PDA

View Full Version : prime95 stress testing - how relevant?


os10
10-22-07, 03:23 PM
question: how important is it for an OC'd cpu to pass prime95 stress test?

i have a q6600 at 9x400, 1.5vcore on air it runs everything just fine except for prime95, which causes temps to soar up to 79c and reboot after 5-20 minutes. iv'e tried various tweaks here and there, it doesn't seem like i'm going to do any better with my current configuration. but i'm still interested in running at this speed because i have no problem with all the usual benchmark apps, SMP-enabled games and video encoding, which don't cause temps to go much over 60c, at most.

what else will stress the CPU as much as prime95? what other applications would i need to be careful of, that might cause as much strain on the CPU..?

obviously, if it were fully stable under prime95 for hours at a time, that would be preferable... but if it runs everything i throw it on a daily basis is there really anything to worry about..?

advice appreciated. TIA.

Evilsizer
10-22-07, 03:48 PM
it shows overall system stability... how well the HS can handle the heat the cpu puts out when oced. wiether the cpu temp is within range when oced.

orthos is another program or intel tat's has a built in stresser as well.

deeppow
10-23-07, 07:29 AM
As Evilsizer said, it is an indicator of system stability. In particular, it is an indicator of calculational stability --- is the machine doing math correctly.

So if you're doing calculations were the answer is important to you then I would say that being able to run Prime95 (or ORTHOS) without failures is a must. If you don't need to calculate math then you can be more tolerant of Prime failures, e.g. listening to music, playing games, word processing. :beer:

aja
10-23-07, 10:57 AM
So if you're doing calculations were the answer is important to you then I would say that being able to run Prime95 (or ORTHOS) without failures is a must. If you don't need to calculate math then you can be more tolerant of Prime failures, e.g. listening to music, playing games, word processing. :beer:

I tend to disagree. That is like saying that a couple of memory errors in memtest never hurt anyone! And then you get all these posts about people that keep getting BSOD for no reason at all, or get corrupted system32 files.

But I get what you are saying - there are varying degrees of stability. If it takes 12 hours prime to crash, chances are the end user will not be able to make it crash while gaming and such...

Surfrider77
10-23-07, 03:57 PM
I personally would rather back an OC down for Orthos (or Prime) stability. BSODs drive me batpoo insane and I have crashed in gaming before as well. Its just not worth it to me. If you say youre stable when using the machine, its up to you. If you start seeing it flake out or crashing apps randomly, its a good sign you need to re-evaluate your overclock.

As for the OCF community, we generally agree a "stable overclock" is Prime or Orthos stable for at least 12 hours... some say 24.

deeppow
10-24-07, 05:50 AM
I tend to disagree. That is like saying that a couple of memory errors in memtest never hurt anyone! And then you get all these posts about people that keep getting BSOD for no reason at all, or get corrupted system32 files.

But I get what you are saying - there are varying degrees of stability. If it takes 12 hours prime to crash, chances are the end user will not be able to make it crash while gaming and such...

Points well made and better stated than I did. :beer:

My criteria is no BSODs. I fully agree with you Surfrider77. Some of this comes down to if the apps properly trap errors so the OS doesn't crash.