PC Graphics

[Tsuki6969]

Greetings, I bought my 6700 XT a few months ago and I totally love it. I recently got into overclocking and undervolting to which I seem to have found an excellent compromise but I'm a litte worried about the junction temperature. I currently reach 80C° max on the card and 96C° max junction temperature with a huge overclock+undervolt.

Are those safe temperatures? What temperature will damage my card or make it thermal throttle?

[elstaci · Accepted solution]

In GPU cards  5000 and 6000 Series models the Junction Temperature in a GPU is referred to as the GPU's  "Hot Spot".  The Maximum Operating Temperature for Hot Spot (Junction Temperature) is 110C.

Once Junction Temperature reaches 110C the GPU will automatically start to throttle or go slower to lower the Hot Spot temperature back to 110C or lower.

The GPU card has several Temperature sensors throughout the card measuring hot spots.

So when you are Overclocking or undervolting you can use 110C Maximum Operating Temperature as a reference.

[RPX100]

@elstaci can you talk about the following scenario, please?

GPU temps under gaming load: (RX 5700 XT, triple fan design, 95-99% load)

• GPU core temp = 55-65°C (68°C max)
• GPU mem temp = 55-65°C (67°C max)
• GPU VRM temp = 45-55°C (61°C max)
• GPU junction temp = well above 90°C even reaching 110, 111 and 112°C
• GPU fan constantly ramping up to 3600 RPM (100%)

card was manufactured in July 2020 and used since Dec 2020.
Temps have always been like that from the start.

Question is: where is the junction temp measured for RX 5700XT?
And does this look like incorrect cooler mounting / bad thermal paste?
I did not take it apart so far - but am thinking about it.

[elstaci]

@RPX100, It is hard to say because the GPU card temperatures can be affected by your Ambient Room temperature and air flow inside your Computer case or could be dirty Heatsink on the GPU card or fan issues.

But as I mentioned before the RX5700 GPU cards are designed to operate up to 110C before it starts to throttle or slow down. So if your Temps are slightly over 110C then the card will simply slow down to bring down the temperature unless it get so hot that the GPU card crashes or causes the computer to crash. Then you have a problem. Personally I would try to keeps Temps. under 110C under any conditions if possible.

Some other User might be able to give a better answer to your questions.

I would post at ASRock Forums and see what other Users with the same GPU card has to say about the temperatures.

Did find this ASRock Forum thread about the RX5700XT running hot: https://forum.asrock.com/forum_posts.asp?TID=14863&title=asrock-rx-5700-xt-challenger-temp-issues

Now Asus RX5700 series GPU card did have a defect from the factory that caused the GPU card to overheat but Asus came out with a fix later on.

I haven't heard of any issues of that sort with ASRock GPU cards though.

[RPX100]

small update regarding my recent post about my own RX 5700 XT's junction temp: @elstaci 

After looking into the topic on YouTube and learning that 90°C already causes throttling,
which I was able to confirm, I decided to follow up on my initial idea and removed the cooler:

• replaced thermal paste with Arctic MX-4
• replaced all thermal-pads with 11W/(m*K)
  
  

result:

• GPU does not throttle anymore and has stable boost clocks
• junction temp stays below 90°C (max 86°C so far under load)
• temp delta (GPU core to junction) is now max 15°C (average < 5°C)
• memory and VRM temps have improved (better thermal pads)
• GPU fans are now around 50%-60% instead of 100% (on auto)

Not sure if the cooler was mounted incorrect, but the paste was kinda dry and unevenly distributed.
This could of course look like it did, because the paste moved, while removing the cooler...

Anyway. Problem fixed. I am happy that I decided to replace the paste+pads.

[Tsuki6969]

So my overclock is fully safe as long as it doesn't reach 110C° junction temp? That's reassuring to hear! I love my card but I was afraid of damaging it with my OC.

Thank you

[elstaci]

This Tech site explains what AMD says about the 110C Hotspot Maximum Operating Tempertures: https://www.techpowerup.com/258228/110-c-hotspot-temps-expected-and-within-spec-amd-on-rx-5700-serie...

"On Hotspot, AMD writes: "Paired with this array of sensors is the ability to identify the 'hotspot' across the GPU die. Instead of setting a conservative, 'worst case' throttling temperature for the entire die, the Radeon RX 5700 series GPUs will continue to opportunistically and aggressively ramp clocks until any one of the many available sensors hits the 'hotspot' or 'Junction' temperature of 110 degrees Celsius. Operating at up to 110C Junction Temperature during typical gaming usage is expected and within spec. This enables the Radeon RX 5700 series GPUs to offer much higher performance and clocks out of the box, while maintaining acoustic and reliability targets."

Like I mentioned you can use the 110C Maximum Operating Temperature as a reference when you overclock the GPU. Like I mentioned the GPU card has several Thermal Sensors throughout the card. If any of the Thermal Sensors detects a temperature at or above 110C the GPU card will automatically start to throttle or slow down to lower the power usages thus lowering the temperature.

Now as to being "safe" it is difficult to answer because you can cause GPU failure or damage if your incorrectly Overclock or change voltages that is not related to temperatures.

I suggest you post at Overclockers.com. There other Users can help you correctly Overclock or change voltages: https://www.overclockers.com/