strange temps with gtx 280 sli loop (PIC)

[sonofander]

my GPU loop components are in my sig...

at idle both cards are at 38-40c - totally ok.

when I put them under load, I notice that my temps are not even between cards. The first card in my loop loads at 45c but the second goes straight to 65c! WTF???

perhaps it's do to the rediculously short run between cards and the fact that the fittings are pretty much touching?

but if the flow was being disrupted, would the temps on both cards suffer??? In the 2 years I've been wc'ing I've never seen a gpu loop with such drastic temp differences between cards.

Any ideas as to what might be wrong? fix it please!!!



here's a pic for reference

http://i37.photobucket.com/albums/e82/sonofander/DSC00345.jpg

[Chruschef]

possibly a remount?.. bad contact?

thats a bit weird, obviously the loop is working quite well because of 45*C on the first card..

have you used this GTX280 before, without water cooling? If so, what was the temperature difference between the two or what where the temps of it before?

[DarthBeavis]

some restriction in the second card . . .take the loop apart and do the blow test . . .I have tri sli in parallel and get MUCH lower delta . . .

You could also try switching the cards . . .you will see higher temps on both if I am right

[sonofander]

possibly a remount?.. bad contact?

thats a bit weird, obviously the loop is working quite well because of 45*C on the first card..

have you used this GTX280 before, without water cooling? If so, what was the temperature difference between the two or what where the temps of it before?

yeah, had both cards on air and water before... never had this issue.


some restriction in the second card . . .take the loop apart and do the blow test . . .I have tri sli in parallel and get MUCH lower delta . . .

You could also try switching the cards . . .you will see higher temps on both if I am right


not sure what you're reccomending here?

[sonofander]

some restriction in the second card . . .take the loop apart and do the blow test . . .I have tri sli in parallel and get MUCH lower delta . . .

You could also try switching the cards . . .you will see higher temps on both if I am right

are you saying that there's something in my block? or that my barbs touching each other are the prob? I'm confused

[thehighlander1]

i think hees saying to switch the card order

so put the bottom one ontop and vise versa

[DarthBeavis]

i think hees saying to switch the card order

so put the bottom one ontop and vise versa

bingo, tells you either something is in ur block or u have a bad mount

[sonofander]

K thanks. Will redo gpu loop tomorrow and let you know

[T_Flight]

The Golden Rule of Diagnoses...isolation. If you ever have a problem remember that word. If the problem moves with the card, then you have isolated it.

[MomijiTMO]

Pretty weird. I'd say bad mount.

Good luck :)

[b@llz0r]

yeah since they are in series both cards will get the same flow regardless of wether one is obstructed.... id go for a re-mount too.

[exe163]

The Golden Rule of Diagnoses...isolation. If you ever have a problem remember that word. If the problem moves with the card, then you have isolated it.

+1. Possible problems: bad mount, fubar card, fubar sensor, software read the temperature wrong.

[hellcamino]

I would go for the re-mount.

[Salad Fingers]

make like a bull on a cow and remount

[septim]

try a remount on that 65c card... (possibly bad contact or uneven pressure on where its needed.)
and since you're taking it apart, you might also want to check the interior, cant hurt to see inside...

[2low4u]

Which card is your monitor plugged into? That should be the card that gets cooled first & it's going to be the hottest of the 2.

[sonofander]

UPDATE:

so it has nothing to do with the water loop... I bled the loop and popped the problem card to check the mount. Everything seemed ok so I put it back in, but just to check fired the machine up without the GPU loop running.

Real quick put some load on the cards and card 1 instantly jumped 20c hotter than card 2.
I am at such a loss right now. taking it apart to reseat it again, but wtf?

Pretty sure I have the damn thing mounted right and as good as can be. Is there a chance that the temp sensor went bad or something?
I'm so confused (

[DarthBeavis]

UPDATE:

so it has nothing to do with the water loop... I bled the loop and popped the problem card to check the mount. Everything seemed ok so I put it back in, but just to check fired the machine up without the GPU loop running.

Real quick put some load on the cards and card 1 instantly jumped 20c hotter than card 2.
I am at such a loss right now. taking it apart to reseat it again, but wtf?

Pretty sure I have the damn thing mounted right and as good as can be. Is there a chance that the temp sensor went bad or something?
I'm so confused (
I had two out of six 280s have a heat bug which caused the card reported temps to skyrocket (sure the actual temps where not what they reported to be). A normal card would idle at maybe 35-40 where the bugged cards would idle 65-70, temps reported at 80 plus at load. I tested on identical loops . . .Problem is NOT vendor specific but goes back to the base card itself (third party companies often make AIB partner cards if they are reference design).

[sonofander]

so what do i do? ignore it?

other thng that bothers me is that this wasn't a prob before i went back to water...
card stock air = fine
move to water = fine
back to air = fine
back to water = fubar???

doesn't make a lot of sense

[sonofander]

LOL WTF ever!

still dry, I swapped the cards so that
gpu1--> slot2
gpu2--> slot1


now gpu2 is 20c hotter! :rofl:

so whatever card is in slot 1 runs 20c hotter???
I'm still laughing over here...

so my issues have nothing to do with my GPUs, GPU blocks, WC loop or anything else that belongs in this section :up:

yay me yay

[T_Flight]

hmmmm, whichever one is in the top slot runs hotter...

Flow problem? Flow rate problem? Bubbles getting trapped? Weak/faulty pump? I'm just thinking out loud trying to guess at what possible causes are. Others probably can help you a bit more.

[Linchpin]

That is just plain bizzarre.

[DarthBeavis]

does not make sense.

[2low4u]

Which card is your monitor plugged into? That should be the card that gets cooled first & it's going to be the hottest of the 2.

Did you try plugging in the diff cards & testing to see which one then ran hotter?

[sonofander]

hmmmm, whichever one is in the top slot runs hotter...

Flow problem? Flow rate problem? Bubbles getting trapped? Weak/faulty pump? I'm just thinking out loud trying to guess at what possible causes are. Others probably can help you a bit more.

I'm running DRY = no water flow, no pump, no nothing. Just wb on GPU


does not make sense.

no it sure doesn't

driver reinstall maybe?

[DarthBeavis]

I'm running DRY = no water flow, no pump, no nothing. Just wb on GPU



no it sure doesn't

driver reinstall maybe?

Do you have another mobo to test?

[DarthBeavis]

also, you have three slots? try not using the slot that gives you bad temps . . .

[2low4u]

*Looks around, knocks on glass Hello anyone there* :rofl:

[Sadasius]

I know what your problem is......It's simple actually. You need the spacers on the copper side of the block or you will block the water getting out and heat it up making your last card hotter. The bottom part of thats Acetal and does not need a spacer, but the copper top does as it is thin and the barbs or plugs will block the inlet and outlet. I may be wrong but I could not see them in your picture.

[sonofander]

Do you have another mobo to test?

one and only one mobo :(


also, you have three slots? try not using the slot that gives you bad temps . . .

3rd slot is only x8 so even if that would work, I don't wanna know it :sofa:


I know what your problem is......It's simple actually. You need the spacers on the copper side of the block or you will block the water getting out and heat it up making your last card hotter. The bottom part of thats Acetal and does not need a spacer, but the copper top does as it is thin and the barbs or plugs will block the inlet and outlet. I may be wrong but I could not see them in your picture.

I think by duplicating my results w/ no water flowing, I've eliminated all possible problems that would have to do with flow. But for what its worth, I used all included hardware in my waterblock mounting - spacers and rubber grommets included



fwiw, I'm refilling my loop right now so that I can boot back in and try a driver reinstal...

[sonofander]

loop back up and running...

now after having done nothing but put them back in my pc with water, I only have a 10c difference between cards per everest and its only being reported as a 4c difference through GPU-Z

oh what the helll

lol

[JMCorp]

my guess given the information to date is over voltage somewhere. my guess is you didnt keep track of which pci-e plug you put in which card during the remounts.

could be your psu or your mobo, but most likely the psu.

[sonofander]

positive that I plugged the same psu plugs into the same card each time (2 are black, 2 are red - red go on top)

anyway, I have no idea why but the prob is completely gone now. after reseating the cards in the pci-e slots and reinstalling the drivers, both everest and gpu-z now report a max 5c diff. between cards when running 3dmark @ 2560*1600 AA and AF maxed

I guess i dont need to know why as long as the problem is solved...

thanks for the help guys