Collatz is BSODing my computer

[Maxwell]

So, I have not tried everything people have suggested, but I've gone through most stuff. Next up is trying individual components, but I didn't get there tonight. I do have a couple updates, though...

- I caught the error on the BSOD before the box restarted: some about not successfully recovering from a driver crash. It seems that when I do something crazy like run an ATI WU, it crashes the driver (both 10.10 and 11.2).
- Interestingly, when I'm in CCC with stock settings, and hit the "Test Custom Clocks" button, the clocks fail. With stock settings. Ugh.

I did switch cards around, and things work the same no matter which card is in the first slot. I'm beginning to think this is a PSU issue (though that is little more than a blind guess at this stage). This is a bit disappointing, since I have this PSU, which is supposed to be good.

Other thoughts/suggestions before I begin to cry myself to sleep?

[Fire$torm]

Damn, not looking good

Have you tried just a single 5970?
This would be a pain but you could try a different OS like Ubuntu or Windows XP. With Ubuntu you could use a USB thumb drive with persistence. This would avoid having to mess with the HDD. Another thumb drive option would be Dotsch/UX

More info on Linux USB live installs at PenDriveLinux.com.

Edit: BTW, try the single card test in each of the PCIe slots.

[Maxwell]

Ok. Testing. Using the 11.2 driver, and everything worked fine on the machine for all tests until I used DNETC WUs to put loads on the card:

1. PowerColor card, x16 slot: failure. Restart.
2. PowerColor card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
3. Sapphire card, x16 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
4. Sapphire card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.

I'm skeptical of the probability of having multiple failures simultaneously (not impossible, I know, but unlikely). Which makes me think the PSU is not giving out enough power... does that sound reasonable?

And it is not just a BOINC thing. Even with one card in the system at a time, I can get the driver to crash after using the "Test Custom Clocks" link within CCC using stock settings still.

There goes my morning...

[Fire$torm]

Ok. Testing. Using the 11.2 driver, and everything worked fine on the machine for all tests until I used DNETC WUs to put loads on the card:

1. PowerColor card, x16 slot: failure. Restart.
2. PowerColor card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
3. Sapphire card, x16 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
4. Sapphire card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.

I'm skeptical of the probability of having multiple failures simultaneously (not impossible, I know, but unlikely). Which makes me think the PSU is not giving out enough power... does that sound reasonable?

And it is not just a BOINC thing. Even with one card in the system at a time, I can get the driver to crash after using the "Test Custom Clocks" link within CCC using stock settings still.

There goes my morning...

Hahaha, Welcome to TroubleshootingVille........ Yeppers, it can burn hours away like nobodies business

OK, three possiblities: PSU, motherboard or OS. I was going to suggest trying a different OS first but do this instead. See if one of those IT guys where you work can "Loan" you a PSU tester!!! They have to have one. If they don't then suggest they buy one and let you test it for them

[Harley]

Ok. Testing. Using the 11.2 driver, and everything worked fine on the machine for all tests until I used DNETC WUs to put loads on the card:

1. PowerColor card, x16 slot: failure. Restart.
2. PowerColor card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
3. Sapphire card, x16 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.
4. Sapphire card, x8 slot: failure. Driver crapped out, but managed to stop DNETC before the computer restarted.

I'm skeptical of the probability of having multiple failures simultaneously (not impossible, I know, but unlikely). Which makes me think the PSU is not giving out enough power... does that sound reasonable?

And it is not just a BOINC thing. Even with one card in the system at a time, I can get the driver to crash after using the "Test Custom Clocks" link within CCC using stock settings still.

There goes my morning...

Max, Good news! Its a driver issue. With MW down today (I was not having problems with MW wu's) I switched over to DNETC. When I switched back to MW I had the same problems that you described. Like I had pushed the OC too far. I was able to run MW on for only a short time before it crashed. I was checking everything. I had the card OC'd at 925/1000, which is low for me. CCC and GPU-Z both said everthing was fine. Then I checked a MW wu. The task info said the gpu's were clock at 1000/1500!!! No wonder it was crashing. I removed CCC 11.2 and used driver sweeper. Then installed CCC 11.1 and now everthing seems to be fine. I'll keep you updated.

Edit: I should have said its a driver and DNETC problem. Try running MW and check the task info. Do not run DNETC.

[Harley]

Its been a half hour now with no problems.

[Fire$torm]

@Harley Thx for the info

@Maxwell: Hmmmm, if your problem is the same as Harley's then......

Uninstall BOINC. Uninstall CCC and reboot. Run Driver Sweeper and reboot. Run CCleaner for Registry & HDD, then Auslogics Defrag, then Free Registry Defrag and reboot.

Now you have driver choices, old or new? I would suggest CCC v11.2 (Win7-x64).

Install BOINC and test. If it fails then get the gas can from your car and......
Joke aside, if it does fail then its still a driver issue but the fault is in the OS. Maybe something in the .Net Framework is corrupted since CCC has .Net dependencies.

So the alternative to the above? Swap out the HDD with a spare and install Win7 on the temp drive and try crunching just Collatz and/or DNETC.

Just remembered that Windows does a horrible job of reporting bad hard drive sectors and other faults. Any recent Ubuntu LiveCD distro will report a faulty HDD immediately after booting to the Desktop.

[Maxwell]

I swear to freaking god...

I almost started shouting from the roof tops. I took Harley's suggestion, uninstalled CCC, driver swept, reinstalled 11.1. Tested a couple MW WUs, and they completed successfully. Woohoo!

Decided to test a couple Collatz WUs next, and the problem came back - driver death.

Did I mention I reinstalled Winders today? Bah!

Ok. Going to try the Fire$torm laundry list of activities now, adding in uninstalling .NET and reinstalling it. And punching a baby. That goes on the list too...

[joker]

Be sure to kick your dog in the balls.

[Harley]

I swear to freaking god...

I almost started shouting from the roof tops. I took Harley's suggestion, uninstalled CCC, driver swept, reinstalled 11.1. Tested a couple MW WUs, and they completed successfully. Woohoo!

Decided to test a couple Collatz WUs next, and the problem came back - driver death.

Did I mention I reinstalled Winders today? Bah!

Ok. Going to try the Fire$torm laundry list of activities now, adding in uninstalling .NET and reinstalling it. And punching a baby. That goes on the list too...

What clock speeds did you use? Did you overvolt? I'll give Collatz a try at your settings and see what happens.