1 hour ago, pgde said:

If you change the pads are there any warranty implications?

 

I recall EVGA does allow you to re-paste the GPU and change the pads. For example, if you decide to build a custom loop and fit a GPU block, or if you decide to fit the EVGA hybrid cooler kit, or the EVGA custom loop block. What they do say is that if you send it back for warranty then you should replace the original cooler with thermal pads of course.

Not sure about other manufacturers.

Of course, if deciding to do this you should always check that it doesn't void the warranty. 

Thanks Martin!

 

On 2/13/2022 at 9:10 AM, BobFS88 said:

Hopefully @Turboken can get the latest version of HW64info and re run his test and measure those VRAM temp junction to see his temps at idle and under load. 

Very interesting, I will try and test this tomorrow.

8 hours ago, TurboKen said:

Very interesting, I will try and test this tomorrow.

OK I managed to try it out. 

So the temp readings at the bottom in GPU-Z are the same as EVGA PX1. However, I noticed at the top there is another sensor called memory temperature and if you hover the mouse over it, it says "Junction Temperature". This reads 68 after running the same 1.5 hr flight as before. During idle it is 32c. Hope this helps.

 

Edited February 15, 20224 yr by TurboKen

10 hours ago, TurboKen said:

and if you hover the mouse over it, it says "Junction Temperature". This reads 68 after running the same 1.5 hr flight as before. During idle it is 32c. Hope this helps.

 

Well that's incredibly low for Memory Junction Temp. I've done lots of research and the only people seeing such low VRAM junction temp are running custom loops.

 I've seen nobody with the standard EVGA air cooler with junction temp as low as yours. Even people like "igorslab" who have tested multiple cards don't see junction temp that low.

Even people wo replace the thermal pads and fit heat sinks to the back plates plus fans blowing on the back plates don't see temp that low. 

For you to get 26 degrees less than everyone else seems weird.

And GPUz showed full load on the card?

Would you mind sharing a screenshot of GPUz sensor page? 

EDIT:  3080/3090 cards have DDR6X. The only thing I can think of is if some cards were fitted with older DDR6 which isn't as hot. 

How old is the card? Red lips or black on the end? I'm thinking newer cards may not be as badly affected. Maybe EVGA have addressed the issue with better thermal pads? Seems that's also speculated on the forums. 

Its said to be purely the heat generated by DDR6X and cooled only by a backplate. In fact the VRM dumps heat into the backplate too, exacerbating the high temps. Cant figure out how your card could be different to everyone else's. Are you blasting a fan at the backplate by any chance?  

If you google 3090 junction temp you will see the extent of the phenomenon and how your results are unusual.

 

 

 

Edited February 15, 20224 yr by martin-w

7 hours ago, TurboKen said:

This reads 68 after running the same 1.5 hr flight as before.

 

 

Power limit should be set to default in Precision X1. Have you kept your power limit low? 

Doesn't have to be a 1.5 hour flight by the way. Even being in the FS2020 menus pushes the VRAM Junction Temp to 90. 

Which version of GPUz? Mine is 2.43.0.

Edited February 15, 20224 yr by martin-w

On 2/13/2022 at 6:09 PM, pgde said:

If you change the pads are there any warranty implications?

 

Changing pads shouldn't be an issue. Its tricky though and the right size of pads is essential. Gelid pads seem to be a popular choice. Get it right and junction temp can drop significantly. Get it wrong and they can be worse. 

Personally, given that the issue seems to be related to the memory chips under the back plate, Id just replace those with Gelid Extreme pads. I'm thinking 2 mm or 2.5mm. Not sure which yet. Easy to get on Amazon.

From EVGA...

Installing third party cooling solutions does not void warranty on our products. Just be sure to keep the original cooling solution as it will have to be on the card if it is ever sent in for RMA. Any physical damage such as burn marks, water residue or damage, or any damage to the PCB will void ALL warranties.

 

https://www.evga.com/support/faq/FAQdetails.aspx?faqid=57720

Edited February 15, 20224 yr by martin-w

6 hours ago, martin-w said:

Power limit should be set to default in Precision X1. Have you kept your power limit low? 

Doesn't have to be a 1.5 hour flight by the way. Even being in the FS2020 menus pushes the VRAM Junction Temp to 90. 

Which version of GPUz? Mine is 2.43.0.

My power limit was set to 100% during these tests.

That is true the MSFS menu still is very demanding, this was supposed to be fixed but was only slightly improved. 

I'm using what ever the latest version is, I downloaded it last night.

I will try the menu and see what readings I get.

By the way my card is about 14 months old but has the black edge, I too remember the early red ones had problems. Also mine is the FW3 Ultra which has the massive heat sink

and I remember EVGA which states on the box how it is supposed to be a very good cooling solution for the RAM and VRM's, but of course that is marketing talking. I would rather know real world results.

EDIT: Yes I do have a fan blowing air across the top of the card, otherwise there was a lot of heat building up in the back where I have an M.2 drive and 4 ram sticks. The fan keeps all these components happy. Maybe I will test with this fan OFF to see how much it improves the temps.

Edited February 15, 20224 yr by TurboKen

22 minutes ago, TurboKen said:

what ever the latest version is, I downloaded it last night.

 

22 minutes ago, TurboKen said:

By the way my card is about 14 months old but has the black edge, I too remember the early red ones had problems. Also mine is the FW3 Ultra which has the massive heat sink

 

 

Okay, thanks for your help Mr Turbo. I've just been reading on Tom's hardware that many RTX 3080 and 3090's were coming out of the factory with misplaced, or faultily installed, thermal pads. Might explained why you are seeing better temps, although they do confirm that under 100 degrees is less likely to be misplaced pads.  The fact that your card is newer than mine might also be significant. Might well be that EVGA have improved the quality of the pads and the way they are installed.

Yes, we have the identical card, hence the significance of your lower temp.

 

https://www.tomshardware.com/news/gpu-thermal-pads-incorrectly-installed

Some graphics cards are coming with improperly installed thermal pads that cause overheating.

Edited February 15, 20224 yr by martin-w

Martin.  My red lip EVGA FTW 3090 also runs hot.  But I haven't noticed any performance related issues.  Just an annoying fan that I typically will not hear over my headset anyway.  I have a 3 year microcenter extended warranty.  So I guess I could bring it in if I want.  I'm just not sure it's worth it. Are you seeing issues?

40 minutes ago, micstatic said:

Martin.  My red lip EVGA FTW 3090 also runs hot.  But I haven't noticed any performance related issues.  Just an annoying fan that I typically will not hear over my headset anyway.  I have a 3 year microcenter extended warranty.  So I guess I could bring it in if I want.  I'm just not sure it's worth it. Are you seeing issues?

 

Yep. Bet if you check VRAM Junction Temp it's close to 90.

In other news, I have some Gelid Thermal pads on the way in various sizes. With great courage and daring... and stupidity I'm going to dive in and swap them.

I'll let you know if I see an improvement.

Just doing the VRAM under the backplate as that seems to be the issue.

EVGA and Nvidia may be happy as long as junction temp stays under 110, but then they would be, all they care about is the card lasting the warranty period.

I have a 5 year EVGA warranty, but its not voided by replacing the pads so worth a shot. My cards are passed on to others after the warranty is up so if I can extend its lifespan worth it.

The interesting point is ambient temp too. I'm in a chilly cottage in winter. So must assume that VRAM will be even hotter in the height of the summer. 

 

Edited February 15, 20224 yr by martin-w

45 minutes ago, micstatic said:

So I guess I could bring it in if I want.  I'm just not sure it's worth it. Are you seeing issues?

 

Nope, no throttling. DDR6X won't throttle till it gets above 110.

Degradation is controversial, with some interpreting the Micron literature to suggest 120. Some however are suggesting a degree of degradation as low as 100. Clearly its not an on off switch, its a sliding scale.

So you can see that my 95 doesn't leave much margin.

I doubt Microcentre would swap your card unless its hitting 110 and throttling.. EVGA told me my temp is warm but not an issue unless it hits 110.

DDR6X is so new that nobody really knows if 94 degrees while gaming will result in shortened lifespan or not. But I suspect that when the 40 series cards arrive they will have much better cooling for the VRAM on the rear of the card.

 

Curious to see your results. Keep us posted sir. 

I hope that cures it, looking forward to your findings once you get the backplate taken off.