Validity Super Pi in testing memory ?

[eva2000]

seems like the place to ask as alot of you guys use super pi as well..

i have 2 x 512MB Corsair XMS3200PT cas2.5 sticks of memory

if both can pass 24hr memtests (all tests) on same motherboards P4B533 and BD7-II

but only 1 stick can pass Super Pi 32M test, the other stick even fails at FSB100/MEM133

does that mean that 1 stick is faulty ?

or does it mean i have to run memtest for more than 24hrs ???

what's your experience with using memtest and super pi in terms of testing whether a stick of memory is faulty (assuming you already tested the cpu at 1:1 and it passes memtests etc)

[PiLsY]

Dont use the whole memtest - just use certain tests m8 :).

If you can loop test 5 alone for 2 or 3 hours without an error then youre 100% stable on memory. If you can loop test 4 for 2 or 3 hours without error youre 100% stable on cpu.

Super Pi is good as an indicator. I tend to run 1m, then 2m, then 4m then 8m - if it completes all those then I run a memtest. I basically find out a rough maximum @ maxed out mem voltage using super pi, then go to memtest test 5 for a few hours to find the actual stable limit.

If you want to burn in memory then again use memtest. Run test 5 for about 8 hours, then run seti for 16 hours. After that up the FSB a notch, then repeat. WOrks quite well - gained me about 10mhz. Remember though that burning in wont increase timings, only FSB. If you want higher FSB with certain timings then burn in from your max FSB at the timings you want. IE 220mhz 2/7/3/2, burn in at that speed and timings to get up to say 225mhz 2/7/3/2. It only gains you FSB, not tighter timings.

PiLsY.