cancel
Showing results for 
Search instead for 
Did you mean: 

PC Fails to POST after swapping GPU's around

Malachor
Level 9
Hi All,

Could someone kindly assist me with the following issue please?

My new PC has been up-and-running now for about 3 months with no issues, apart from GPU2 running way too hot (I have 2 x MSI 7970 Lightning’s in Crossfire). By hot I mean during gaming GPU1 (PCI-E x16 Slot 1) is ~42 degrees and GPU2 (PCI-E x 8 Slot 3) is ~78 degrees. Even though it’s been running hot, I’ve had no issues apart from micro stutter/tearing in Skyrim when enabling crossfireX mode within CCC.

I decided to re-apply thermal paste to GPU2 so I spent about 4.5 hours dismantling my water loop and dismantling the card. I think on my first attempt I used way too much paste, so I removed it using thermal compound remover and then re-applied the TIM (Prolimatech PK-1) using the spread method /w credit card; paying close attention not to create any air bubbles or make the layer too thick or uneven. I previously used the “X” method. I also took this opportunity to swap the GPU’s around so GPU2 was now GPU1 and vice versa (both cards linked by an EK Plexi bridge). I re-assembled to card, applying uniform pressure and double checked everything was ok with the thermal pads/screws etc. Looked good, so I re-assembled my PC, refilled the loop, did some basic leak testing before attempting to power on fully.

So, before I continue, the ONLY things that have changed from an otherwise working system are:

• Thermal paste re-applied to GPU2 (much neater and half as much TIM used, but GPU die completely covered)
• GPU’s have been swapped around so the above card, which was on the bottom (slot 3) is now on the top (slot 1)

Upon powering up the PC, the system fails to POST and the motherboard (ASUS Rampage IV Extreme BIOS 3404) “Boot Device LED” is red with error code 62. I get no visual display on either of my two monitors.

Prior to swapping the cards around, both monitors were connected via DVI to the card in slot 1, and this worked fine. Now the cards have been swapped, I get no display if I plug the DVI cables into the top or the bottom card, either single or double.

So, in order to start troubleshooting, I disabled PCI-E lane 1 (which now houses the card that was too hot and previously in PCI-E slot 3). I then connected one monitor to the card in slot 3 (previously GPU1) and successfully booted into windows. I run a quick Heaven benchmark to test temps and they were fine (as they always were on this card). Whenever there is power to the other GPU, the system fails to POST and I get no display. One thing I might add is, when GPU1 (previously GPU2) powers on, the blue lights on the back of the card light up in sequence (I think there’s about 8 of them) then they all go out and the card goes into what looks like power saving mode? This card used to do this before when it was in the other position until I played a game in crossfire /w full screen, and then the card would light up fully and look like its being utilised. I would have thought that swapping the cards would have swapped this feature i.e. what used to be GPU1 now powers down when not in use, but it seems to have “remembered” the previous positions relative to each GPU because this one just won’t stay on longer than ~20s after powering the machine up.

Things I have tried:

• Reseating the cards and power cables (there is power to the EZ-Plug 1 and 2 on the mobo)
• Unplugging USB devices before/during boot in case there was a conflict somewhere
• Disabled USB front panel headers
• Unplugged all peripherals i.e. Blu-Ray drive, fan controller and all other non-boot HDD
• Booting into Windows (by disabling PCI-E Lane 1 or unplugging power to GPU1 – only way to get into Windows) using the AMD Driver Cleanup tool to uninstall everything, reboot, then re-install latest 13.3 Beta Drivers

What I haven’t tried (because of the amount of time it requires to disassemble/reassemble the loop, drain the GPU’s/Plexi bridge, refill etc etc [I don’t have much time as I have a hectic life])

• Putting the GPU’s back in their original positions

If there’s anything you guys can think of that I might have missed or some further testing I can do to isolate the problem I would be eternally grateful. I can provide much more detail if required, and photo’s of the before/after TIM on the card that’s giving me grief if needed. Thanks very much for all your time and I look forward to your responses with great enthusiasm.

Kind Regards

Mal.
40,343 Views
61 REPLIES 61

Arne_Saknussemm
Level 40
If I were you I would start up with the card in slot 3 and un-install all drivers with driver sweeper or Driver Fusion (in safe mode) and then re-install.

But I have to say...if you have a card that runs hot maybe putting it as the first card in SLI or Xfire might not be the best idea since that card always works harder....

Malachor
Level 9
Thank you for the prompt response and for your suggestion.

I will try that however I thought the temps purely attributed to bad TIM application so thought it wouldn't matter which card is the primary so long as the temps are the same for individual benching.

If the driver sweeper suggestion fails; is there anything else I can try?

Many thanks.

Mal.

graphic
Level 11
have you tired the other card on its own ?
I can appreciate been water cooled its hard to do but ...... i can't think of anything else
except maybe some setting bios ?

Malachor
Level 9
I have tried the card on its own in so much as I have disabled all other PCI-E lanes using the mobo trip switches (which essentially stops power going to those devices) and it doesn't work. I haven't tried it on its own in the manner you mean (just that card in the system and nothing else). The only options in the BIOS relating to GPU's are for PCI-E Lane simulation and setting the speed to GEN3 or GEN2 (which I've also tried to no avail).

Arne_Saknussemm
Level 40
Sorry Malachor I missed that the 78 degrees was on water cooling! That temp for water is way too high and suggests some problem with the loop.....as you suggest with mounting etc. is your inlet and outlet port on that card the right way round etc.

Just make sure it is getting flow before anything else

According to the instructions provided with the EK Plexi Bridge, you can use the inlet as outlet and vice versa (basically any permutation); but, a picture speaks a 1000 words. Here is a picture of how my system is set up. It's still exactly like this put the GPU's are swapped around. I should also add that I don't think it's a flow issue, because, when GPU 1 was in the top slot, it idled at 32 degrees. Now it's in slot 3, it idles at 28 degrees (so if anything, the hotter card benefits from being here). Check the pic below and see what you suggest.

20466

graphic
Level 11
hmm, something up with that card
you mentioned you took waterblock off, I'm wodering if something went wrong installing block?
what about putting original casing back on that card, or re installing block

Malachor
Level 9
If I have to I will but it's 4~5 hours per attempt and finding the time to do that amidst a heavy work/gym/family lifestyle is really tough.

Arne_Saknussemm
Level 40
I would try the driver wipe and clean install first and see if it fires up and at what temp and then go from there....just make sure it is seated properly in it's slot etc. and power connections are seated well.....

Looks a nice setup but that temp on that card was a high load air cooling temp not a water cooled temp.

Are your GPUs cooled in series or parallel?