GPU Hot Spot Temperature 105 Celsius, fans at 3000 RPM, while GPU Temperature is max 70 Celsius

[Mitohondri]

Hi, I have a rather puzzling problem that I cant resolve for 2 weeks now.

All was good, when all of the sudden while playing WoW my GPU fans started spinning at 100% on my 7900 XT gpu
HWinfo showing 65 C GPU temp, shile hot spot temp was above 100 C

checked all psu cables connections, GPU PCIE slot, moved SSD from 1 slot to another, pulled memory and put it back in

Computer is all fine, when its used for anything except for gaming, its even ok in new Cinebench GPU bench, but now while actual gaming

My rig where the problem first came up:

Asus TUF B760M DDR4
i5 12400f
LE720 360mm water cooling
Corsair Vengeance DDR4 3200MHz 2x8GB
Kingston SX2000 1TB NVMe
SF750 PSU
Lian Li O11D Mini
PowerColor RX 7900 XT

I did this:

BIOS update, updated all drivers for everything, mbo, gpu, peripherals, monitor etc, same problem
put in my 3070 card and repeated all of the above, same problem
put in different memory, same problem
put in different psu, same problem
reinstall win 11, repeated all of the above same problem
reinstall to win 10, repeated all of the above same problem

Sold CPU, water cooling, case, ssd and 7900 XT - went to shop, and bought new MBO, CPU, MEMORY, COOLER, SSD and Case:

My rig now:

Asrock B760M-itx
i5 12400f
NH-L9i
Fury DDR4 3200MHz 2x8GB
Seagate Firecuda 512GB NVMe
SF750 PSU
Lian Li Dan A4-H20
Gainward RTX 3070

Plugged it in - same problem

BIOS update, updated all drivers for everything, mbo, gpu, peripherals, monitor etc, same problem
put in different memory, same problem
reinstall win 11, repeated all of the above same problem
reinstall to win 10, repeated all of the above same problem
changed my psu cable
plugged it in a different outlet

computer passes all the stress test like, AIDA, OCCT, Cinebench, CPU-Z, y-cruncher etc
It ramps up the fans to 100% and hot spot temperature is 105 C, only in gaming and in OCCT 3D stress test, while its ok in cinebench GPU test


I am all out of ideas , and i was not able to find solution, or similar problem, so this is my last resort

is it possible that my PSU does something to my CPU, and then it send some instructions to GPU that makes the GPU do some work that is maxing hot spot but not "normal" gpu temperature

PSU is the only variable that I didnt change, i did try with different PSU, but only after the problem was there, but now when i have all the news stuff, and the problem is the same with the original SF750 PSU its the only thing i suspect

Help if you can guys

 

[mama]

Sounds like a cooling issue. Blockage or water pump maybe?
Edit: sorry, just saw the air cooler on the second setup.

 

[Frick]

I do believe I read some stuff about the Powercolor cards specfically had a tendenecy to be poorly assembled, . Check the card for loose screws.

 

[Vya Domus]

I don't understand you had this problem before as well with a different GPU ? So both the 3070 and 7900XT were hitting the same temperatures ? That seems really unlikely.

 

[freeagent]

I don't understand you had this problem before as well with a different GPU ? So both the 3070 and 7900XT were hitting the same temperatures ? That seems really unlikely.

If that is the case, then it has to be his lack of case flow. If he does have good case flow, then silence and high performance do not go hand in hand.

 

[Vayra86]

I don't understand you had this problem before as well with a different GPU ? So both the 3070 and 7900XT were hitting the same temperatures ? That seems really unlikely.

Yeah I had a lengthy post regarding that, I thought he had those 7900XT temps in two systems, and that would have been plausible, seeing as its the same card. Deleted my post

 

[droopyRO]

Lian Li Dan A4-H20

Did you try with the side panels removed, are you still getting above 100ºC hot spot ?

 

[Nike_486DX]

You already applied liquid metal and torqued the mounting screws to spec, right?

 

[Mitohondri]

i have a feeling you didn't read or understand the post guys
what liquid metal, what torqued, what case flow, what side panels , that is not the issue

the problem is not with the case flow, or the GPU temperature, since that is not getting above 70C, with both 7900 XT and 3070, the problem is the GPU HOT SPOT TEMP, that is getting 105+ C
two different GPUs, 2 different MBOs, two different memory, two different CPUs, two different SSDs, two different OS, in all possible combinations, two different case, only variable that is not changed is PSU


Lian Li O11D Mini, and A4-H2O have insane good airflow and i used in one case 7xA12 and 2xA14 Noctuas, and in other 2xA12 Noctua fans

HOT SPOT temp 105+ C and fans at 100% happen when i play games or run heavy GPU test

so my question was, since i changed everything two fold, could it be PSU doing something or CPU doing something, that is making HOT SPOT that hot?

and like i said al was fine, until one day it was not fine

 

[Vya Domus]

could it be PSU doing something or CPU doing something

No, if you get the same temps on two different GPUs from two different manufacturers the only possible explanation is that the case cooling is really bad.

Remove the case panel like suggested and see what happens. The only other plausible explanation is that you have a crypto miner or something running.

 

[Chomiq]

Gpu temps won't be solved be ditching your entire setup and swapping to even smaller case. Stop using your wallet.

 

[evernessince]

At any point have you tried breadboarding the system? It's not something you noted as having done. To do it you simply take your motherboard, CPU, 1 stick of RAM, and in your case the GPU and hook it up to the PSU outside the case. You then only connect what input perifs you need (which should just be a mouse / keyboard). Then run whatever application that was causing high hotspot temps and see if the issue persists. It's possible you got two GPUs with high hotspot temps but improbable. I have never seen PSU selection impact something as specific as GPU hot spot temp but even if we assume it could, it would have to be extra-ordinarily rare and yet you are exhibiting it across two different GPUs. Probabilities tells us that it's far more likely something else. On the list of potential issues, the PSU causing high hotspot temps is below that of metal whiskering.

 

[Mitohondri]

to repeat again.
Lian Li O11D Mini with 7(seven) 120mm Noctua fans, 2(two) 140mm Noctua, of which 3 120mm fans are below GPU as intakes, 3 120mm fans at the top as exhaust, 2 140mm fans on the side as intake, and 1 120mm at the rear as exhaust was the case i used for 6 months before this issue.

Never ever was CPU temp above 60C , GPU temp above 70C, GPU Hot Spot temp above 80C, GPU memory junctions temperature above 90C, while gaming at 4k resolution in WoW, Diablo 4, Cyberpunk, Apex etc.

and to miner issue, as I wrote i formatted 2 disks, and installed win 11 and win 10 on both of them 2 times clean

it is not case issue, airflow or any similar thing since the problem is the same on the table, when the system is not in the case

yes i have done that also, as I wrote, peripherals, memory stick, 2 motherboards, 2 gpus, 2 cpus, 4 memory sticks, 2 ssds, 2 operating system 2 different power socket

i would not come here if i didn't do every single thing i could come up with

gpus work perfectly fine in other systems that i have which is in Meshlicious with worst cooling than this two are

my question was, is it possible that the SF750 did something to CPU, that makes the cpu send instructions/signals to GPU that makes the GPU do heat up the hot spot, or something

thing is, when the problem popped up i did try with with different PSU, it was the same thing, so i thought that its not psu

but if the original PSU did the damage, then the other psu would would not matter

and when i bought new components i put he SF750 again, i didnt not try first with different PSU, because i didnt think the problem is with PSU

 

[Vya Domus]

Then you are really unlucky and got two GPUs that just have bad temps. My reference 7900XT would hit 100C when overclocked.

 

[Mitohondri]

i wrote above that both GPU work just fine in, since i tried them both in different system, and there is no problem with any of them

 

[Dragam1337]

Repaste gpu, and make sure the cooler is equally tightened on each corner.

 

[Mitohondri]

Repaste gpu, and make sure the cooler is equally tightened on each corner.

please read what I wrote, there is nothing wrong with any of the gpu

 

[taka]

My 3070 never pass the +10~15 Celsius hot spot from the core temp. are you sure the 12V line is ok?
What about the Vcore?

About the 7900 not sure wht to say dont have one...

 

[Mitohondri]

ok, good idea

here is my screen shot.

do you see anything that is off, or anything that can indicate what the problem is?

only thing that i notice is different from yours, is you have full green line, while mine is grey/green, is that something that shows that my PSU is faulty?

 

[Dragam1337]

please read what I wrote, there is nothing wrong with any of the gpu

And have you tried repasting the gpu, since you can say that with 100% certainty ?

When there is a huge delta between gpu temp and hotspot, it is essentially always a bad mount or bad TIM that is the issue...

But for all i care you can just do whatever you want... buy a new psu, that will surely fix your gpu hotspot temp... lol

 

[Mitohondri]

And have you tried repasting the gpu, since you can say that with 100% certainty ?

When there is a huge delta between gpu temp and hotspot, it is essentially always a bad mount or bad TIM that is the issue...

But for all i care you can just do whatever you want... buy a new psu, that will surely fix your gpu hotspot temp... lol

i can say that with 100% certainty, since both gpus where tested in my other computer, and they both work fine, and don't have problem with hot spot or anything else in same testing conditions, in same software and games

i know that i wrote a lot of text and it can be boring to read through all of it, but please try, just to fully grasp the issue i am having

thank you for understanding

 

[droopyRO]

Were those cards new or ? It looks like a bad mount or thermal paste application. Do they still have those "warranty stickers". Do you have warranty for them ? If not i would repaste them or take them to someone that can do that for you, if you have not done that before. As a general rule the difference between GPU temp and hot spot should be around 15ºC with a good mount.
PS:
Until then you can undervolt them an lower the temperature and fans rpm.

 

[taka]

ok, good idea

here is my screen shot.

do you see anything that is off, or anything that can indicate what the problem is?

only thing that i notice is different from yours, is you have full green line, while mine is grey/green, is that something that shows that my PSU is faulty?

is your cards bent or sagged, all looks good on monitoring side

 

[droopyRO]

all looks good on monitoring side

No, a 36ºC delta between the GPU temp and GPU hotspot is not good.

 

[meatwar]

A 7900XT packed in a "Glass box" really nice... that's a non-breathing case for such hw set...
And definitely to much delta values as previous droopyRO mentioned...