PDA

View Full Version : Install watercooling, and BSOD?



ColonelCain
07-18-2009, 11:04 PM
I've been running my new i7 rig under air cooling in an open-bench set-up for a bit now, while I was waiting for the HK 3.0 LT to get back in stock.

I just finished installing my WC parts, and then booted up. The DFI backsplash was slightly artifacted, but go into the BIOS, and it is fine. Boot to windows, and as soon as it gets to desktop, BSOD with atikmag.exe (I think this is the spelling) at fault.


I know that I have the both the CPU and GPU blocks very tight, but neither are warping.

Nothing else was changed.

jkresh
07-18-2009, 11:21 PM
did you have an overclock on air? If so get it back to stock, either way get into the bios and check temps, I would suspect that you don't have a good mount (quick google search suggests atikmag is related to the catalyst control center and might mean something is up with your gpu block).

SoulsCollective
07-18-2009, 11:22 PM
Did you leak-test?

I'd pull the blocks off and look for any water or even just crud, hair, etc caught somewhere, as well as giving the PCI-E contact fingers a good clean.

hellcamino
07-18-2009, 11:30 PM
If you are using a full cover block on your card it probably is shorting something.

biglipzit
07-19-2009, 12:03 AM
I can guarantee that you either got a leak or a short on the full cover block. You could just by a very rare chance have a display driver corruption. Did you attempt to start up in safemode and see if it BSOD's? If it still does then you can narrow it down to something with the cooling system and not software.

ColonelCain
07-19-2009, 09:22 AM
@ jkfresh: No, it is not overclocked, still at stock.
@Souls: I did leaktest, though I'll admit, not 'thoroughly'.
@ hellcamino: It isn't a full cover, just an mcw-60, and a set of iandh's 4870/4890 heatsinks. I know that I didn't do a completely clean job with the MX-2, but that isn't thermally conductive, I think.


@biglipzit: I was able to start it into safemode just fine, actually. That points slightly to display driver corruption, but, how woukld that explain the 'artifacting' in the back-splash, and the BSOD screen itself?

Holst
07-19-2009, 10:00 AM
@ jkfresh: No, it is not overclocked, still at stock.
@Souls: I did leaktest, though I'll admit, not 'thoroughly'.
@ hellcamino: It isn't a full cover, just an mcw-60, and a set of iandh's 4870/4890 heatsinks. I know that I didn't do a completely clean job with the MX-2, but that isn't thermally conductive, I think.


@biglipzit: I was able to start it into safemode just fine, actually. That points slightly to display driver corruption, but, how woukld that explain the 'artifacting' in the back-splash, and the BSOD screen itself?

It sounds like some sort of GPU problem then.

Can you revert back to aircooling and retest?

Most likely cause is going to be the ram heatsinks you have fitted.
If your getting artifacts then something might be shorting out.
How did you fit them? Adhesive or tape?

ColonelCain
07-19-2009, 06:54 PM
It sounds like some sort of GPU problem then.

Can you revert back to aircooling and retest?

Most likely cause is going to be the ram heatsinks you have fitted.
If your getting artifacts then something might be shorting out.
How did you fit them? Adhesive or tape?
Neither, I attached them with MX-2.

I'm going to first try a 4870, then I will try taking the sinks off of the 4890.

Mr. Spoon
07-19-2009, 07:56 PM
Are they individual ram heatsinks, or one of the full-board ram heatsinks? I know at least with my old 8800GTX that had a full-board ram heatsink on it, thermal paste wasn't viable due to the space between the contacts; I needed thermal pads.

ColonelCain
07-19-2009, 09:09 PM
Well, the damn thing is dead now. :mad:

Tried swapping slots, would let me boot, but after I installed CCC (when you switch slots, windows considers it a new device, and you need to re-install drivers), it exhibited the same behavior.
Tried blowing out the slot with compressed air, and dusting off the PCI-E contacts. No dice.


Tried a 4870. Worked fine.

Tried removing iandh's sinks, and just booted it long enough to see the boot-up back-splash. Still had artifacts.
Tried loosening the MCW-60 mount a little. Nothing.
Using q-tips, and pieces of paper-towels, wiped up everything that had 'splatted' onto the PCB, and cleaned the individual ram IC's, and VRM's. Then, installed the stock cooling again. Same problem.
Decided to let try load to desktop, BSOD right after the windows loading screen went away.
Re-started, decided to let it try again. This time, half-way through windows loading, 2 non-uniformally positioned bars, one with a lot of artifacts in it, appeared. When it "appeared" to have finished loading, screen went to standby for even longer than before, but then restarted, w/out a BSOD.

On this boot, post code display on my said "FF", (which means OS start when you have a X64 OS), but still standby. Shut it down, tried again. Nothing. It's dead.


Now, as I wait for tomorrow morning to file an RMA, what in hell went wrong?

Remember, this "artifacting" that I described earlier came on the first boot after installing WC'ing. So VRM's overheating is not a factor.
Also, no traces of water anywhere on the card.

Any ideas?

SoulsCollective
07-19-2009, 09:14 PM
Sounds like you over-tightened something and it either caused a solder crack somewhere, or something overheated causing permanent damage, or you did get some TIM, metal flakes, water, or something conductive somewhere.

Remember to give the board a good scrub before you send it back. It's dead now, so go nuts with soapy water and/or meths/turps, then dry it off with a hairdryer and leave in the sun.

sirheck
07-19-2009, 09:14 PM
HMMM??
Maybe ESD:shrug: