Question Palit RTX 3070 Ti temperature issue ?

[InSaneJeSter]

Hello yesterday after games and video editor, I ran into a problem with the temperature of the video card, in a strange way it is almost always at a hot spot of about 90 degrees and above.
Then I checked in different programs and tests the temperature does not correspond to the use or frequency of the video card.
Screenshots and video:
https://drive.google.com/drive/folders/1dUrQ5sOzifuekbJJMzsupC6Lb8WYLPR9?usp=share_link
I also tried to blow the dust of the video card, and the entire computer.
The settings in Nvidia are all standard, like Windows, Windows hardware acceleration is disabled.
That is, my fear is that over time the chip will become more degraded or the thermal sensor on the video card will simply fail.

My system specs:
Case: Aerocool Tor Pro
Case Fan: Aerocool Edge 14 PRO x7
Motherboard: Asus ROG Strix B550-E Gaming Wi-Fi [Bios: 2806 AM4 AGESA V2 PI 1.2.0.7.]
GPU: Palit GeForce RTX 3070 Ti GameRock Drv528.49 0.925v 1950mhz Mem 10236mhz
CPU: AMD Ryzen 7 3700X BOX 3,6/4,4GHz (3950GHz\1.175 Volt) AMD Wraith Prism.
Water cooling: Gamer Storm CASTLE 240 RGB V2
RAM: Kingston HyperX FURY Black 8x2 Hynix CJR.Rank one.3400 16-19-19-19-36
PSU: Toughpower GF1 ARGB 750W Gold - TT Premium Edition [PS-TPD-0750F3FAGx-1]

Drives:
M.2 Samsung 970 EVO Plus 500 GB [MZ-V7S500BW]
M.2 WD Black SN750 250 GB [WDS250G3X0C-00SJG0]
SSD WD Blue 1000 GB [WDS100T2B0A]
SSD WD Blue 500 GB [WDS500G2B0A]
HDD 1 TB Western Digital Blue [WD10EZEX]
HDD 2 TB Seagate BarraCuda [ST2000DM008]
HDD 4 TB WD Ultrastar DC HC310 [0B36040]
HDD 4 TB Western Digital Elements Desktop [WDBWLG0040HBK]

Monitor: AOC 27" 27G2U5/BK
Network Router: ROG Strix GS-AX3000
UPS: PowerCom Macan MAC-2000
OS: Windows 10 Pro 22H2 Licence

 

[Fix_that_Glitch]

What is at 90 and above?

 

[KyaraM]

What is at 90 and above?

He said what. The hotspot temperature reported by HWInfo. What part does it tell on? Simply the hottest on the GPU.

 

[InSaneJeSter]

What is at 90 and above?

It's strange, but I saw the difference after different testing, in Furmark, OCCT, there is no difference in the undervolt and its absence, everything is the same, but in games there are tests in Metro Exodus Enhanced Edition, Cyberpunk, Total War Warhammer 3 IE, Dying Light 2 , Red Dead Redemption 2 and Hitman 3 all with beams where possible, anyway, but the video card heats up about 71 degrees at 60-78 cooler revolutions.
In hotspot its 90-94c.

 

[Fix_that_Glitch]

60-70 are fine temperatures. 78 is getting a little warm. The hotspot temps are not as important, unless the memory is heating up to those temps. Can you increase the fan curve to be more aggressive at cooling when the temps start to climb above 70?

 

[Fix_that_Glitch]

He said what. The hotspot temperature reported by HWInfo. What part does it tell on? Simply the hottest on the GPU.

When you write AT a hotspot, instead of the hotspot IS then the question to clarify is an appropriate response. Being unhelpful is not.

 

[KyaraM]

It's strange, but I saw the difference after different testing, in Furmark, OCCT, there is no difference in the undervolt and its absence, everything is the same, but in games there are tests in Metro Exodus Enhanced Edition, Cyberpunk, Total War Warhammer 3 IE, Dying Light 2 , Red Dead Redemption 2 and Hitman 3 all with beams where possible, anyway, but the video card heats up about 71 degrees at 60-78 cooler revolutions.
In hotspot its 90-94c.

Hotspot is a little warm, but not critical. 71 is completely uncritical. As long as hotspot doesn't get to over 100 and the memory is below that mark, too, it should be fine. Personally I prefer if the memory doesn't get over 80, but 90, for example, should be still okay.

 

[InSaneJeSter]

60-70 are fine temperatures. 78 is getting a little warm. The hotspot temps are not as important, unless the memory is heating up to those temps. Can you increase the fan curve to be more aggressive at cooling when the temps start to climb above 70?

Yeah i change a little after some tests, and the temperature dropped by about 7-10 degrees.
https://drive.google.com/file/d/1UKF_FRorCZA7nJXPNiYzetcEvxN4cBWr/view?usp=share_link

Hotspot is a little warm, but not critical. 71 is completely uncritical. As long as hotspot doesn't get to over 100 and the memory is below that mark, too, it should be fine. Personally I prefer if the memory doesn't get over 80, but 90, for example, should be still okay.

Yes, unfortunately, as I understand it, the problem of the vendor itself is burning, the revision of the second version in the first generally took off the temperature for 100-110.
But it became strange to me that the temperature in tests and games differs, that's exactly what haunts me, I also couldn't change the whore for myself normally.
Let me explain that when buying, I immediately began to change the whore to a minimum, I wanted to be up to 900v and 1950, but it didn’t work out and errors were pouring even without activating the rays, thinking that the video card could not work, so I did 950v and boost the frequency to memory +850 everything worked stably, but after a couple of months, I decided to try lowering it a little bit and was able to reach 925v at 1950 MHz and +850 to the memory, but still I couldn’t get 900v, and now 8 days ago I started trying again to find suitable points in the curve and was able to Weird! Before that, there were deviations in the curve almost the same, but still I was able to get 1935 frequencies at 900v (1950 in boost) and +850 to the memory frequency, all tests and OSST were passed even with volume rays!
But I was alerted that the temperature in tests and games differs, and in tests both with and without undervolt, it does not differ at all.
I'm sorry if a lot of things are not clear, I just use google translator and can't communicate in English so well.
https://drive.google.com/file/d/1xoXGEgYyu9h6fzqU0F6El3FD1a_PKc6q/view?usp=share_link

 

[Phaaze88]

Fix_that_Glitch said:
[Can you increase the fan curve to be more aggressive at cooling when the temps start to climb above 70?]

Yeah i change a little after some tests, and the temperature dropped by about 7-10 degrees.

So, case airflow is lacking.
A ~20C gap between gpu core and hot spot isn't unusual at all. I noticed it's more common with Turing and up. Probably the work of the higher thermal density compared to older dies.

I'm not seeing anything particularly alarming here.




EDIT: Aerocool Edge 14 PRO
These fans are just showpieces, and nothing more.

 

[InSaneJeSter]

In general, everything is not as clear as it was, I made a little mistake with the conclusions and checked more tests.
As I found out that it's not the temperature, but something else, it's strange that I do an undervolt or not, but the temperature does not change, even lowering the power limit to 80% has no effect.
https://drive.google.com/drive/folders/1vncfCPKuyalGGs4NaQ1YCAM6ur3nITWN?usp=share_link

 

[Phaaze88]

If nothing changes regardless of the profile you've set, then it's most likely a physical limitation, like paste on the die or case airflow - maybe both.