Actually, I sent EVGA an email and they responded in like 10 minutes. Impressive as usual.

This was the response..

 

.

Quote

Hello Martin,

thank you for contacting our support team.

The temp of the VRAM is high but it is very safe as long as this temperature is reached when gaming/underload.
It should however not reach the 110°C.

If you have any further questions or concerns, please let us know.

Best regards,
EVGA EU

 

EVGA aren't bothered as long as it doesn't exceed 110 degrees. 

 

17 hours ago, martin-w said:

Not for 3090 memory temp. Mid 80's is a good temp for 3090 memory and safe. 3090 memory idles at 50 to 60. 80's isn't that high for GPU temp either really these days. 

I was referring to GPU over all temperature under load, although I agree the memory will be higher and if the temps are in the 80's would be fine. However in practice the GPU for over all performance and efficiency is still better to be lower so that with the high heat tolerances with these new GPU, I believe you have a more heat margine in the event that your won't spike up into TDP area as oppose to running hot to begin with with less margin. 

I maybe a little spoiled on how I would run my GPU that would average around 60 to 70c under load loaded conditions when running with the CPU at about 50c running Battlefield 4, for maybe an hours just for a test.

I manage to run the system much colder or under cooler conditions  starting out (40c to 50c on the GPU on idle) and (27c to 35c on the CPU at idle) it was most stable and not reaching TDP condition that sometime cause it crash at about 80c or better. My old system was 3930k OC to 4.6 ghz under load in adaptive mode state using a custom loop water cool with a fan cooled GTX Titan pascal. 

The  Ambient temp in the room about 75 degrees will allow for the CPU temps to be about 32c to 37c with hose connect to the back of the cabinet, extended to the window, I get CPU lower to about 27c and GPU about 45c depending on how cold it is. I cheated but it works and did it for years. 😁 Of Course I have to clean and filter out and thing coming in. The cabinet has a side fan with a filter so sometimes just raising the window was enough. The system would heat the whole room so I would close the vents. The system died from a leak from the water block that fell on the GPU and seep through the metal that took out the GPU and motherboard where it would not post vga. It was not seen until it was too late otherwise I would still be running it. I was due for an upgrade but was not force to do it this soon. However I am glad i am able to and it would be out of the way going forward.

 

36 minutes ago, BobFS88 said:

I was referring to GPU over all temperature under load

 

That wasn't my issue. GPU temp is around 60'ish. Its the VRAM temperature that's high. Nothing else.

 

36 minutes ago, BobFS88 said:

The  Ambient temp in the room about 75 degrees will allow for the CPU temps to be about 32c to 37c with hose connect to the back of the cabinet, extended to the window, I get CPU lower to about 27c and GPU about 45c depending on how cold it is.

 

That's idle not load. My systems idle at those temps too. That's normal. With an NH-D15s. Loads temps are the consideration not idle temps. 

 

Re the memory junction temp, I contacted EVGA and this was the response. They aren't bothered at all about 94 degrees under load. 

 

 

 

Edited February 9, 20224 yr by martin-w

On 2/8/2022 at 8:38 AM, martin-w said:

Thanks Jason. Useful info. 👍

Coincidentally with your reply... I just dived into Precision X1 and dropped the power target down to 80 with the temp linked. The result is that memory temp drops to 88 and lower fan noise. Much better.

I'll do some more experimentation tomorrow. I want to find out what sort of impact dropping the power target has on frame rate. I suspect that exactly what you did. 

 

Hi Martin, I talked about this in another thread also but I dropped my power level down to 90% which dropped the temps and fans speed a little. However I did this more just to not make it work too hard as I cannot just go buy another if it quits working. I don't have any numbers but I wouldn't expect to much. I will look next time I get a chance. I think the most I would see is temps in the 70s but I have a lot of fans in my case so it stays fairly cool.

23 hours ago, TurboKen said:

I think the most I would see is temps in the 70s but I have a lot of fans in my case so it stays fairly cool.

 

Nope, not what I'm referring to.

GPU temp will be in the 60 - 70 range. Its Memory Junction Temp we are talking about, not GPU temp. VRAM is hitting the 90's for most people with 3080 or 3090 cards, and 60 at idle. DDR6 is running hot.  

As I mentioned above, EVGA tell me its not a concern under load and that 110 is the temp they would be concerned about.

As for reducing power limit, I ran some tests last night and even at 80% power limit it still eventually crept up to 90. So only 4 degrees lower. I'm not worrying about it to be honest, after lots of research it seems most people are seeing such high temps with DDR6, it just runs hot, simple a s that. 

Now if someone has a custom loop and a double sided CPU block they might see lower temp. 

If you check yours, use Hardware Monitor, or GPUZ, not Precision X1. Precision X1 isn't reporting the temp right. Or rather, its from a different location. 

 

Quote

I cannot just go buy another if it quits working.

 

I went for EVGA's extended warranty, pretty cheap, so its nor warrantied for 5 years. 

Edited February 10, 20224 yr by martin-w

1 hour ago, martin-w said:

 

Nope, not what I'm referring to.

GPU temp will be in the 60 - 70 range. Its Memory Junction Temp we are talking about, not GPU temp. VRAM is hitting the 90's for most people with 3080 or 3090 cards, and 60 at idle. DDR6 is running hot.  

As I mentioned above, EVGA tell me its not a concern under load and that 110 is the temp they would be concerned about.

As for reducing power limit, I ran some tests last night and even at 80% power limit it still eventually crept up to 90. So only 4 degrees lower. I'm not worrying about it to be honest, after lots of research it seems most people are seeing such high temps with DDR6, it just runs hot, simple a s that. 

Now if someone has a custom loop and a double sided CPU block they might see lower temp. 

If you check yours, use Hardware Monitor, or GPUZ, not Precision X1. Precision X1 isn't reporting the temp right. Or rather, its from a different location. 

 

 

I went for EVGA's extended warranty, pretty cheap, so its nor warrantied for 5 years. 

Yeah I'm sorry I meant to mention I was referring to GPU temps, I still haven't checked the memory temps yet but I will. 

I too got the extended warranty, but I'm not sure even with that how fast it could be repaired/replaced given the current conditions. Hopefully I won't have to find out. I have never had an nvidia card fail yet.

15 hours ago, TurboKen said:

I too got the extended warranty, but I'm not sure even with that how fast it could be repaired/replaced given the current conditions. Hopefully I won't have to find out. I have never had an nvidia card fail yet.

 

EVGA do their cross ship thing, were you pay for a new card and send the old one back, that way you get your new card ASAP. They then of course refund you when they receive your defective card. I would imagine they have a supply of cards for RMA but yes, who knows with the current issues. 

OK Martin I did some testing for you. I'm running 3840x1600 @ 60FPS. Power level was @ 100% for these tests vs the 90% I usually have it. I did a flight from flightbeam PDX to SFO in the AS CRJ on both sims. About 1.5hr flight, here are the results read from EVGA PX1.

               Idle              P3D              MSFS

Mem1     27                57                 51

Mem2     27                56                 51

Mem3     28                58                 52

GPU       26                63                 58

So these results seem really good to me but I have (2) 140's and (2) 120's for intake plus an 80 inside blowing cool air across the top of the card, which also keeps the hot air from going to the 4 memory sticks. Then (6) 120's in a push pull on the top of the case and a 140 in the back. The luxury of having a big case I guess. Maybe you could add some more fans to try and get the temps down? 

 

17 minutes ago, TurboKen said:

OK Martin I did some testing for you. I'm running 3840x1600 @ 60FPS. Power level was @ 100% for these tests vs the 90% I usually have it. I did a flight from flightbeam PDX to SFO in the AS CRJ on both sims. About 1.5hr flight, here are the results read from EVGA PX1.

               Idle              P3D              MSFS

Mem1     27                57                 51

Mem2     27                56                 51

Mem3     28                58                 52

GPU       26                63                 58

So these results seem really good to me but I have (2) 140's and (2) 120's for intake plus an 80 inside blowing cool air across the top of the card, which also keeps the hot air from going to the 4 memory sticks. Then (6) 120's in a push pull on the top of the case and a 140 in the back. The luxury of having a big case I guess. Maybe you could add some more fans to try and get the temps down? 

 

Are am assuming you are you a liquid cool solution for your cpu, correct?

Just now, BobFS88 said:

Are am assuming you are you a liquid cool solution for your cpu, correct?

Yes, I'm using a H150i which is a (3) 120mm radiator. The temps shown above are all regarding the video card just to be clear.  The CPU was probably in the low 60's range.

That seem to be what I would expect for a setup like that with the temps. 
Thanks for sharing.

7 hours ago, TurboKen said:

OK Martin I did some testing for you

 

Thanks, but that looks like Tcase not T-junction.

Pretty much everybody is seeing high junction temp.

What did you measure it with?

Hardware info, or Hardware Monitor you need. Or keep it simple and use GPUz.

 

https://linustechtips.com/topic/1299399-hot-news-more-like-hwinfo64-now-shows-hot-vram-for-rtx-30803090/

 

"Summary

With a very recent update to HWInfo64, users are now able to see directly via the popular hardware monitoring tool the temperature of their GDDR6X memory modules for any RTX 3080/3090 cards which they own, and boy will they (likely) see a surprise waiting for them.  Even with DLSS enabled, maximum temps of GDDR6X memory modules can reach as high as 100 C while gaming.  And this is even worse if someone is attempting to use their RTX 3080/3090 for memory intensive workloads like cryptocurrency mining; the VRAM in those cases would shoot up to 110 C and severely down-clock itself to keep itself from being literally roasted to death, which makes sense as 110C is already (sort of) higher than the "95+C" operating temperature listed on Micron's official website for GDDR6X.  Igor's Lab (the first afaik to bring up this "hot VRAM" issue back in September of 2020) estimates that the Micron chips themselves will likely have to hit 120 C before sustaining immediate permanent damage, but (also afaik) the general consensus among anyone who knows at least a little bit about computer hardware is that triple-digit positive (or negative too in many cases afaik) Celsius temperatures are typically never good for the lifespan of the hardware itself."

Edited February 13, 20224 yr by martin-w

As you can see, some individuals have managed to reduce VRAM Junction Temp by changing the thermal pads. Determining the best thickness is tricky though, get it wrong and temps go up rather than down.

Some direct fans directly on the cards back plate too, which drops temp by a couple of degrees.

 

 

If you want to test your memory junction temp then you must use the latest version of GPUz, Hardware Info, or Hardware Monitor. Otherwise you won't see the right temperature reading. Precision X1 from EVGA doesn't not measure Junction Temp. 

 

This issue is one of the reason why EKWB are now selling double sided Graphics card blocks. 

 

Edited February 13, 20224 yr by martin-w

1 hour ago, martin-w said:

As you can see, some individuals have managed to reduce VRAM Junction Temp by changing the thermal pads. Determining the best thickness is tricky though, get it wrong and temps go up rather than down.

Some direct fans directly on the cards back plate too, which drops temp by a couple of degrees.

 

 

If you want to test your memory junction temp then you must use the latest version of GPUz, Hardware Info, or Hardware Monitor. Otherwise you won't see the right temperature reading. Precision X1 from EVGA doesn't not measure Junction Temp. 

 

This issue is one of the reason why EKWB are now selling double sided Graphics card blocks. 

 

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. 

If you change the pads are there any warranty implications?