MSI GeForce RTX 3090 Ventus 3X Hot Spot temperature issue

[Hendrikx]

So I was one of the few lucky ones to get my card in the early October last year. It's been amazing to be able to drive a samsung odyssey g9 and play games on a high refreshrate/high resolution monitor.

However since I bought a racing wheel and pedals and started play various racing games, one of which is DiRT Rally 2, the game started to stutter after a while, which has never happened in any game before. First thing I though was: "temperatures", so I loaded up HWMonitor to check the GPU temperatures.
 

That's when I noticed that my "Hot Spot" temperature reached beyond 100 degrees, up towards 106 degrees (celsius).

One of my thought was that since I use my 240mm radiator as intake, warm air reached the inside of the case and the GPU didnt have the possibility to cool down. However after some more research only the Hot Spot accually went extremely high, "GPU temp" and "Memory" still stays around 65-75 degrees while gaming.

This led me to install another air cooler for the CPU instead of the AIO I previously had, to see if triple intake fans with "cool" air (since ambient temps in the room is about 21 degrees), would help lower the Hot Spot temps. GPU temp and Memory temp went down, however the "Hot Spot" temp remained the same.

I've read  that thermalpads could be bad and wrongly sized on some of these GPU:s, and replacing them could benefit in lowering the "Hot Spot" temperatures.

Right now, as I am afraid of damaging the card, I'm forced to run "Power Maximum: 70%" in the Geforce Experience settings, as that will atleast keep the Hot Spot barely under 100 degrees.
However this is not a solution, as I'd really want to benefit of the full power this card is able to give.



General temperatures:

Temps after boot and some browsing:

Temps after runnings Path of Exile for 1 minute.

So my question here is, what to do?
Thanks in advance

[Falkentyne]

4 hours ago, Hendrikx said:

So I was one of the few lucky ones to get my card in the early October last year. It's been amazing to be able to drive a samsung odyssey g9 and play games on a high refreshrate/high resolution monitor.

However since I bought a racing wheel and pedals and started play various racing games, one of which is DiRT Rally 2, the game started to stutter after a while, which has never happened in any game before. First thing I though was: "temperatures", so I loaded up HWMonitor to check the GPU temperatures.
 

That's when I noticed that my "Hot Spot" temperature reached beyond 100 degrees, up towards 106 degrees (celsius).

One of my thought was that since I use my 240mm radiator as intake, warm air reached the inside of the case and the GPU didnt have the possibility to cool down. However after some more research only the Hot Spot accually went extremely high, "GPU temp" and "Memory" still stays around 65-75 degrees while gaming.

This led me to install another air cooler for the CPU instead of the AIO I previously had, to see if triple intake fans with "cool" air (since ambient temps in the room is about 21 degrees), would help lower the Hot Spot temps. GPU temp and Memory temp went down, however the "Hot Spot" temp remained the same.

I've read  that thermalpads could be bad and wrongly sized on some of these GPU:s, and replacing them could benefit in lowering the "Hot Spot" temperatures.

Right now, as I am afraid of damaging the card, I'm forced to run "Power Maximum: 70%" in the Geforce Experience settings, as that will atleast keep the Hot Spot barely under 100 degrees.
However this is not a solution, as I'd really want to benefit of the full power this card is able to give.



General temperatures:

Temps after boot and some browsing:

Temps after runnings Path of Exile for 1 minute.

So my question here is, what to do?
Thanks in advance

 

Please use HWinfo to post your stats, not HWmonitor.  HWmonitor is completely awful.  HWinfo also has expanded fields that you can click to show many other readings in the sensor areas..

 

Looks like one of the VRM's are not being cooled properly.

Hotspot is the hottest of multiple scattered sensors across the GPU Core and VRM's.

Bad paste or terrible mounting pressure on the core would cause both the average core temp and hotspot to rise.

You have a low average core temp but bad hotspot temp, so there is a problem with possibly one of the thermal pads covering the VRMs.

 

Unfortunately, you will have to disassemble the card (as well as have replacement thermal paste available) to determine the problem.

 

Take a very clear, high res picture of both the card PCB and the heatsink side and the thermal pads, and maybe one of us can find out what's going on.

No I do not know the thickness required for replacement pads.  I believe a few people have posted measurements on the Nvidia reddit, but I'm not sure.  I've only directly saved measurements for 3090 FE, 3090 Strix and 3090 Palit / Gainward cards.  eVGA people have their own measurements on the eVGA forums.  Gigabyte and MSI I am not sure about but I did see discussions.  I know one card (I believe a MSI model) had dimensions posted before though.

 

You have a Ventus.  I did find Suprim X measurements though.  I also don't know about the Trio either.

 

When you do replace the pads, make sure you use Gelid Extreme (NOT Thermalright, nor Gelid Ultimate).  You need compressible soft pads.  Laird (Tflex) 90000 pads also work very well.

 

 

https://imgur.com/a/kc84a0R

 

https://imgur.com/a/ozHghWw

 

 

 

https://www.digikey.com/catalog/en/partgroup/tflex-hd90000-series/71539

 

Anyway I DO NOT suggest using Themalright Odyssey.  You can see some people had bad contact.  Either buy Gelid Extreme or buy the Laird pads.  

It seems a combination of 1.5mm, 2mm and 3mm is needed.  For 2.5mm you can stack a 2mm and a 0.5mm OR maybe you can manually compress a 3mm (if you can find one in Gelid Extreme or the Tflex 90000).

[Falkentyne]

Found this also.

 

 

Just remember the thermal pads I told you to use.

 

[Hendrikx]

So I decided to open up the GPU and replace the thermal pads.

I used Gelid Extreme 3mm and Thermalright Oddysey 2mm.
I didnt take any pictures as it accually before I replaced anything looked quite OK, no pads were missing and it looked to have good pressure.

 

I replaced all the memorypads on the backplate side with 3mm (which I squeezed some for a good fit) and on the die side I used 2mm pads.

I replaced thermal paste with Nocuta NT-H1, there were alot of paste used prior which took forever to wipe clean, a mess.

(However my GPU temps were ok before replacement)

This took about 45 minutes to complete, and the results looks promising:
The temps before you can see in my screenshots above, which is with "70% Power Maximum" for the card.

Now I can run "100% Power Maximum" with alot better temps: