This is a weird problem I've been having recently. GPU is an EVGA XC ULTRA 2070 SUPER, purchased new November 2019 (so it's out of warranty as of a month ago.) This is either a new issue or I am way less observant than I thought, because I only noticed it recently.
Randomly when the GPU is under load (around 75C), the RPM of fan 1 will spike to around 3400RPM. This is independent and irrelevant of whatever fan curve I have set. Outside of these spikes, fan 1 obeys the fan curve at all times, and fan 2 never has any problems (note: this changed while writing this post, see below.) These spikes happen but the actual power going to the GPU as '% speed' does not change. If I use Fan Control to set the fans to 100%, they cap out at around 3000RPM, so whatever is happening is making fan 1 spin faster than it's meant to be able to.
There is no consistent relationship between temperature and these spikes - when gaming, the GPU is pretty consistently at around 75C, and the spikes appear to be entirely random. The spikes are roughly double the 'intended' RPM - at those temperatures with my current fan curve, fan 2 is at around 1700-1900 RPM. There are little tiny 'blips' in fan 2's speed at the same time as the fan 1 spikes, but they are imperceptible without looking at a graph, very different from the very loud spikes on fan 1.
If I use Fan Control to manually set fan speed to, say, 50%, it doesn't matter - the fan spikes somehow override however much power is going to the fan (or at least however much power the fan controller says is going to the fan.)
Here is a screenshot of MSI Afterburner, showing the spikes and how the fan speed % and fan 2 do not change, and how it's seemingly independent of GPU temperature. The rise early in the graph is when I launched a game, so you can see there are no spikes when there is no load/high temperature on the GPU.
NOTE: As I was writing this, something new happened. For the first time, both fans spiked really high without the fan speed going up. This is the first and only time this has happened, until now it has only ever been fan 1. See below:
Here is my fan curve in Fan Control - both GPU fans are set to it. The curves in MSI Afterburner and Precision X1 are disabled, but they are identical - I made them match just so there couldn't be funny business there.
Latest GPU drivers, clean reinstall. Windows 10 Pro with all updates. Card is only and has only ever been used for gaming and some very light CAD, no mining or AI or anything. I've never done anything but some slight overclocking, no extreme OCs, no repasting, etc. I am going to repaste it soon, though, since in investigating this I found that I have a hotspot temperature of 105C, which is a little spicy. That might be related, though I have no idea how to test that without repasting and seeing the results.
I found one other person on Reddit who has the same issue with a GIGABYTE Vision 3070, and their friend has the same issue with a different PC/card (not specified), so it's not just an EVGA or 2070 thing.
My hunch is that the fan controller on fan 1 (and maybe fan 2 now?) is crapping out - I hope not, since as I said earlier the card is out of warranty, but I don't know what else it could be. As far as I can tell it doesn't affect the card's performance, it's just incredibly loud and can't be good for the bearing.
Has anyone seen anything like this before, and might know what's going on?
EDIT: Also maybe related maybe not, my GPU is limiting itself to 75C. Afterburner and Precision X1 are both set to limit it to 83C. GPU-Z reports that the limit is 83C. But whenever it hits 75C, the "Temp Limit" in Afterburner goes to 1, and I've never seen it hit 76C except for maybe half a second. Maybe the fans are spinning like crazy because the card is thermal throttling? How do I set the GPU temperature limit if Afterburner/Precision X1 aren't doing it?
Randomly when the GPU is under load (around 75C), the RPM of fan 1 will spike to around 3400RPM. This is independent and irrelevant of whatever fan curve I have set. Outside of these spikes, fan 1 obeys the fan curve at all times, and fan 2 never has any problems (note: this changed while writing this post, see below.) These spikes happen but the actual power going to the GPU as '% speed' does not change. If I use Fan Control to set the fans to 100%, they cap out at around 3000RPM, so whatever is happening is making fan 1 spin faster than it's meant to be able to.
There is no consistent relationship between temperature and these spikes - when gaming, the GPU is pretty consistently at around 75C, and the spikes appear to be entirely random. The spikes are roughly double the 'intended' RPM - at those temperatures with my current fan curve, fan 2 is at around 1700-1900 RPM. There are little tiny 'blips' in fan 2's speed at the same time as the fan 1 spikes, but they are imperceptible without looking at a graph, very different from the very loud spikes on fan 1.
If I use Fan Control to manually set fan speed to, say, 50%, it doesn't matter - the fan spikes somehow override however much power is going to the fan (or at least however much power the fan controller says is going to the fan.)
Here is a screenshot of MSI Afterburner, showing the spikes and how the fan speed % and fan 2 do not change, and how it's seemingly independent of GPU temperature. The rise early in the graph is when I launched a game, so you can see there are no spikes when there is no load/high temperature on the GPU.
NOTE: As I was writing this, something new happened. For the first time, both fans spiked really high without the fan speed going up. This is the first and only time this has happened, until now it has only ever been fan 1. See below:
Here is my fan curve in Fan Control - both GPU fans are set to it. The curves in MSI Afterburner and Precision X1 are disabled, but they are identical - I made them match just so there couldn't be funny business there.
I'm sure this is an unoptimized curve and is no good but it's just something I made to learn why things are loud, not the point of the post.
Latest GPU drivers, clean reinstall. Windows 10 Pro with all updates. Card is only and has only ever been used for gaming and some very light CAD, no mining or AI or anything. I've never done anything but some slight overclocking, no extreme OCs, no repasting, etc. I am going to repaste it soon, though, since in investigating this I found that I have a hotspot temperature of 105C, which is a little spicy. That might be related, though I have no idea how to test that without repasting and seeing the results.
I found one other person on Reddit who has the same issue with a GIGABYTE Vision 3070, and their friend has the same issue with a different PC/card (not specified), so it's not just an EVGA or 2070 thing.
My hunch is that the fan controller on fan 1 (and maybe fan 2 now?) is crapping out - I hope not, since as I said earlier the card is out of warranty, but I don't know what else it could be. As far as I can tell it doesn't affect the card's performance, it's just incredibly loud and can't be good for the bearing.
Has anyone seen anything like this before, and might know what's going on?
EDIT: Also maybe related maybe not, my GPU is limiting itself to 75C. Afterburner and Precision X1 are both set to limit it to 83C. GPU-Z reports that the limit is 83C. But whenever it hits 75C, the "Temp Limit" in Afterburner goes to 1, and I've never seen it hit 76C except for maybe half a second. Maybe the fans are spinning like crazy because the card is thermal throttling? How do I set the GPU temperature limit if Afterburner/Precision X1 aren't doing it?
Last edited:
Facebook
Twitter
Instagram