Question RX 6600 Junction Temp

[hejawi]

Computer Type: Desktop

GPU: RX 6600 XFX Speedstr SWFT 210

CPU: RYZEN 5 3600 6 CORE 12 THREADS

Motherboard: Asrock B450m HDV R4.0

BIOS Version: 4.9

RAM: 16GB Viper steel 3000Mhz (8GbX2)

PSU: Seasonic S12III

Case: Montech X1 (3 fan in, 1 fan Out) open side

Operating System & Version: WINDOWS 10 PRO latest update

GPU Drivers: Tested on 22.6.1 and 22.5.1

Chipset Drivers: AMD B450 CHIPSET DRIVERS latest one dated on 27th of June


Description of Original Problem:
So I am now using XFX speedstr 210 RX 6600 for 2 months, the temp was perfectly fine as the GPU temp is 60 - 65 degree and Junction 85 - 90 degree without any overclocking, only power draw of 120 watt

After installing a new CPU cooler (Thermaltake toughair 310) the Junction temp changed and increased and reached even 105 in heavy computing, I have noticed this during my game play session there was some freeze using 22.6.1 which the freeze disappeared after reinstalling 22.5.2 but the temperature still high weirdly (GPU temp is 62 - 65, Junction temp 95 - 105)

I don't know what exactly happened, and is it normal?

Can the new CPU cooler affect the GPU temp or the temp caused from the super weather temp increase as it is 33 degree out there?

 

[Cyberat_88]

Define "Junction" for me.
If your temps went up with new cooler, it's not a good one is it ? Any AMD CPU of that value should have Liquid Cooling.
Don't be scared, they don't leak inside the case, breakdown cases are extremely rare, especially out of the box.
Running mine for 5yrs. no slow down in cooling, Corsair before any other brand.
Also lower CPU fan 100% function to 60C or below in Bios.

 

[hejawi]

Define "Junction" for me.
If your temps went up with new cooler, it's not a good one is it ? Any AMD CPU of that value should have Liquid Cooling.
Don't be scared, they don't leak inside the case, breakdown cases are extremely rare, especially out of the box.
Running mine for 5yrs. no slow down in cooling, Corsair before any other brand.
Also lower CPU fan 100% function to 60C or below in Bios.

Thanks mate for your reply, the issue is not with my CPU, it is with my GPU
The Hot spot temp or Junction temp increased from 85 - 90 to 95 - 105 and the only thing changed is the CPU cooler, that is why I can't understand what is happening with my GPU, I tried to troubleshoot it, did many thing including different Bios version, chipset version, switched to quite bios instead of performace bios on GPU, swapping the GPU Driver, the temp still high

 

[Phaaze88]

Junction temp as in memory junction, or the hottest reading off the gpu's die? It relates to what app you used to read that value.

I think in either scenario: Did you install that cooler with the gpu still inside, and bumped it by mistake?

 

[Cyberat_88]

You mean the PCI-E slot temp ? You have sensors for that ? Where is that report from ?

 

[Cyberat_88]

"Quiet" is bad and inducing Heat. Wait, you flashed the GPU Bios ?

 

[Cyberat_88]

Also, a hotter CPU socket makes everything around hotter, 105C is quite extreme.

 

[hejawi]

Junction temp as in memory junction, or the hottest reading off the gpu's die? It relates to what app you used to read that value.

I think in either scenario: Did you install that cooler with the gpu still inside, and bumped it by mistake?

No the hottest read, used HWinfo64, hot spot reading or in AMD Software it is Junction temp
And now it was not installed since the cooler kinda big so I had to take it out

 

[hejawi]

"Quiet" is bad and inducing Heat. Wait, you flashed the GPU Bios ?

No the same GPU version has two Bios option with a hardware switch on the same card

 

[hejawi]

Also, a hotter CPU socket makes everything around hotter, 105C is quite extreme.

The CPU socket after the new cooler never runs more than 75 degree, but the GPU hottest spot reading reach 105 degree

 

[Cyberat_88]

Where are you getting the 105C reading from what program ?

 

[hejawi]

Where are you getting the 105C reading from what program ?

From HWinfo64 under GPU hot spot Temperature and from AMD Adrenaline GPU Junction Temperature

 

[Phaaze88]

No the hottest read, used HWinfo64, hot spot reading or in AMD Software it is Junction temp

Gotcha.
I would conclude that the gaps filled in between the gpu die and heatsink's cold plate was disturbed when you took it out... or when you put it in. Might have applied a little too much grip on it when you were doing either.
When there's a large gap between gpu die and hot spot temperature like that, it's typically caused by bad cooler mount(you didn't take the cooler off), or the paste 'filling' has broken.
You'll have to repaste it. Got some spare MX-4 lying around or something?

 

[Cyberat_88]

Are you sure you're not reading 105F from the software ? What does Bios say ?

 

[hejawi]

Gotcha.
I would conclude that the gaps filled in between the gpu die and heatsink's cold plate was disturbed when you took it out... or when you put it in. Might have applied a little too much grip on it when you were doing either.
When there's a large gap between gpu die and hot spot temperature like that, it's typically caused by bad cooler mount(you didn't take the cooler off), or the paste 'filling' has broken.
You'll have to repaste it. Got some spare MX-4 lying around or something?

Well, I didn't touch the GPU cooler or done anything to it,
But wouldn't that also affect the whole temperature of the GPU it never exceed 65 at full load, the problem only with the hottest read was 85-90 degree and increased not less than 10 degree
I have some thermaltake paste used when I installed CPU cooler, but wouldn't that remove the warrenty, XFX offered me RMA but it will cost me 150$ for shipping and tax that is why I am trying to troubleshoot before doing that

 

[hejawi]

Are you sure you're not reading 105F from the software ? What does Bios say ?

Dear,
There is no reading for GPU temp in the Bios,
This reading for GPU hottest spot not the CPU

 

[Cyberat_88]

That is what I am trying to lock onto, I think your "Junction" reading is Bogus.
But Celsius vs. Fahrenheit does matter.
Yes, it should melt the connectors to MB and raise temp of your GPU, once, before dying.

 

[Phaaze88]

There are multiple sensors on the gpu core. Hot spot or junction comes from which ever is the hottest one, just like cpu package is the reading from the hottest cpu core.
The gpu temperature is taken from an average.

I have some thermaltake paste used when I installed CPU cooler, but wouldn't that remove the warrenty, XFX offered me RMA but it will cost me 150$ for shipping and tax that is why I am trying to troubleshoot before doing that

Aye, I suppose they do have that obligation - some even have that 'warranty void if removed' stickers in some places - depending on where you're from.
They can't get away with that everywhere, as some countries have laws against that: you could still do a repaste, and use the warranty for another issue later.
Other than that, this problem can't be fixed without opening the card up.

 

[Cyberat_88]

Final advice:

1. Go with the better CPU cooler (lower temps).
2. Find the most compatible GPU driver for your games (not the latest).
3. Ignore bogus readings from unverified software.
4. Ignore glitches that only happen once every 6 months or more.
5. Buy a UPS power protector.

 

[hejawi]

There are multiple sensors on the gpu core. Hot spot or junction comes from which ever is the hottest one, just like cpu package is the reading from the hottest cpu core.
The gpu temperature is taken from an average.



Aye, I suppose they do have that obligation - some even have that 'warranty void if removed' stickers in some places - depending on where you're from.
They can't get away with that everywhere, as some countries have laws against that: you could still do a repaste, and use the warranty for another issue later.
Other than that, this problem can't be fixed without opening the card up.

So it will be a risky move in order to solve the issue, I will try it on my own or at least give to a good technical in my country to check it out

 

[hejawi]

Final advice:

1. Go with the better CPU cooler (lower temps).
2. Find the most compatible GPU driver for your games (not the latest).
3. Ignore bogus readings from unverified software.
4. Ignore glitches that only happen once every 6 months or more.
5. Buy a UPS power protector.

I don't any issue with the CPU anymore, it is now with the GPU,
I am using the previous one which was stable for me but the problem still persist
The reading from AMD Adrenaline which is the official software for AMD GPU
It is been 4 days on this problem

 

[Cyberat_88]

I happen to run 21.5.2 and I don't see any "Junction" readings.

 

[hejawi]

I happen to run 21.5.2 and I don't see any "Junction" readings.

I happen to run 21.5.2 and I don't see any "Junction" readings.

What's your GPU, this option was added to GPU rx 5000 series and newer

 

[Cyberat_88]

Mine is much older and according to AMD, last driver made for my OS, not the most stable. My system is in the signature below.

 

[Cyberat_88]

AMD says Junction temp is good to 110C, I'm not so comfortable with such temps. But do you overclock/underclock the GPU ?