I've got a homebuilt that i've been using for a few weeks:
6gb Kingston DDR3-1600
2 x HD 5850
Coolermaster HAF X
CNPS10x Quiete Cooler
Corsair TX-750W PSU
I had it OC'd to 4.2 GHz, passing prime 95 stress tests w no problems- completely stable. I've been using the computer to play a lot of games (Bioshock 2, Mass Effect 2, and WoW) for the past few weeks with no problems. Yesterday I started getting random shutdowns/restarts with several failed POSTs in between. I assumed maybe it was an OC problem so I set everything back to stock frequencies/voltages. I'm continuing to have the problem. It seems to only be happening during games and there is no BSOD, just a restart with no message. Could this be a graphics card problem, or perhaps a memory problem? CPU and GPU temps appear to be fine all the way until a restart happens. Any ideas would be great. Thanks.
I would start with the GPU's. Since you are running 2. try taking one out and see if the same thing happens. then try the other one by itself too. If the problem still occurs, then try the memory. Take it all out except one stick then try the other stick. For the RAM, you could also run memtest. It can be found on www.ultimatebootcd.com burn the cd and boot from it.
Does it just restart or do you get a bluescreen error first? It might be quick but you should still see the flash of blue before it restarts.
The power supply could also be the problem. If your problem goes away after removing one GPU, but they both work fine individually, it may be your PSU not delivering enough power. I would be surprised though because Corsair makes excellent units and 750W should be plenty.
I set clocks and voltages back to stock, and went to 1 gpu. No problems when using either card. Put things back into crossfire, no problems. Then I overclocked again, nearly the same figures (about 4.1GHz, 1.4V). I ran stress tests in Prime95 for 1.5 hours with complete stability, however when I started playing games again I continued to get the shutdowns. Went back to stock-clocks/voltages and it went away again. This seems to indicate that the problem is with the overclock after all.
For now I'm running at stock speeds and will go for a slightly more modest overclock this weekend.
Other things I did to try to rule out overheating:
- I installed the wheels on my HAF X case to allow more airflow to the PSU (fan pulls from the bottom of the case- it now sits with about 1-1.5 inches of clearance now that the wheels are installed).
- I moved the 2nd graphics card from the 2nd 16x slot (the cards are RIGHT up against each other) to the last slot, which is 8x. I've read that this results in a few less FPS, but I wanted to make sure the card with the fan right up against the other board wasn't causing problems.
Sounds more like a software problem. If it runs stable in prime for hours at a time that tests all the hardware w/ settings. a software glitch will cause shut downs and no bsod. did you recently install any new software or get a system update? New driver update? These will cause that kind of problem.
If you want to continue to inspect your hardware another stress test is OCCT HERE. It will test CPU, GPU, memory and PSU independently and give you readouts of power consumption on the 12v and 3.3v rails as well as temps for CPU and GPU in realtime. Put your video cards back together and take the side off your case first for better unrestricted airflow during the tests and you can see the GPU temps during stress testing to determine if temps are the problem. Halfway through you can put the side back on and see how much difference that makes and if it's that big a difference you might consider more cfm fans.
I started realizing that I was experiencing the problem more frequently after unplugging and re-plugging my headset in. I uninstalled/reinstalled the drivers, and to be safe reduced my OC to a more moderate 3.9 GHz (which I don't mind since it was doable with far less voltage and far lower temps). Haven't seen the problem pop up in a week. Thanks for the feedback- selecting best answer.