GTX580 shows crazy temperatures and shuts down

Colonel_Panic

Limp Gawd
Joined
Oct 5, 2009
Messages
328
Hello [H] friends, I'm hoping to probe the collective intelligence of [H]ard to solve a problem that has flummoxed both my friend and I for quite some time. Previously I posted about an issue with triple SLI GTX580 and since then I have some additional information.

The initial issue is that my friend's system with 3x GTX 580 cards would not function in SLI mode. After loading the cards, the entire system would shut down. Tonight, aided by using EVGA Precision to display % and temps on a 3DConnexion SpacePilot Pro, we were able to see that the temperature of the card spikes to 100C or more. Since the entire system is cooled by water, we have no doubt that the temperatures are incorrect. What could cause that?

The strangest bit is that we disabled SLI and still have the same behaviour when using individual cards. When I ran Furmark with a single card, the temperature climbed rapidly up to 100C and the system shut down. Since it then happened again when using the second card as the primary, we thought it would have to be something further upstream, perhaps mobo BIOS.

Google searches yield similar problems. Any suggestions? Where else should I look?
 
Why are you so sure those temps aren't correct? Is it possible that your waterblocks are not correctly installed? What's the setup of the loop?
 
Either a big air bubble in the water system or a block inst installed right.
 
Why are you so sure those temps aren't correct? Is it possible that your waterblocks are not correctly installed? What's the setup of the loop?
The physical cards don't feel nearly hot enough, and are merely warm to the touch. We recently removed all of the blocks and installed them again. Additionally, if Card 1 has the monitor, and is at 70C, when we switch the monitors to Card 2, Card 1 is 50C and Card 2 shows 70C.

The water system is kind of complicated, as we're using the opportunity for some research and development. As a result, there are 3 loops:
1. 1x CPU > 3x GPU > heat exchanger > Zalman Reserator > Pump
2. 2x mosfet > 1x NB/SB > RAM coolers > Zalman Reserator > Pump
3. Heat exchanger from #1 > Powered radiator > Pump
 
I returned a Gigabyte GTX 580 because of a similar issue. I don't actually think the GPU was getting over 100 deg C but maybe the temperature sensor wasn't accurate and was reading higher than actual. The video card must have some thermal protection so if it thinks its over 100 for a certain period of time then it will shut down.
 
I returned a Gigabyte GTX 580 because of a similar issue. I don't actually think the GPU was getting over 100 deg C but maybe the temperature sensor wasn't accurate and was reading higher than actual. The video card must have some thermal protection so if it thinks its over 100 for a certain period of time then it will shut down.

That was my thought too. But then testing a second card resulted in the same problem, which sounds like something different.
 
Back
Top