Troubleshooting Suggestions Needed

[dms1971]

I'm having issues with my homebuilt pc. I'm experiencing system hangs and occasional BSOD's at random intervals when the system is a heavy load (usually running p95's torture test or while 3d gaming). Sometimes they'll happen after 5 minutes, other times after 40 minutes.

My system specs are:

Phenom II x4 940be cpu (stock speeds)
Xigmatech S1283V Dark Knight fan/heatsink
Jetway HA-06-GT mobo
Radeon 4850 (also stock speeds)
4 2gb stcks of AllComponents DDR2 800 ram
Lite-On Sata DVD-ROM
WD 320gb HDD
Samsung 750gb HDD
Antec 900 case (w/ 3 120mm led fans, and 1 200mm non-led fan)
Sigma Monster SP750C 750w PSE
Win7 RC1 and Ubuntu 9.04 dual boot (same crashes in both)

I've tried to rule out as many possibilities as I could.

Temps on the cpu and gpu are excellent, even under load.
I tested the psu with a multimeter at both idle and load and came up with very good results on both the +5v and the +12v (5.1v and 12.34v @ idle, 5.1v and 12.33v @ 100% cpu load)
I've tried swapping out videocards as well as running with no discrete cards using onboard graphics.
I've swapped out the drives, running the system with 1 of them at a time to rule them out as a possibility.
I've run memtest86+ through several passes on the ram with no errors.
I just put in a new mobo thinking it was my old one that was causing the problems (obviously it wasn't since I'm still having the crashes )

I've tried upping the cpu voltage a bit and got errors in p95 (errors, not crashes).
Same with the ram.

At this point the only thing I can think of is the cpu itself... Anyone else have any ideas or suggestions before I drop the cash on a new chip?

TIA

 

[tecmo34]

Your system is fine under normal load correct?

Your issue still points to two things... PSU or Heat.
■What are your temps during the brief time you run Prime95 before you crash?
■Do you have another PSU that you can switch out with? It seems to me as you put your system under load, your PSU gives out under the stress. Your power supply has 4 rails of 12v at 20amps, which under stress you may be over doing one rail.

 

[dms1971]

CPU temp is 31c idle, 42-46c under p95 (depending on when it crashes). The temp readings are consistent between CoreTemp, HWMonitor, Everest, and Speedfan (as well as lm-sensors when I run the linux version of p95). The lcd temp readout on the mobo itself matches these temps as well.

COmpletley out of any other ideas, I just downclocked the ram from 800 to 667 and I'm running the p95 test again; so far it's been going for a little over an hour without any issues. That's a lot longer than it has been running without a crash in the past so I'm not quite sure what to make about that. Going to let it run until it crashes or overnight (whichever comes first lol).

But, after an hour of running the In-Place Large FFT test (the one that claims max heat and power consumption) the cpu temp is currently sitting steady at 46c so heat shouldn't be the problem.

Strange though, if it does turn out that the crashes were caused by the ram, shouldn't the memtest86+ passes have picked up the problems?

One way or the other though, I'll follow up here either tomorrow or when the system crashes.