I usually avoid posting because I know someone has had similar problems fixed before, it's just a matter of searching...but I've been searching for days, and have not been able to fix my problem.
I have a brand new, week old system (Specs below) that I built myself that I haven't been able to get stable under load at stock speeds. When using Prime95 blend test, the computer will blue screen after about 5 minutes. I can run 4 cpu clients of folding at home all day without a problem, but when I try and run the GPU client, I'm greeted by the BSOD within 5-10 minutes. I've seen 2 different blue screens, IRQL_NOT_LESS_OR_EQUAL and MEMORY MANAGEMENT.
When not under load, I can run all day and night without problems.
Asus P6T deluxe V2
i7 920 D0
6GB OCZ 1600MHZ DDR3
2x 640GB Caviar Black
EVGA GTX260 core 216
Vista Home Premium 64-bit
Things I've tried:
-Testing 1 stick of ram at a time. Everything is fine, Prime95 runs without errors until I stop it. I've tested each stick this way.
-Testing 2 sticks of ram at a time. Again, everything is fine. I've tested every combination of sticks this way.
-Memtest86+ passes without errors on 1, 2, and all 3 sticks
-In the BIOS, everything to Auto...does not help stability
-DRAM voltage to 1.64v...does not help stability
-QPI/DRAM voltage to 1.35...does not help stability
-I've done a fresh install of Vista, which didn't help
-I'm currently on Vista service pack 1, but had service pack 2 before the fresh install
I've been going crazy trying to find the solution to this problem...any help would be more than greatly appreciated!!
1) Corrupted drivers (unlikely due to several installs, one fresh OS).
2) Heat. You do not mention any temps. Using CPUID Hardware Monitor, what are your cpu and gpu temps at blue screen? (Most likely a gpu temp issue if any. However, hotter case air under graphics load may push cpu temps too high.)
3) PSU that cannot deliver cleanly when asked for graphics power. Only test is use a different psu, or test this psu in a comparable system. (Most likely IMO)
4) Bad graphics card. Same test as in #3. (Second most likely)
I may have fixed it, but I did 2 things at once so I'm not sure what did the trick.
I finished installing some windows updates and restarted the computer. But while the computer was off, I rigged up a spare fan that came with my motherboard to blow directly on the RAM. When I restarted, I was able to run Prime95 for over an hour until I stopped it. I folded with 4 cpu clients and 1 GPU client all night without any problems.
I think my problem may have been overheating RAM. I'm not sure if this is a problem that people have (especially at stock settings), but it explains a lot. This would explain why it takes 5 minutes or so of heavy load before it blue screens. This would also explain why 1 and 2 sticks worked...they produced less heat. Also explains why changing voltages in the BIOS didn't help (maybe even made it worse). Anyway, everything is good for the time being. Thanks for the help.