memory passes memtest86 but fails prime95, why?

Thats right, kingston hyperX PC3500.

NF7-S v2 2.9 memory voltage 1.7 chipset voltage.

217 FSB (defualt FSB) 2 - 3 - 3 - 7 (default timings)


Passes memtest, fails prime95 in an instant.


(and it cant be cpu because i can test it at a high multiplier low fsb and it passes)

very odd

Does it pass 3Dmark? Might loop it a few times and see how it holds up.

If you get Microsoft's memory tester and chose the advanced mode it will probably fail after a little while too.

Memtest86 just doesn't do the trick anymore.

what is suposed to happen if the memory fails 3dmark? Does the comp crash? does 3dmark exit? Do you see artifacts?

Right now im 70% through 3dmark2001 and so far so good...


so if this works, why does prime95 fail?


edit: will do microsofts memory tester as soon as 3dmark is done.

3Dmark is one of those tests that seem to test the memory AND the cpu at the same time, and is a good indicator of stability. Prime95 basically tests the CPU only.

You'll know if 3Dmark fails.. but one of these things will happen. 1. Your puter will reboot. 2. The screen will freeze and you'll have to reboot. 3. 3Dmark will crash back to the desktop.

My guess is your memory is fine, but your CPU is borderline. Try dropping the fsb 2 or 3 Mhz and see what happens.

How much voltage do you have runnning to your CPU?

1800+ JIUHB DLT3C running 1.8 volts

passed 3dmark2001 twice

I'm guessing your using 10 x 217 or 11 x 217 for the CPU (or thereabouts)?

blackjackel said:

Thats right, kingston hyperX PC3500.

NF7-S v2 2.9 memory voltage 1.7 chipset voltage.

217 FSB (defualt FSB) 2 - 3 - 3 - 7 (default timings)


Passes memtest, fails prime95 in an instant.


(and it cant be cpu because i can test it at a high multiplier low fsb and it passes)

very odd

Click to expand...

Memtest86 doesn't stress RAM near as much as Prime95. On an NF7-S, the P95 stable FSB will be typically at least 3 MHz lower than the "Memtest86-stable" FSB.