• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Prime95 gives errors SP2004 doesn't.

P

Pez D Spencer

Banned
I've currently got my Q6600 at 3.0GHz using a 333MHz bus and a 9x multi. Voltage is at 1.35. Even at 1.4v, Prime95 spits out errors within 5 minutes. SP2004 ran for almost 2 hours without a single error. And even at 2 hours SP2004 was still going strong and error free. I only stopped it because 2 hours is stable enough for me. If I have problems down the road I'll regroup, but right now things seem fine.

With both Prime95 and SP2004 I used small FFT's of 8k. I used Prime95 version 25.5 which tests all four cores simultaneously. With SP2004 I opened four instances of the program and set the "CPU" drop down menu in each instance to it's own core. Someone mentioned on another website about opening four instances of SP2004 and using task manager to set the affinity of each instance to it's own core through the task manager, but from what I could tell, the "CPU" menu in SP2004 did this so I didn't bother with task manager. Speed fan shows all four cores as loaded up when running SP2004 using only the in program CPU menu, so I assume this is just as good as using task manager.

Both Prime95 and SP2004 are doing the same testing aren't they? I don't quite understand Marsenne (sp?) prime numbers but I'm pretty sure they do the same calculations with the CPU. I just don't understand why Prime95 crashes so easily and SP2004 doesn't. The only explanation I can think of is that Prime95 v25.x has bugs in it. On the Prime95 main site, they only have a v24 for download. I got the v25.5 I used from somewhere else. Maybe the v25 I was using isn't fully tested and is still considered a Beta. I don't think the site I got it from mentioned beta though. Who knows.

So what's the deal? Anyone know?
 
CPUs are very complicated little things. Very small differences in the input can cause very large differences in internal state - even though both programs are doing about the same thing from a high-level perspective, the exact instruction sequence is different, which can easily cause one of them to sensitize a critical path while the other one doesn't. That's one of the things that makes silicon validation so difficult - you can't exhaustively test a CPU, so you have to come up with test that exercise as much of it as possible in a reasonable amount of time.
 
And the same Orthos test may fail tomorrow if you are very close to the upper limit of your overclock. Remember that Orthos does not guarantee stability. I run Orthos for about 1 to 2 hours. The real stability test comes when I run my applications.
 
did u try memtest? or unlinking ur memory in ur bios and running it at stock? if ur memory is a bit oc'ed that might be why prime95 is spitting out those errors.
 
oh and how are your cpu temps? if it is overheating maybe that might be why it is erroring in prime95.
 
Temps hit the low 60's during a torture test. Memory is unlinked as well. To hell with it. The PC has been running stable as a rock all day so I'm calling it good anyway.
 
Sounds like the new Prime95 might be better than before at ferreting out errors? Interesting find. I was worried that it might be the opposite.

I still wonder about your methodology with SP2004, I think that you have to set the affinity with Task Manager for it to properly test all cores.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top