Crossfire 7970 Top card hotter then bottom

D

Deacon_Jones1988

Limp Gawd
Joined
Dec 8, 2012
Messages
159
I have 2x7970s in CFX, During Idle and gaming I have noticed a big difference in temp on the cards. For instance the top card is always warmer then the bottom card. Right now at idle my top card is idling at 36*C and the 2nd card is at 28*C.

The fan speeds on both cards are at 55%. I have a fan installed on the inside that is at the end of my 2 gpu's its the bracket that the Phantom 820 has that tilts. Its attached to the HDD/SSD cage.

Would the fan be causing my heat variance? (Because given it's location the air from the fan would move over the 2nd cards GPU, it flows between both cards.) Or would the heat from the 2nd card be traveling upward onto the other card causing the variance?

If pictures would better explan my setup let me know and I ll take some and post. Thanks in advance.

Stay [H]
 
Your bottom card is using cool air to cool itself, your top card is using hot air from the card below it to cool itself, therefore, the variation is expected and normal. In general, good aftermarket coolers will work just fine and you won't have to worry about taking any extra measures to protect it - if you see the top card trending above the 90-95c mark at 100% fan speed, that's when you can start worrying...
 
Yeah, completely normal. I also had 2x7970 at one point, and the top one was always 10-15C higher under load.
 
Thanks for fast responses. So: "make sure top card stays below 90*C." Got it. Depends on what game i play it can (top card) get into the 80s. I have always watched my card temps very closely. Because I have had a few get too hot before by being careless.

Once again Thanks for the replies. I wasn't sure if this was normal, given my understanding of physics and such I suspected this could be normal.

Stay [H]
 
Deacon_Jones1988 said:
Thanks for fast responses. So: "make sure top card stays below 90*C." Got it. Depends on what game i play it can (top card) get into the 80s. I have always watched my card temps very closely. Because I have had a few get too hot before by being careless.

Once again Thanks for the replies. I wasn't sure if this was normal, given my understanding of physics and such I suspected this could be normal.

Stay [H]
Click to expand...

The response is a bit more nuanced than "make sure the card stays below 90". If the fans are running at 50% and the card is at 90c, then that is not a problem. If the fans are running at 100% and you're at 90c and climbing, then that could be a problem. (as the cards have fan profiles that are designed to protect them, and 90c is a safe operating temperature). The point of the 100% + 90c+ is that your cooling unit has no more capacity to dissipate the heat without help...
 
David_Schroth said:
the cards have fan profiles that are designed to protect them, and 90c is a safe operating temperature). The point of the 100% + 90c+ is that your cooling unit has no more capacity to dissipate the heat without help...
Click to expand...

Wow, I wasn't aware that a gpu could operate at that temp safely. Just to clarify, are you saying that 90*c is a safe temp for a card to reach momentarily, (just for a few short minutes.) or are you stating that you can run a gpu at 90*c for hours on end at 90*c and not be in danger of bricking the card? I'm not knocking your opinion, your knowledge, or doubting the merit to your statement/theory or whatever, I see your title up there and number of years. :) I have just never heard that before. I have always been told that anything over 85*c should always be avoided.

Thanks in advance.

Stay [H]
 
Deacon_Jones1988 said:
Wow, I wasn't aware that a gpu could operate at that temp safely. Just to clarify, are you saying that 90*c is a safe temp for a card to reach momentarily, (just for a few short minutes.) or are you stating that you can run a gpu at 90*c for hours on end at 90*c and not be in danger of bricking the card? I'm not knocking your opinion, your knowledge, or doubting the merit to your statement/theory or whatever, I see your title up there and number of years. :) I have just never heard that before. I have always been told that anything over 85*c should always be avoided.

Thanks in advance.

Stay [H]
Click to expand...

The temperature limit will vary from GPU to GPU and manufacturing process to manufacturing process. For example, the reference 290X cards are configured for a full load temperature of 95c, and the fans and GPU clock will adjust to hit that temperature target. When you look at the whitepapers for the silicon, you'll see that TSMC process is rated up to 125c (if I'm remembering right), so there's still plenty of safety margin built into the card's design. AMD and the AIB partners wouldn't be willing to engineer, produce and provide a warranty for the cards if they didn't have a reasonable expectation that it would last through the entire 3 year warranty period for heavy users. I also ran a pair of 5870's at 90c doing Bitcoin mining for 2-3 years and had no issues with their longevity.

I'm sure some previous generations of GPUs had a 85c maximum threshold before trouble began, but for the most part, a lot of people will come up with an arbitrarily selected number, such as 85c, that a particular component should never exceed even if a higher temperature was selected by the manufacturer of the part based upon their analysis and testing.

For the tl;dr answer, I'll go with "it depends".
 
Check what voltages do the 2 cards run at (if you are using afterburner make sure you enable "unlock voltage monitoring") and put the one with the highest voltage on the bottom, that will reduce the temperature delta between them.
 
Thanks for info. Given what you have said, I should be good, the cards should never die due to temps, my cards never get over 82*C for the most part. Also, shouldn't the GPUs have a fail safe built in where they will just shut off when they get to an unsafe temp? Like cpus do, my 8350 shuts off when it hits 67*C

I do use Afterburner, I tried to enable the voltage monitoring feature, it just won't work for some reason, it doesn't give an error just doesn't enable it at all. Maybe my cards aren't supported for some reason. I am not sure which voltage reading i should refer too but the vddc of the cards range from 0.90-1.10 i believe when they are in high usage they go up to 1.10

Thanks in advace.

Stay [H]
 
You must log in or register to reply here.
Back
Top