GPU Temp: 68°C, Hotspot Temp: 100°C (Sapphire Pulse RX 6700XT)

[Macro Device]

Temperatures below 110C on your hotspot are basically safe. Below 100C is the comfort zone. Below 90C = overclocking party invitation.
As far as you're within 110C and your junction-to-hotspot delta ain't all around the place (70C junction and 95C hotspot is sane; 70C junction and 110C hotpot is off) you don't need to worry about it much. Your GPU will die... about a dozen years from now. And what can I say, 12 y.o. GPU rarely do deliver playable FPS in recent titles.

 

[67Elco]

My RX 6800 XT records hot spot at 74 c in Starfield...all default settings except the fan curve via Afterburner.

 

[torkema61]

Temperatures below 110C on your hotspot are basically safe. Below 100C is the comfort zone. Below 90C = overclocking party invitation.
As far as you're within 110C and your junction-to-hotspot delta ain't all around the place (70C junction and 95C hotspot is sane; 70C junction and 110C hotpot is off) you don't need to worry about it much. Your GPU will die... about a dozen years from now. And what can I say, 12 y.o. GPU rarely do deliver playable FPS in recent titles.

Of course, I don't expect the GPU to work for life. If I can use a GPU for 5 years, that is enough time for me. Before switching to AMD GPU, the hotspot temperature of the GPUs I used (GTX 1070, RTX 2060 Super, RTX 3060Ti) was 75°C at most. But when I switched to the RX 6700XT, I was disappointed when I saw 100°C. When I sent it to technical service due to a heating problem, they sent me a brand new card. Of course, since the fan curve in the Sapphire Pulse 6000 series was problematic, I managed to keep the GPU temperature between 68-72°C and the hotspot temperature between 78-83°C by manually adjusting the fan.

I undervolted, set the MHz value to 2581 and the mV value to 1150, and now the GPU temperature is between 55-63°C and the hotspot temperature is between 74-78°C.

I'm happy with the GPU now. However, I feel like I will have problems with this issue again in the future.

 

[Zach_01]

Of course, I don't expect the GPU to work for life. If I can use a GPU for 5 years, that is enough time for me. Before switching to AMD GPU, the hotspot temperature of the GPUs I used (GTX 1070, RTX 2060 Super, RTX 3060Ti) was 75°C at most. But when I switched to the RX 6700XT, I was disappointed when I saw 100°C. When I sent it to technical service due to a heating problem, they sent me a brand new card. Of course, since the fan curve in the Sapphire Pulse 6000 series was problematic, I managed to keep the GPU temperature between 68-72°C and the hotspot temperature between 78-83°C by manually adjusting the fan.

I undervolted, set the MHz value to 2581 and the mV value to 1150, and now the GPU temperature is between 55-63°C and the hotspot temperature is between 74-78°C.

I'm happy with the GPU now. However, I feel like I will have problems with this issue again in the future.

You cant really compare hotspot temps between AMD/nVidia. Apples and oranges...
Both brands reveal exactly what they want users to see. So you cant be sure whats really going on inside the GPU die.
I think AMD has been more honest about it since Zen2/RDNA1 first came out in 2019.

Of course there is the cooler/contact implementation subject by vendors.
I think most of them are just seeking ok-ish cooling performance because of the 110C operating limit.

 

[torkema61]

Of course there is the cooler/contact implementation subject by vendors.
I think most of them are just seeking ok-ish cooling performance because of the 110C operating limit.

Absolutely I agree.

 

[ChrisM255]

Of course there is the cooler/contact implementation subject by vendors.
I think most of them are just seeking ok-ish cooling performance because of the 110C operating limit.

It is absolutely a die/coldplate contact issue.
I found a way to bring my 6700 XT to what it was supposed to be. Once I solved my contact issue it basically unlocked tons of OC and UV potential. The before and after was insanely different. I also don't cringe anymore when I peak at HWinfo64. Seeing 100-110°C and/or 15-30°C delta was sooo unsettling... undervolting was OK, but still was seeing high deltas at times and poor freq behavior, poor OC and UV performance. The stable ranges are WAY larger, I can hit much higher freq and lower voltages now. Even at stock settings after the fix I was seeing a performance increase due to higher and sustained clocks.
You will void the warranty though... personally, if I can't fix it myself I replace it. To each his own. Personally don't like RMA-ing, never had a good experience. I also work in the Tech industry and take electronics/computers/servers apart to service/refurbish regularly so this was right up my ally.

RX 6000 high Delta and temps fixed!

First, I have been desperately trying to find a post like this for MONTHS. I am only trying to help the gaming community. Not looking for 15 minutes of fame, an overclocking competition or ways to improve my overclock. Just want to help people with this GPU temp/delta issue that I was suffering...

www.techpowerup.com

 

[torkema61]

Unfortunately, after 8 months, the GPU's hotspot temperature rose again. On the first day of the GPU, the normal temperature was 70 and the hotspot temperature was 81. Of course, I used it with undervolt since the first day and while the normal temperature was around 62, the hotspot was in the range of 80-85. When I tried to test without undervolt last night, the normal temperature was in the 62-65 range, but the hotspot was close to 90 degrees again.

I think this will be an inevitable end for me.

 

[#22]

Unfortunately, after 8 months, the GPU's hotspot temperature rose again. On the first day of the GPU, the normal temperature was 70 and the hotspot temperature was 81. Of course, I used it with undervolt since the first day and while the normal temperature was around 62, the hotspot was in the range of 80-85. When I tried to test without undervolt last night, the normal temperature was in the 62-65 range, but the hotspot was close to 90 degrees again.

I think this will be an inevitable end for me.

Keep in mind that hotspot temps temps and size of a difference tend to rise with growth of ambient temp. Maybe your case.

 

[torkema61]

Keep in mind that hotspot temps temps and size of a difference tend to rise with growth of ambient temp. Maybe your case.

I have a mesh case. There are 2 Akasa Viper AK-FN063 fans on the front of the case as intake fans. There are 2 Arctic F120 exhaust fans on the top and 1 Arctic P120 exhaust fan at the back. My cafe is this;

The air temperature outside is in the range of 18-20 degrees.

 

[#22]

I have a mesh case. There are 2 Akasa Viper AK-FN063 fans on the front of the case as intake fans. There are 2 Arctic F120 exhaust fans on the top and 1 Arctic P120 exhaust fan at the back. My cafe is this;

The air temperature outside is in the range of 18-20 degrees.

It doesn't matter what cooling config you have - you won't fight much temperature growth from higher ambient with airflow. Then it just cools your components with warmer air

Ofc there's always option that thing responsible here is paste degradation, but if your ambient change recently, I would first pay attention to it when doing measurements.