System unstable in 3d games/benchmarks

[ForMatt]

System:
Xp pro 64bit.
129.61 Geforce Drivers.
Intel Q6600 2.4ghz stock settings
2x msi 8800gts 512 oc (factory overclocked settings. Msi software not installed)
1x dual layer sony DVD-RW
4x WD320g Hds in raid 0+1 array
4gb G.Skill 6400cl4d (2x2gb sticks at 800mhz, no overclock)
BFG 800 watt PSU
Asus Striker Extreme MB Bios: 1305
Sound: packaged add-in sound card with the striker board.

Cooling:
Cpu: Zalman 9700led
Case fans: 2 85 cfm air flow fans in front of case, blowing over hard drives.
GPU: Fans pumped to 100/100 (99/99% most likely) via the nvidia control panel. Profile set to load this at OS boot.

* System is stable at desktop, browing, video watching, music, etc.

* Running 3dmark06 or the Fur benchmarking test from Ozone causes system to turn off in 2 to 10 minutes.

* Any 3d game (Wow, Witcher, TF2 crashes in a time frame simlar to the 3d benchmarks.

running single card seems to give me *slightly* more time when running a test/game, but running both kills it quicker.

Temperatures:
CPU 42C unloaded
Gpu1: Unloaded 50c
Gpu2: unloaded 51C
GPU1 ambient: 39C
GPU2: ambient 39C

Loaded temps on CPU were getting up to about 50-51
GPUS were popping up to about 10C more as well.

I'm really frustrated here. At first I thought it was the video cards ( had 2 asus 8800gt 1gb cards ) and returned those to fund the 8800gts purchases

I've tried this with the old cards on XP32bit. Same drivers. Same issue.

I'm at my wits end. I've been trying to troubleshoot this rig for almost 2 months now. Everytime I think it's stable i prove myself wrong

 

[Twizted_3kgt]

Probably have some bad hardware, run a memory test program.

 

[Nitestick]

generally instability in 3D games is due to either power or graphics card issues. however it can also be due to just about anything else! i say first stop is running a memory test (memtest86+) and a cpu stress test such as prime95 or orthos. if they both produce no errors over a period of time (a good few hours) then start suspecting the psu or graphics card. it's easier to do software stability tests than swap out hardware.

 

[Jorsoft]

nitestick is right on the money here. Those temps don't seem high to me so I'd start off with what he's saying. It's doubtful both of those graphics cards are bad based on what you're describing so if you can't single it out with Memtest and prime95 I'd be looking at the psu as the culprit. I see that it's 800w but what are the amps/rails on it? I'm sure they're more than sufficient but just out of curiosity.

 

[ForMatt]

800w. 56 amps on the 12v rails.

Memtest86+ passed. Tested two seperate occaisions in the A1+B1 and A2+B2 slots. 2 full test suites in both. No errors.

Pulled one card. to make sure the 28 amps required of the GTS wasn't somehow combining with the peripherals and killing the system. Still crashed.

Numerous everest stress tests 30 minutes to 1hr30 minute batches. All fine. showing stable temps and stable voltage.

Updated to latest beta 169.44 video drivers. Might be coincidence, but system seemed stable for a while. Was about to get 40mins of gameplay out of it. (This was after I'd stress tested for 1.5 hours on everest to make sure everything was hot).

:/ I'm at my wits end. I'm about to call the MB bad as I'm seeing nothing in the event viewer. This happened with 2 other video cards (8800gt from asus with the glaciator fans).