One GPU fan is ignoring custom fan curve and revving loudly at around 73-75 degrees

[xenon20]

Hi All,

I have submitted a support ticket to EVGA, but I thought I'd ask here also if anyone has encountered similar issues in the past.

I have an EVGA RTX2070 XC Ultra GPU, I have recently been doing some stress testing on my PC and have noticed that when the GPU temp (as measured in HWInfo) reaches around 73-75C, one GPU fan starts revving up to max speed every 5-10 seconds. The second fan on the GPU is not doing it.

I've set a flat line of 50% fan duty across all temps of all fans in the BIOS (via ASRock Fan-Tastic Tuning), and then I use Rem0o's Fan Control software to create custom curves for CPU/GPU/and mix of these for case fans. As far as I understand, Fan control overrides the BIOS fan settings when windows boots up.

When I have been GPU testing with Furmark and monitoring Fan Control / HWInfo, The GPU temp is only reaching a point which calls for ~50% fan duty (On GPU fan 1&2), but the RPM of GPU fan 1 keeps Revving up (and down) to around 3200 RPM while the fan 2 remains at around 1500 RPM:

I've done some searching around and found a number of people on the net with similar issues, the most common thing I've heard is to check and reapply thermal paste to the GPU's chip. Also this VIDEO suggests that although the GPU temp as reported by HWInfo is ok (eg 73-75C), the VRAM temp may be overheating and the GPU's switching into "panic mode" to prevent damage etc (although based on RTX3XXX).

As there are lots of people with similar issues across different cards (mostly RTX by the sounds of it), I'm wondering if it's an inherrent issue and and RMA for a replacement unit would display the same behaviour? I can always start taking the GPU apart to check / reapply paste etc, but a bit wary about doing so before knowing this is the problem.

Any experience with the same issue for anyone?

Thanks,
Paul

 

[Disco_P]

It might be that Fan control bugging out but more likely is that 106C hotspot temperature is putting the card in panic mode.

 

[Alt0153]

repaste needed stat!

 

[FlyingFish]

With it being an EVGA card, have you tried EVGA's own program to control the fans?
I think it's called Precision X but it's been a few years since I last had a EVGA GPU.

 

[z10m]

It might be that Fan control bugging out but more likely is that 106C hotspot temperature is putting the card in panic mode.

That.

 

[xenon20]

Thanks guys,

With it being an EVGA card, have you tried EVGA's own program to control the fans?

Yes I've tried disabling fan control and trying Precision X1 with a custom curve too, but the same behaviour happens.

I could do without the GPU going into panic mode when it only reaches 73C core temp as it will do that regularly.

When we talk about the hotspot temp, is that likely to refer to the VRAM temp that was suggested in the video? Somewhere it said that HWInfo gave GPU memory temp readback but I dont see it in the list?

I saw some video's that suggested to replace the thermal pads inside the GPU and repaste the GPU. Is that the best option to follow here? If the suggestion of RMA is made, would a new unit likely have the same issue?

 

[Disco_P]

Some cards don't have a VRAM temperature sensor, yours probably doesn't.

Hot spot is usually the hottest point on the GPU die. Due to the massive difference between core and hot spot, your paste is probably not making good contact.

You can either repaste or RMA if still under warranty. Note that EVGA have pulled out of the GPU market but have kept stock of some cards for RMAs.

 

[muon]

How old is the card?

You can see at 106C the card is throttling down and maxing out the fan.

 

[mburn_83]

had this on a 2080 - had to repaste and replace the memory heat pads.