• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Rx 6800 XT card hotspot temp is 102+ c

Joined
Nov 25, 2023
Messages
202 (0.33/day)
Processor 5900x
Motherboard MSI 570s
Cooling AIO 240
Memory 32 GB G.Skill @ 3200MHz
Video Card(s) 6800 xt
Storage Many
Display(s) Two QHD
Power Supply SilverStone Hela 1200W
Software Windows 11
Hi all , @torkema61

Just recently I found my Rx 6800 XT (Gigabyte gaming OC) fans are so loud and it reach 2500 ~ 2900 rpm. I cleaned everything but did not work also I placed a new thermal paste from noctua but the tmep did not want to get down. Should I use thermal pad on the GPU chip ? GPU temp is 40c idle and ~74 c load. I did not OC anything.

I also saw this thread and I have the same problem so this is common but what to do to fix ? in addition my QHD gaming monitor stopped display any picture so I dont know if the GPU heat is the problem but the monitor did not want to work on any other device please advice, has power but no picture.
 
Hi all , @torkema61

Just recently I found my Rx 6800 XT (Gigabyte gaming OC) fans are so loud and it reach 2500 ~ 2900 rpm. I cleaned everything but did not work also I placed a new thermal paste from noctua but the tmep did not want to get down. Should I use thermal pad on the GPU chip ? GPU temp is 40c idle and ~74 c load. I did not OC anything.

I also saw this thread and I have the same problem so this is common but what to do to fix ? in addition my QHD gaming monitor stopped display any picture so I dont know if the GPU heat is the problem but the monitor did not want to work on any other device please advice, has power but no picture.
If your GPU's warranty period has not expired and it has never been repaired, send it to the technical service under warranty and state that the temperatures are high and that you want it to be cleaned.

However, if the GPU's warranty has expired, then you can perform thermal repast. Replace the GPU's thermal paste with one that does not "pump-out". If possible, also replace the vram pads. Your card will run almost as cool as the original. Also clean the cooling block and fans thoroughly. Dust and some invisible obstacles (hair, etc.) can also reduce cooling performance.

Of course, I have no information about how efficient the original GPU is in terms of cooling.
 
Thermal pads work worse than paste, so don't. You could try undervolting the gpu to lower the temperatures.
Since the monitor doesn't work on any device, I would suspect the monitor itself or the cable. It doesn't make sense if it was due to a hot GPU.
 
I bought a Vega 64 from Gigabyte and had the exact same issues described. I took it apart and noticed that at least 20% of the GPU die had no paste. Here is the kicker. I bought a 6500XT Gaming Gigabyte card and it started doing the same thing. So I took the card apart and noticed that once again on the same part of the die there was no Thermal Paste. It is my reflection that you only buy Gigabyte cards if you know how to take GPUs apart but also at the same time have like the least amount of screws to remove. The good thing is on AMD they are cheaper than everything else but since As Rock has jumped in they offer better built and quieter cards than Gigabyte.
 
If your GPU's warranty period has not expired and it has never been repaired, send it to the technical service under warranty and state that the temperatures are high and that you want it to be cleaned.

However, if the GPU's warranty has expired, then you can perform thermal repast. Replace the GPU's thermal paste with one that does not "pump-out". If possible, also replace the vram pads. Your card will run almost as cool as the original. Also clean the cooling block and fans thoroughly. Dust and some invisible obstacles (hair, etc.) can also reduce cooling performance.

Of course, I have no information about how efficient the original GPU is in terms of cooling.
No warranty. What you mean "pump out" ?

Thermal pads work worse than paste, so don't. You could try undervolting the gpu to lower the temperatures.
Since the monitor doesn't work on any device, I would suspect the monitor itself or the cable. It doesn't make sense if it was due to a hot GPU.
I tried my best thermal paste but temp did not change and the gpu original thermals were pads but they were friable. For the monitor I bough different cables but monitor did not want to work, this happened with the gpu hot problem

Under volting is dangerous ?

I bought a Vega 64 from Gigabyte and had the exact same issues described. I took it apart and noticed that at least 20% of the GPU die had no paste. Here is the kicker. I bought a 6500XT Gaming Gigabyte card and it started doing the same thing. So I took the card apart and noticed that once again on the same part of the die there was no Thermal Paste. It is my reflection that you only buy Gigabyte cards if you know how to take GPUs apart but also at the same time have like the least amount of screws to remove. The good thing is on AMD they are cheaper than everything else but since As Rock has jumped in they offer better built and quieter cards than Gigabyte.
In my location I have very limited choices and very bad warranty. When I opened my 6800 card did not found any thermal paste just friable pads. I did not know there is a gpu called vega 64 what it used for ? graphics ?
 
Last edited:
How I know if certain thermal paste can "pump out" or not ?
By monitoring temperatures for about a week. If they get considerably worse day by day, it means your paste is pumping out.

With that said, 102 °C hotspot is not the end of the world. If it reaches 110, and your card's performance starts to degrade, then you have a problem.
 
there is not a lot that you can do.
RDNA (1,2 and 3) have extreme hotspot deltas even under water...
all of my GPUs incl. a 6900XT and 6950XT after a repaste sometimes hit 113°C and throttle.
it's something you have to live with except when you buy really high end models which are normally in the high 90s under heavy load.
 
all of my GPUs incl. a 6900XT and 6950XT after a repaste sometimes hit 113°C and throttle.
it's something you have to live with except when you buy really high end models which are normally in the high 90s under heavy load.
is that in the summertime with an gpu intensive game?
 
is that in the summertime with an gpu intensive game?
even in winter with 18c ambient. my 6950XT can stay at around 90-95°C for maybe 10 minutes until it starts to hit 110-113°C and throttles.
same for my 5700XT (when overclocked), 6800XT Aorus Master, 6900XT Speedster and 6900XT Nitro+. all are at least hitting 100°C on the hotspot after a while. the Nitro + and 6950XT hits 113°C even at 100% fanspeed with a 40°C delta after three repastes.
 
Thanks alot guys for fantastic answers I will try under volt and post the result
 
If you're comfortable opening up your card, alongside the suggestions in this thread, PTM7950 is great for hotspot delta reduction, in my experience.

Also no pump-out.
 
Something is clearly wrong with your card. I have a 6800 XT from XFX (319 Merc) and it doesn't have "the best" cooler. Still, after properly applying thermal paste (AC MX-6) and overclocking, I never go above 90 °C hotspot with approx. 2500-2600 RPM on the fans. I do have an open case tho, but my card is overclocked, +15% power limit and 300W mod with MPT, so I'm getting about 350W max power on the card.

If it's under warranty → RMA the card
if it is not under warranty → re-apply the thermal paste (quality one and put a lot on, be careful not to over-tighten the screws, because you will get pump-out)
P.s.: there's no such thing as too much thermal paste ;)
1701124281922.png
 
You could take it apart once more and this time be very careful to screw the heatsink back in as evenly as possible, go in a diagonal X pattern and in each corner you only twist the screwdriver by the same amount, over and over until completely tightened. If it's still just as bad as this I bet the heatsink just sucks and the cold plate isn't very even, once again gigabyte cards seem to be worse of them all.

Definitely undervolt, it will lower temperatures and increase performance.

I do have an open case tho
That makes a huge difference actually, like 10C or more.
 
You could take it apart once more and this time be very careful to screw the heatsink back in as evenly as possible, go in a diagonal X pattern and in each corner you only twist the screwdriver by the same amount, over and over until completely tightened. If it's still just as bad as this I bet the heatsink just sucks and the cold plate isn't very even, once again gigabyte cards seem to be worse of them all.

Definitely undervolt, it will lower temperatures and increase performance.


That makes a huge difference actually, like 10C or more.
I am aware of that. But if I put off the overclock and stick to stock clocks I'm down to 75 °C on the hotspot and 60 °C on the GPU so that's not it I believe.
 
nd overclocking, I never go above 90 °C hotspot with approx. 2500-2600 RPM on the fans
Choosing fan level is not working it keep raising despite what I choose.
 
Choosing fan level is not working it keep raising despite what I choose.
Do you get these high temperatures during gaming regularly, or do you run a benchamark like Time Spy, or Furmark? Furmark is a GPU killer, don't use it.
 
Do you get these high temperatures during gaming regularly, or do you run a benchamark like Time Spy, or Furmark? Furmark is a GPU killer, don't use it.
Time Spy is also quite a power sucking benchmark on RDNA2. I've seen 220W from my 6700 on Time Spy, while even DOOM Eternal only goes up to 185W (and most games do around 160)
 
Time Spy is also quite a power sucking benchmark on RDNA2. I've seen 220W from my 6700 on Time Spy, while even DOOM Eternal only goes up to 185W (and most games do around 160)
Holy sh*t, didn't know Time Spy was that abusive. One of my friends has a 6700XT, never seen it go above 200W, another one has a 6750XT and never saw it go above 220W, those are some crazy numbers for a 6700 :') I usually check my overclock with Superposition, and yes I know it's power hungry but it's very good at showing core clock stability.

yes during gaming. I dont like Time Spy, or Furmark for that reason
I'd very much like to keep the discussion on the forum, but if you want to, I will go out of my way and try to connect to your PC via Anydesk, TW or some program of your liking, try to help you out and if we fix it, OFC post the results here.
BUT first, put in your PC specs and maybe give us a picture or two of your rig so we know what we're working with here ;)
 
Holy sh*t, didn't know Time Spy was that abusive. One of my friends has a 6700XT, never seen it go above 200W, another one has a 6750XT and never saw it go above 220W, those are some crazy numbers for a 6700 :') I usually check my overclock with Superposition, and yes I know it's power hungry but it's very good at showing core clock stability.
If you make the power limit effectively unlocked using MPT, Time Spy will draw as much as it can, seemingly. I've seen some insane power draw from it.

220W seems about right for the power limit of a 6750XT.

Cannot screenshot it, so have a picture of it drawing 200W, which is the max power limit I set this time around.
The clock is still not reaching what it should (2660MHz effective, 2710 set on software), so it could draw even more
 

Attachments

  • 20231127_195409.jpg
    20231127_195409.jpg
    5 MB · Views: 265
Last edited:
If you make the power limit effectively unlocked using MPT, Time Spy will draw as much as it can, seemingly. I've seen some insane power draw from it.

220W seems about right for the power limit of a 6750XT.

Cannot screenshot it, so have a picture of it drawing 200W, which is the max power limit I set this time around.
The clock is still not reaching what it should (2660MHz effective, 2710 set on software), so it could draw even more
Using up to your power limit and killing your GPU with a highly unbalanced load are not the same thing. Time Spy is aggressive on power, but not dangerous with a properly configured GPU, or one running at default settings (unless it's defective). Furmark can be dangerous even with default settings.
 
Using up to your power limit and killing your GPU with a highly unbalanced load are not the same thing. Time Spy is aggressive on power, but not dangerous with a properly configured GPU, or one running at default settings (unless it's defective). Furmark can be dangerous even with default settings.
Oh yeah I said it was a power sucking benchmark, not that it was akin to Furmark.

More so "it will make your GPU hot way more than games will"

In another note, don't drivers "nerf" Furmark nowadays so its less nocive?
 
Back
Top