I work in the RMA dept. of a computer product reseller. I do testing on "defective" products for verification purposes. I am wondering if I can get some feedback on testing procedures for motherboards.
I use a "workbench" setup of known good components(ie. CPU/heatsink, RAM, power supply, hard drive, cdrom, keyboard, mouse, AGP video card). If the motherboard will POST, installation of Win XP is succesfull, necessary drivers install o.k I proceed to stability testing.
Currently, I favor Prime95 for 14+ hours. I also use Sandra Burn-in, looping 3dmark2001se.
With "known good" components a prime95 error would have to point to the motherboard, but I find it a intermittent/unreliable source for a defective motherboard. What do you think?
Prime95 is much more a test of the CPU than anything else.
Sandra's burn-in testing everything would probably be the best all-round motherboard test I guess.
For my own stability testing (when overclocking usually) I tend to leave Sandra burn-in (CPU & memory) looping, while simultaneously leaving 3dMark2001 looping & playing Mp3s in Winamp. This of course doesn't really thrash the LAN, PCI bus, or HDDs much, but I'm not overclocking those :smile: . The MP3s are useful because you can then hear if it's crashed (the music stops playing), plus it's actually providing some entertainment.
So I would say ditch prime95, and stick with Sandra's burn-in on a loop for a few hours, but running ALL the tests (not just CPU/MEM like me)
I have a stick of ram that works fine on my abit board, but fails prime95 on an asus board. The asus board is fine with a different stick of ram by the same manufacturer. There just isn't a surefire test.