• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

GPU Temp - 6900XT suddenly running hot.

Soldato
Joined
12 Sep 2003
Posts
11,108
Location
Newcastle, UK
Hey all

Fired up dead space 2, as I've never completed it and I haven't gamed in a few weeks! I recently bought a new monitor so I thought I'd try the game out in 4k with HDR. Loving it so far, but... what isn't "cool". See what I did there lol is that my GPU fans are running really fast. I loaded up the AMD metrics as I thought perhaps my undervolt had reverted back to stock (400w), but nope, still sipping just 200w.

GPU-Metrics.jpg


I was going to potentially sell the card but not with fluctuating 85 - 90ºC hotspot (I think max temp is 110ºC). I'm not sure what the fan curve takes into account. GPU Temp or GPU Hotspot but at 60ºC I should only be running at 35%. Over 80ºC is 60% (which is loud). So it seems to be using Hotspot.

I've never noticed a 30ºC difference before, and this isn't after a long gaming sesh, this is straight away it shoots up to that temperature. I saw this - https://community.amd.com/t5/pc-graphics/6900xt-max-temp/td-p/441952 - but I read that as if the card was running stock. Mine is undervolted and hits that 90ºC hotspot instantly.

Time to crack it open or just leave it alone? The card is only 2 years old.

Cheers.
 
Hey all

Fired up dead space 2, as I've never completed it and I haven't gamed in a few weeks! I recently bought a new monitor so I thought I'd try the game out in 4k with HDR. Loving it so far, but... what isn't "cool". See what I did there lol is that my GPU fans are running really fast. I loaded up the AMD metrics as I thought perhaps my undervolt had reverted back to stock (400w), but nope, still sipping just 200w.

GPU-Metrics.jpg


I was going to potentially sell the card but not with fluctuating 85 - 90ºC hotspot (I think max temp is 110ºC). I'm not sure what the fan curve takes into account. GPU Temp or GPU Hotspot but at 60ºC I should only be running at 35%. Over 80ºC is 60% (which is loud). So it seems to be using Hotspot.

I've never noticed a 30ºC difference before, and this isn't after a long gaming sesh, this is straight away it shoots up to that temperature. I saw this - https://community.amd.com/t5/pc-graphics/6900xt-max-temp/td-p/441952 - but I read that as if the card was running stock. Mine is undervolted and hits that 90ºC hotspot instantly.

Time to crack it open or just leave it alone? The card is only 2 years old.

Cheers.
A big delta between the gpu temp and hotspot usually means a re thermal paste job unfortunately. Although you bought it two years ago. It could been in a warehouse for who knows how long.
 
Last edited:
Thanks both. Played for an hour and it didn't go past 90ºC, but yeah not ideal when drawing so little wattage vs stock.

Re-paste here we come. :)
 
Some GPU makers are famous for using cheap thermal paste on their boards. Or cheap poorly fitting thermal pads.

For some tips, check out Northwest Repairs on YouTube for how to open up your GPU, and apply the paste.
 
Thanks both. Played for an hour and it didn't go past 90ºC, but yeah not ideal when drawing so little wattage vs stock.

Re-paste here we come. :)
Or, I hear an under-volt of the card might buy you some time. In temperature jumps.
 
Cheers all. Spot on advice. Cracked it open this evening. Two points I noted. First, the lack of paste. lol. Second, one of the caps had no thermal pad. Managed to scrounge a bit of excess from elsewhere thankfully and stretch a bit across. Now making contact with the metal base.

Popped on some MX-6. Fired up Dead Space 2 again for 5 minutes or so but instantly better.

GPU-Old-Paste.jpg

GPU-Missing-Pad.jpg

GPU-Clean.jpg

GPU-Bodge.jpg

GPU-New-Paste.jpg

GPU-Temps.jpg


We are back in business. 200w, 98% utilisation, is seeing 52ºC and +10ºC for hotspot/junction. That's a GPU reduction of 5-6ºC, and a hotspot reduction of over 20ºC! :)

Cheers for the help.
 
As far as I'm aware, the caps are the little silver/orange/silver things to the side, the big bit is the VRM (or maybe inductor)... either way, fairly concerning that one of them had no thermal pad coverage!
Glad you got it sorted... I really need to check my temps and play with an undervolt!
(only monitoring core temp and using probably 100W more than I need to for the same performance... but I barely ever get around to gaming so it's fiiiinnee :D )

Edit: I thought from the first naked pic that it had burnt the die... guessing that was just a reflection from your lights!
 
Last edited:
As far as I'm aware, the caps are the little silver/orange/silver things to the side, the big bit is the VRM (or maybe inductor)... either way, fairly concerning that one of them had no thermal pad coverage!
Glad you got it sorted... I really need to check my temps and play with an undervolt!
(only monitoring core temp and using probably 100W more than I need to for the same performance... but I barely ever get around to gaming so it's fiiiinnee :D )

Edit: I thought from the first naked pic that it had burnt the die... guessing that was just a reflection from your lights!
Thanks for the clarification. Yeah pretty shocking that it was naked. I thought perhaps it had snapped off and was on the underside of the heatsink, but it was clear. To be running 2 years like that... glad I sorted it. In a way. I had some thermal pads and I binned them a few months ago. Doh! Yeah just reflection thankfully lol.
:p
 
Back
Top Bottom