PC Graphics

[FoxStrike]

Hi everyone, i hope you are well. I came here to see if you could help me with my graphics card. I have a rx 5500xt and usually playing the temperatures rise up to 70 degrees celcius and the junction temperature goes up to 110 degrees, with the fan at 100% of its speed being really loudy. The idle temperatures are 50 degrees.  It is normal to reach these temperatures? how can i reduce them?. thanks to whoever reads this and can help me.

pd: English is not my native language and i hope you can understand me.

System Configuration
Motherboard: Asus a320m-k
CPU: AMD Ryzen 3 pro 3200g
PSU: EVGA 500W 80 plus white
System Memory: 16GB Dual Channel 3000 MHz
GPU (VBIOS): Gigabyte rx 5500xt 4gb
OS (Version): Windows 10 x64 (Version: 19043.1348)
Driver: Radeon Software Adrenalin 2020 Edition 21.10.2
Applications: When im playing any game
Background Apps: Steam
Radeon Software Settings: Instant Replay OF, Anti-Lag OF, Radeon Image Sharpening OF, Enhance Sync OF

[elstaci]

At 110C the GPU card is probably throttling to keep the temperature at 110c or lower.

Do you have good  air flow inside your computer case?  Remove the computer case side panel and see if the GPU temperature lowers any. If it does that indicates you have poor air flow inside your computer case.

How old is your GPU card?  Make & Model please?

Try stress testing the GPU card and see if it passes or how hot the GPU card gets during the testing. Check the all temperatures, CPU and GPU, and PSU outputs during the test.

Another possibility could the Thermal paste or pads are not working as well as when it was new.

[FoxStrike]

The airflow its good, my cpu its around 50-60 degrees when i am playing. Disassemble all the case panels and continue to dial that same temperature when I play. The "actual" temperature its fine, around 70 degrees, but the one that worries me is the "junction temperature" and and that the gpu fan has to be at 100%

[elstaci]

The Junction Temperature is call the "Hot Spot" of the GPU Card. The Maximum Operating Temperature for the Hot Spot on the GPU is 110C.

There are several Thermal "Hot Spot" sensor on the GPU card. 

Whenever one of those Sensor is reaching or reaches 110C, the GPU card starts to throttle or slow down to keep the Hot spot temperatures at 110C or below.

The GPU card is designed to run at those very high temperatures but it really isn't a good thing. The cooler the GPU card runs the longer it will last.

[RPX100]

This is a temperature delta of at least 40°C (GPU core to hot spot),
and from personal experience (RX 5700 XT) I would point my finger towards:

tried up and/or not equally applied thermal paste on the GPU chip.

---
Please note: if you want to apply new thermal paste, then you also need new thermal pads.
Because the old ones will most likely not survive the dismantling of the cooler.
-> You also need to make sure that the new thermal pads have the correct thickness

Watch some repasting videos on YouTube to see if you want to try it yourself.
Otherwise you should contact a PC-repair shop for this task.

[FoxStrike]

What is the thickness of the thermal pads that you recommend for this card? and also what thermal paste you recomend?

[RPX100]

I can not recommend the correct thermal pad thickness, because:

- thermal pad thickness can be different for each graphic card
-> this could even be different for different RX 5000 series cards from the same manufacturer

- they also can be different for each part (memory / VRM / caps)
-> pads on memory are usually 1.5mm or 2mm thick
-> but it is really important to actually get the correct ones
-> if you fail to match the correct thickness, you will end up with too much/not enough pressure
-> and then you either have bad thermal paste contact + air bubbles or memory is not cooled at all

example: my RX 5700 XT from ASRock uses 3 different types of pads, all with different thickness.
-> which is annoying when you have to pay for large cut-out pads and then have leftovers
-> also note that you have to cut these pads to size by yourself - they do not come pre-cut

---
There are different ways to approach this problem:
1) have a look at your graphics card from the side and check if you can see the pads
-> the important ones are between cooler and board (PCB)
-> you should be able to see pads on memory, VRM and caps
-> this will however just give you basic knowledge about the 3 different pad types used
-> you would still have to measure them to make sure 
-> if you have the time: dismantle your card, measure pads, order them online and then rebuilt when they arrive

2) find a YouTube video with your exact make and model (look for teardown or repaste videos)
-> if the cooler design matches other cards from the same manufacturer, you could search their videos too

3) send a mail to the manufacturer support and ask them which pads have been used
-> success rate depends on the knowledge and information available to support 
-> some manufacturers are happy to provide the info, others will just tell you that you would void your warranty

---
The actual process of dismantling + cleaning + reapply + rebuilt took me around 90 mins.
Cutting the pads to the correct size and cleaning everything can take some time.
It is important to be careful and do not apply too much force when tightening the screws.

Good luck if you decide to try this. Make sure to watch some of these repaste videos.
The videos can provide some general overview before you try it on your own card.