Mining attempts cause a crash/reboot

[Therilith]

I can play games and such just fine, but whenever I start cgminer or phoenix (or any kind of GPU intensive work like password cracking) my computer instantly crashes and reboots.
For some reason I used to be able to get it working by starting a game (any game) in fullscreen before I started mining, but that no longer works.
One time I actually got a bluescreen instead of just a black sceen and reboot. It identified "ati2dvag" as a problem, so I did all the usual stuff (reinstalled the drivers etc.), but to no avail.

Any help would be appreciated.

Code:

XFX RADEON HD6950 2GB
AMD FX-6100 BE 6-CORE 3.3GHZ
ASUS CROSSHAIR V

[1714899255]

1714899255

[1714899255]

1714899255

[1714899255]

1714899255

[1714899255]

1714899255

[1714899255]

1714899255

[jake262144]

Just this one card in your PC?
What PSU? Please specify wattage, make, and model. Saying just something like "550 W" isn't enough.

Has the card been overclocked? Did you ever modify its BIOS (eg. the shader unlock trick)?
It is possible that this card isn't at 100% health. Mining puts the equipment under a exceptionally high strain. Perhaps games don't provide enough load to make the defect obvious?

Have you tried running Furmark? Do so and report whether it crashes your GPU as well. The test should take at least an hour.
Furmark is free and can be downloaded here: http://www.ozone3d.net/benchmarks/fur/

EDIT:: Is that a Bulldozer I see in your gaming rig? You poor soul...

[Remember remember the 5th of November]

Sounds like a driver problem to me. Downgrade!

[Therilith]

I'm running Windows XP by the way.

Just this one card in your PC?

Yes.

What PSU? Please specify wattage, make, and model. Saying just something like "550 W" isn't enough.

http://www.microcenter.com/single_product_results.phtml?product_id=0376717 (Not the place I bought it)

Has the card been overclocked?

No.

Did you ever modify its BIOS (eg. the shader unlock trick)?

I had exactly these problems for weeks before I tried to unlock the shaders.
The card is fairly new and turned out to be a non-reference model. Didn't even have the dual BIOS switch.
The unlock didn't work, probably because some of the newer 6950s seem to be properly locked.
I restored a backup of the original BIOS after it became apparent that the unlock wouldn't work.

Have you tried running Furmark? Do so and report whether it crashes your GPU as well.

Was in a bit of a hurry, so I only ran it for 15 min. I guess I could do it again for a full hour if needed.
I changed the resolution to 1280x1024 and ran it. The GPU temp reached 95c after about 10 minutes. While mining, it never goes higher than 88.

The test should take at least an hour.

My computer crashes as soon as the miner gets to the hashing part. Shouldn't a GPU stress test cause the same problems just as fast?

Sounds like a driver problem to me. Downgrade!

I might.

[jake262144]

Sounds like a driver problem to me. Downgrade!

I fully agree that this is the path of least resistance. Just fully uninstall the driver (make sure AMD APP libraries are removed) and install older versions, perhaps 11.9 and SDK 2.5.

I changed the resolution to 1280x1024 and ran it. The GPU temp reached 95c after about 10 minutes. While mining, it never goes higher than 88.

88 is pretty darned high if you ask me, even for an inherently flawed product as the 69xx series is. They weren't meant to be built in 40nm you know.
Definitely not the temp to do serious mining at.
Still, it does not necessarily have to be the cause of your crashes.

My computer crashes as soon as the miner gets to the hashing part. Shouldn't a GPU stress test cause the same problems just as fast?

Not necessarily. A stress test might use more memory and clock it higher than a miner does but at the same push the core SPs a little less.

I had exactly these problems for weeks before I tried to unlock the shaders.

You tried to flash an unstable card? Good heavens, you're either extremely brave or just crazy! Welcome to the club