[SOLVED] 3060Ti FE major thermal issues? Is this terrible paste/pads from nvidia?

Status
Not open for further replies.

TheRandomGuy7

Distinguished
Oct 16, 2012
95
0
18,640
Brand new 3060Ti arrived today and my quest for a GPU is finally over. But my excitement has been a bit ruined...

Under a stress test (Kombustor and others) the sensors immediately read 80c and the fans instantly ramp up to 100%. 'Hot spot' reads around 105c.

At 100% fan speed the temperatures never even drop below 78c or so (this is under full load but still). I've never seen cooling on a GPU this ineffective before.

A case of FE cards being bad at cooling? Well I doubt it, since every single review raves about how the 3060Ti FE runs extremely quiet and cool.

Example: View: https://i.imgur.com/UMzDnoq.png
-techpowerup review
View: https://i.imgur.com/x94sdzE.png
-my card under heavy load

My card is running at MORE THAN DOUBLE the rpm, is almost 10c hotter, and is probably 40db louder.

Any ideas or similar experiences?

EDIT: Undervolt your card people!
 
Last edited:
Solution
Stress tests are not realistic results of the average use your card will see. These will usually make the card run louder and hotter than normal.

You can't compare an "at home" test with a controlled environment most reviewers use. They control room temperature, speed on all fans in the case (front, back, top, CPU, GPU, etc.)

Play a game for an hour or so with your preferred temperature/performance monitor. This should give you a more realistic result of what your card will see on a daily basis.

Udyr

Honorable
Mar 3, 2021
254
106
9,690
Stress tests are not realistic results of the average use your card will see. These will usually make the card run louder and hotter than normal.

You can't compare an "at home" test with a controlled environment most reviewers use. They control room temperature, speed on all fans in the case (front, back, top, CPU, GPU, etc.)

Play a game for an hour or so with your preferred temperature/performance monitor. This should give you a more realistic result of what your card will see on a daily basis.
 
Solution

TheRandomGuy7

Distinguished
Oct 16, 2012
95
0
18,640
Stress tests are not realistic results of the average use your card will see. These will usually make the card run louder and hotter than normal.

You can't compare an "at home" test with a controlled environment most reviewers use. They control room temperature, speed on all fans in the case (front, back, top, CPU, GPU, etc.)

Play a game for an hour or so with your preferred temperature/performance monitor. This should give you a more realistic result of what your card will see on a daily basis.
What case are you using.

First post is the correct answer...

To be honest these FE cards are masterpieces.. unboxing it was like discovering an alien relic, the packaging and card itself are both so elegantly designed. But because of this, on top of the gleaming reviews, I was expecting nothing short of absolute perfection in terms of performance and cooling.

Very first thing I did (as I would testing a GPU normally) was run the stress tests and they made me panic and post here as I've never seen a card get this hot so fast before.

But you're right, after actually gaming on it the performance is fine, and while still fairly hot at 95c on the memory junctions, the cooler holds up just fine in normal usage and fans seem to stick between 60-80% instead of 100. Also testing out NiceHash and getting 62c and 72c on core and hotspot respectively.

Problem is, even though I've used maybe 15 different (older) GPUs in this case before, I didn't realise they've all been high-end Strix and MSI and Sapphire models which blow away any reference card, even the 30 series which is maybe the best yet.

The case is a Fractal R5 with huge empty space (drive bays removed) and a good positive airflow setup with 3 140mm noctuas. In addition to a NH-D15 pumping even more air towards the exhaust.

So basically I rushed into posting this as I was a bit angry and concerned. But for any FE owners who might be in the same boat, I'll leave my experience up for you to see! I still suspect there is a non-optimal job on the paste/pads/mounting but nothing as major as I feared.
 

Zerk2012

Titan
Ambassador
First post is the correct answer...

To be honest these FE cards are masterpieces.. unboxing it was like discovering an alien relic, the packaging and card itself are both so elegantly designed. But because of this, on top of the gleaming reviews, I was expecting nothing short of absolute perfection in terms of performance and cooling.

Very first thing I did (as I would testing a GPU normally) was run the stress tests and they made me panic and post here as I've never seen a card get this hot so fast before.

But you're right, after actually gaming on it the performance is fine, and while still fairly hot at 95c on the memory junctions, the cooler holds up just fine in normal usage and fans seem to stick between 60-80% instead of 100. Also testing out NiceHash and getting 62c and 72c on core and hotspot respectively.

Problem is, even though I've used maybe 15 different (older) GPUs in this case before, I didn't realise they've all been high-end Strix and MSI and Sapphire models which blow away any reference card, even the 30 series which is maybe the best yet.

The case is a Fractal R5 with huge empty space (drive bays removed) and a good positive airflow setup with 3 140mm noctuas. In addition to a NH-D15 pumping even more air towards the exhaust.

So basically I rushed into posting this as I was a bit angry and concerned. But for any FE owners who might be in the same boat, I'll leave my experience up for you to see! I still suspect there is a non-optimal job on the paste/pads/mounting but nothing as major as I feared.
Stress test can be fairly close to gaming depending on the game and monitor resolution.
The FD 5 does have bad airflow.
 

Zerk2012

Titan
Ambassador
That's why I was concerned tbh. Performing far worse than other cards I've tested in the same R5 case.
Example Cyberpunk pushes my card to the max..
The FD 5 case just has small vents on the side to let air through.

Your core temp is fine I would use MSI afterburner to set a custom fab curve to kick the fans up to 100% to help cool the rest. You have to check the box to make it active when the PC boots.

The memory on those cards runs a bit hot anyhow no just the FE editions.
 

TheRandomGuy7

Distinguished
Oct 16, 2012
95
0
18,640
Example Cyberpunk pushes my card to the max..
The FD 5 case just has small vents on the side to let air through.

Your core temp is fine I would use MSI afterburner to set a custom fab curve to kick the fans up to 100% to help cool the rest. You have to check the box to make it active when the PC boots.

The memory on those cards runs a bit hot anyhow no just the FE editions.

I don't think the R5 has inherently worse airflow than other cases, with all the sound dampening engaged sure. If I open the front door panel the temps are 1-3c better though. Possibly if I also remove the noise padding from the top of the case then this thing should have pretty excellent airflow?

Anyway done a lot more tweaking now, done a solid undervolt and realised this is how the card should run from factory. Undervolted and +1000 on memory gives almost 5% performance increase and very surprisingly a whopping 10c cooler...

I definitely couldn't cope with 100% fan for any period of time. But thankfully 60% is more than enough now.
 
Status
Not open for further replies.