by darkfa8.
m
0
l
!