need help with tricky OC

[OS]

I have a mobile 2600/NF2 combo. P95 detected a fault after 24 hrs. Temp was ~45C. I don't think this is heat related and voltage was at 1.7V. If I jack up the voltage another .05V, will this give absolute stability or is it more the cpu is fundamentally unstable?

This situation also shakes my confidence in having an OCed system since if it took 24 hrs to detect an error, who's to say that if I increase the voltage, it might take 1 week to get an error. 1 error is always 1 too many.

 

[Pohemi]

It depends on how concerned you are about Prime95. You could possibly run another test(or game application) for weeks on end and not have any failures. In theory, Prime95 will always fail, whether after one hour, one day, or one month. Personally, I would probably be satisfied with 24 hours on Prime95, but only if the system passed every other test and benchmark without ever failing. Prime95 seems to be the system killer even when there's apparently nothing wrong. I guess it might be personal opinion that would decide this one.

BTW, what type of NF2 board are you using?

 

[OS]

Originally posted by: Pohemi420
It depends on how concerned you are about Prime95. You could possibly run another test(or game application) for weeks on end and not have any failures. In theory, Prime95 will always fail, whether after one hour, one day, or one month. Personally, I would probably be satisfied with 24 hours on Prime95, but only if the system passed every other test and benchmark without ever failing. Prime95 seems to be the system killer even when there's apparently nothing wrong. I guess it might be personal opinion that would decide this one.

BTW, what type of NF2 board are you using?

8rda3i

I think it's safe to say in this case the failure is probably overclock related is what I'm getting at.

It's too bad the multipliers on this mobo seem broken, otherwise I'd just run stock 2.0 ghz. 😛

 

[Toro 45]

Originally posted by: Pohemi420
Personally, I would probably be satisfied with 24 hours on Prime95, but only if the system passed every other test and benchmark without ever failing.

I agree 24 hours and plays my games and runs other bechies to my satisfaction

 

[Uhtrinity]

If everything runs fine I wouldn't be concerned. On my main gaming system I had two sticks of single side (2 X 256) DDR 500 ram, prime 95 was rock solid. With 2 sticks of double sided memory (2 X 512) prime 95 fals within 109 minutes, but I have yet to have had any game crashes (long sessions of WOW mainly). A failure at 24 hours shouldn't be a big deal unless you are doing some high precision work.

 

[NewBlackDak]

My ultimate stability test now is bootstrapping Gentoo. If you have any hardware problems it'll show you very quickly.

 

[OS]

updated, I'm running 2.5 ghz at 1.85V, P95 small test stable for 3 days lol. Should be ok for now, running 53C, but I wonder how stable it's gonna be when summer rolls around.

 

[VirtualLarry]

Originally posted by: Pohemi420
It depends on how concerned you are about Prime95. You could possibly run another test(or game application) for weeks on end and not have any failures. In theory, Prime95 will always fail, whether after one hour, one day, or one month.

That's a bit illogical - if that were true, then Prime95 would be basically useless as a distributed-computing project, because then it would be essentially guaranteeing that "bad data" would be returned to the project, at some eventual point in time. I disagree with that. If you system is running stably, you shouldn't have errors in Prime95 - ever. If that implies using ECC RAM to avoid cosmic rays from flipping bits in your DRAM array, then so be it, perhaps your long-term usage patterns of your PC dictate that such types of memory are required for your application.

Originally posted by: Pohemi420
Personally, I would probably be satisfied with 24 hours on Prime95, but only if the system passed every other test and benchmark without ever failing. Prime95 seems to be the system killer even when there's apparently nothing wrong. I guess it might be personal opinion that would decide this one.

That's what makes it such a good testing tool - it's so stressful to most systems, that if Prime95 works, there is a high probability that everything else will work as well. (*)

(*) Not necessarily true for P4-based systems, as Prime95 stresses the FPU primarily, which is the weak point on Athlons, but not P4s.