GTX 580 running hot

[Fastidious]

So I just swapped out my p8p67 asus board and a 2500k i5 for a 2700k i7 coupled to a fatal1ty AsRock board.

I reinstalled my 2 x GTX 580 Gainward 3gb cards and noticed a massive jump in temps on 1 card. Weirdly it was gpu2 not gpu 1.

I'm guessing I've damaged the card somehow but wanted to check with the gurus first (that's you btw).

Any thoughts?

GPU 1 running at circa 50 degrees C
GPU 2 running at 85-90 degress C

Screenshot hopefully attached.

Follow-up: Planning tomorrow to swap the cards over to check if the 'hot' card still performs the same in the other slot. If it does, and i assume it will, should i be looking to RMA or is there a known area of weakness with these cards that I've been unable to find on the net?

 

[M34]

That is quite interesting, I have a 5970 so from past experiences I could tell you, check your drivers, cords, make sure everything is plugged into it, 6pin/8pin whatever.. and also check the airflow that its getting in enough air suction, make sure its not crammed in.

 

[Fastidious]

Thanks for the input M34.

The drivers are currently the beta Nvid drivers and seemingly work very well indeed. I did roll back to a 'release' version and got the same results.

i swapped the cards over and checked for damage, badly connected fans etc., and everything seems fine. Interestingly the 'hot' card now performs fine and the other card now runs hot. I guess this means that the burn-in software I'm using isn't spreading the load between the 2 cards.

So is it normal for a gfx card to run at 90C? Is it going to cause any long term damage?

Should I find a bottle of whiskey and some sleeping pills?

 

[Ivy]

It totaly depends on the conditions, if the 90° is caused at 100% load then the GPU will still be fine, even if the temperature is at the upper limit.

A GPU basically should be able to handle up to 100° or even more without getting any damage, however, its not good for PC health when it is continuously at such a temperature level. Take into account that the entire PC can heat up and its not good for the electronical parts inside. So for safe and longtime use, 80-90° is surely the temps which should not be exceeded, and i mean it at 100% load, not at idle nor what else.

Can there be to much cooling? Basically yes in term the cooler gets very loud and unpleasant and if the cooling is higher than truly needed. As i said, a good tuner is trying to keep a GPU at 80-90° at max load, and every cooling stronger than that can cause a pretty noisy fan. Unless its water cooled, but thats usualy not the case for the usual user, or any user with small and compact systems.

Just one thing: If the GPU is running at 50° and its a high end GPU such as the 580 GTX, then it is NOT at full load. Its impossible to have such low temps over air cooling... forget about. Its roughly 250 W TDP, thats very heavy for air cooling. Especially because the nowadays most effective vapor chamber cooling can NOT work effective at 50°, they run more effective as hotter it gets. At above 70° they got max efficiency because the pipe need a certain temperature to start to work in a powerful manner. From what i know the heat works kinda like a "engine".

 

[Fastidious]

Hi Ivy,

I've just installed a Corsair H100 WC system for the CPU and it runs sub 50 degrees at 5.2ghz.

The GFX card test I posted yesterday were at 99% load as per a usual stress-test. I am less worried about the high temps now that I have read your post so I thank you for that.

I ran Furmark overnight and studied the log this morning. Both cards performed OK. 11 fps on full settings with GPU1 maxing at 75C and GPU2 typically registering 5-8 degrees lower. 11fps seemed low but after trawling through several more forum posts, it seems that this is OK.

Your thoughts, as always, are appreciated.

 

[hydrata]

It sounds like you've solved your issue, but just in case: check that SLI is enabled. I'm no expert or even knowledgeable in this area, but I recall from the Battlefield 3 thread I've been reading that some people had issues with upgrading where SLI would be disabled when drivers were updated.

 

[Fastidious]

Thanks Hydrata, SLI is enabled after update but I had to check just to make sure

 

[Ivy]

Indeed. it seems like the second GPU was somehow disabled, and as i said dont worry about a 100% load peak of 80-90° thats pretty normal for high end GPUs over air cooling when the cooler is not at full load (and i hope it isnt since it can hurt the ears lol). My cooler at 80-90° tops at about 60-75% fan speed and thats inside a SFF PC and the GPU is 10% OC. Its ok to play with but i do not wish to have the cooler at 100%, thats just to loud. Sapphire cards for example, tuned by factory, only go to 100% speed in term the GPU is exceeding 100°... and i did never reach that (im glad). At idle (watching video files or browsing) the cooler is at 24% speed or even lower, then its barely hearable. Only when i play games it need lot of cooling.

 

[Fastidious]

Similarly, ambient temperature at about 20C and gpu's running at late 30's without stress. In-game with DX11 particularly, running at around 70-75C. From what you've said that sounds reasonable and enables the vaping chambers to begin working properly.

Thank you for the input, I'll report back if there are any other issues (for documentation purposes ofc ^^)

 

[Zantal]

SLI gets disabled after update. always.
that was the problem. And sometimes i forget to enable it again too (like playing for an entire week with SLI disabled)

 

[RADIO_ACTIVE]

And sometimes i forget to enable it again too (like playing for an entire week with SLI disabled)

I have so done this before