Prime95 Stable?

What is the worth of being prime95 stable? The rig in my sig is stable in everything including Doom3, Farcry, and COD except Prime95.

I guess it would be useful for people running apps witch require a lot of calculations and need them perfect... for instance, I wouldn't run FAH on any machine that is not prime stable as you may be reporting inaccurate results.

It is also usful if you happen to be a perfectionist, obssessive compulsive, or anal retentive, like me.

It is just annoying that I can't even last 10 minutes in Prime95 but everything seems fine while gaming and benchmarking. Maybe it could have something to do with my powersupply +12v averaging +11.67v. Maybe I should take the time to adjust the pot to raise this line.

I am having the same problem on my setup.. nf7-s and corsair 2X512 3200...
I can memtest error free up to 241 and with 3 errors per pass at 242, but can't prime over 236-237 stable for more than 10 minutes either...
My TT 420W PSU is running a low 12v rail as well, avg between 11.6 and 11.7, sometimes as low as 11.34v, do you really think it is that? I already RMA'd this thing once, and it came back worse than before..not really worth the shipping to do it again..
If so.. how do you up the pots?I'd like to do it.. sick of this thing undervolting..

Isn't the theory supposed be that if the computer can't generate accurate calculations then the computer just isn't running right? As in either the programs you run won't work well as in generate artifacts in the image, etc. or if the program you are using rejects bad calculations and forces the CPU to do it over then it just slows you down?

So what I'm saying is if there are two identical systems, and one prime 95s just fine and the other one only lasts 10 minutes, the one that is stable on P95 should be faster? Now does this hold up in any material differences? No idea or am I just dead wrong?

-Leon

Frankly, I'm just not intelligent enough to answer that, but it seems logical.. I never thought of it that way....

LCheung - I've heard of results like that in 3DMark2001, especially with tRAS values. Some people were getting lower scores with a value of 5 and better scores with a value of 6, even though the 5 should theoretically perform better. The value of 5 in that case causeed incomplete memory reads and forced the system to redo calculations. So yeah, you're either going to get noticable errors or your CPU is going to do extra work fixing the not noticable errors.

P95 is quick and easy for finding the edge when you are OCing. However, there have been
many times when P95 has failed and everything else ran flawlessly.

For instance, with pc3200 Buffalo (ch-5) sticks I can play UT2k4 and Halo all
day at 220fsb. (Cas 2.0 - 11/3/2)
But P95 fails at 215fsb with the same RAM timings.
If you can use the computer for your every day apps, including intensive games, and not
have any instability issues, who cares if P95 is 100% stable? It's just a convienient tool for
finding the edge ... in my book anyway.