Very high 3080 VRAM temps, anyway to fix it?

Associate
Joined
2 Oct 2020
Posts
23
I recently managed to get a 3080 FE for my new pc, however I have noticed very high temperatures for the GDDR6X VRAM with it reaching up to 106 degrees Celsius, I primarily use it for gaming but I want to mine on it while im not playing games to make some money back on the purchase however I am concerned such high temps will damage the card in the long run so wanted to check if anyone had a practical solution short of watercooling it?
I have already attempted undervolting without the desired results, it only ends up reducing the core temp however that is not much of an issue as that tends to never go over 65-70 anyway.
 
Soldato
Joined
29 May 2005
Posts
4,899
take the card apart and put some premium thermal pads on. and set higher fan speed and ensure nice cold air is flowing over the GPU fans.

thats all you can do. these VRAM are meant to get pretty hot. even under water i imagine they will get hot as well
 
Associate
OP
Joined
2 Oct 2020
Posts
23
take the card apart and put some premium thermal pads on. and set higher fan speed and ensure nice cold air is flowing over the GPU fans.

thats all you can do. these VRAM are meant to get pretty hot. even under water i imagine they will get hot as well
that seems to be the general consensus i found online after a bit of research, will thermal grizzly minus pad 8s be good enough or should I try and find some gp ultimate or something similar?
 
Soldato
Joined
29 May 2005
Posts
4,899
that seems to be the general consensus i found online after a bit of research, will thermal grizzly minus pad 8s be good enough or should I try and find some gp ultimate or something similar?
Those pads seems to be the prevailing choice. But you void your warranty by doing that. So weigh that up before you go down that path.
 
Soldato
Joined
28 Aug 2017
Posts
2,779
Location
United Kingdom
take the card apart and put some premium thermal pads on. and set higher fan speed and ensure nice cold air is flowing over the GPU fans.

thats all you can do. these VRAM are meant to get pretty hot. even under water i imagine they will get hot as well

i can confirm under water on my 3090 fe my core sits around 48-55 in games at 2070mhz and my ram runs at 72-75 with a oc applied @10750mhz, backplate on my bitspower block is quite warm when the card gets going, all via hwinfo64 latest version that added memory temp for gpu's :).

plus mining for any kind of duration will affect the card's performance by anything upto 10%, meaning the card can be 10% slower due to mining, i really wouldn't think of doing that as your loosing vital frames in games
 
Soldato
Joined
29 May 2005
Posts
4,899
i can confirm under water on my 3090 fe my core sits around 48-55 in games at 2070mhz and my ram runs at 72-75 with a oc applied @10750mhz, backplate on my bitspower block is quite warm when the card gets going, all via hwinfo64 latest version that added memory temp for gpu's :).

plus mining for any kind of duration will affect the card's performance by anything upto 10%, meaning the card can be 10% slower due to mining, i really wouldn't think of doing that as your loosing vital frames in games
Interesting to know if you mined on that card what vram temp it would get to.
 
Soldato
Joined
28 Aug 2017
Posts
2,779
Location
United Kingdom
Soldato
Joined
29 May 2005
Posts
4,899
oh dont worry i wont attempt it, i found a article online when 2080ti were mined on and afterwards the article said the cards dropped performance in games by 10%, not gonna gimp my card

here is the article

https://www.tomshardware.com/uk/news/mining-card-rtx-2080-ti-tested-gaming
Lol, I think performance of card drops over time due to degradation anyways. There is no metric to say if something should drop by x amount.

More over that article said the cards were pretty old and had a lot of accumulated dust etc etc as it was run in open air rig. So I would think that would cause pretty hefty performance hinderance. More over no one in the world has actually done any testing on GPU performance at day 0 to day 365. So who is it to say cards don’t drop performance overtime anyway. The tested cards didn’t get their benchmarks on day 0. They were benched against a database. So not really indicative of what the card can do or not do. In mining GPU core is under volted and down clocked . So GPU core usually run very cool. So it is strange that article is saying those cards are limited by GPU core clocks. That part of the card is least likely to be going off tbh.

there are people here with vega cards that have been mining for years and still got the same performance in terms of hash rate. So a single non-normalised test doesn’t mean much tbh.

any how. I was just asking as a quick and dirty test that’s all.
 
Last edited:
Associate
OP
Joined
2 Oct 2020
Posts
23
i can confirm under water on my 3090 fe my core sits around 48-55 in games at 2070mhz and my ram runs at 72-75 with a oc applied @10750mhz, backplate on my bitspower block is quite warm when the card gets going, all via hwinfo64 latest version that added memory temp for gpu's :).

plus mining for any kind of duration will affect the card's performance by anything upto 10%, meaning the card can be 10% slower due to mining, i really wouldn't think of doing that as your loosing vital frames in games
from what ive seen i think the thermal limit is 110 degrees so i wouldnt be surprised if it got to that(if u tried mining) so it might be worth not mining on it. 3090 VRAM cooling is supposed to be better though
 
Soldato
Joined
28 Aug 2017
Posts
2,779
Location
United Kingdom
Lol, I think performance of card drops over time due to degradation anyways. There is no metric to say if something should drop by x amount.

Maldonado that article said the cards were pretty old and had a lot of accumulated dust etc etc as it was run in open air rig. So I would think that would cause pretty hefty performance hinderance. More over no one in the world has actually done any testing on GPU performance at day 0 to day 365. So who is it to say cards don’t drop performance overtime anyway.

there are people here with vega cards that have been mining for years and still got the same performance in terms of hash rate. So a single non-normalised test doesn’t mean much tbh.

any how. I was just asking as a quick and dirty test that’s all.

fair point, for me i couldnt justify it, was lucky to get a card so i'm doing all i can to keep it running at top performance, cleaning etc :)
 
Associate
OP
Joined
2 Oct 2020
Posts
23
Those pads seems to be the prevailing choice. But you void your warranty by doing that. So weigh that up before you go down that path.
oh? does removing the shroud and backplate remove the warranty? i couldnt actually find much information online regarding founders edition cards
 
Soldato
Joined
28 Aug 2017
Posts
2,779
Location
United Kingdom
oh? does removing the shroud and backplate remove the warranty? i couldnt actually find much information online regarding founders edition cards

as long as you dont damage anything (backplate, screws or thermal pads) the warranty is good but as pc-guy said any mods you do that are found out will void the warranty 100%, you do it at your own risk/expense
 
Soldato
Joined
29 May 2005
Posts
4,899
That’s why there is a big big price increase for GPUs come out of the factory with water blocks and back plates that have manufacturer warranty.

for all the custom water cooling guys ripping the shiny RTX 30 HSF off to put on EK blocks and saving a few hundred quid they are basically doing it with the knowledge of things go wrong with the card, they will likely have to take that loss themselves. Thus the original HSF is kept in near vacuum state in case of the fateful day. But if you get water damage on it, you are out of luck even if you didn’t touch the card.
 
Associate
OP
Joined
2 Oct 2020
Posts
23
That’s why there is a big big price increase for GPUs come out of the factory with water blocks and back plates that have manufacturer warranty.

for all the custom water cooling guys ripping the shiny RTX 30 HSF off to put on EK blocks and saving a few hundred quid they are basically doing it with the knowledge of things go wrong with the card, they will likely have to take that loss themselves. Thus the original HSF is kept in near vacuum state in case of the fateful day. But if you get water damage on it, you are out of luck even if you didn’t touch the card.
that makes sense
 
Back
Top Bottom