Question Ventus 3x OC one fan stuck spinning at 100% under zero load, even after reboot

Jun 3, 2021
Heads up: A similar but not identical thread:
In my case, I can get all fans to spin under load. It's just that Fan #1 can't stop spinning. :(

TL;DR: One fan on my (already once RMA'd) Ventus OC 3X 3090 always spins at 100% speed whenever the PC is on, even under zero load. Any software solutions? Should I RMA?

Hi everybody,

In early May, I was fortunate enough to snatch an MSI 3090 VENTUS 3X OC from BestBuy, here in Canada. Paid an arm and a leg for it, but at least it wasn't from a scalper, and I need the large amount of memory for my work in deep learning.

Five days later, my GPU died somewhat randomly, and it was so bad that it even prevented my PC from POSTing when it was slotted in. The computer would not boot with the GPU in, and the VGA debug light (B550 Tomahawk mobo) was on. I tried friend's 1080, system worked like a charm.

I RMA'd the GPU. Pretty annoyed I had to pay for the shipping out of pocket on such an expensive GPU, but whatever.
They fixed it and shipped it back. All in all, this took about two weeks from me shipping the GPU to getting it back, and that's with a long weekend in the middle. Not too bad, I suppose.

As soon as I got my GPU back, I installed it back into my PC and, to my joy it worked well and my computer was able to boot! Everything looked fine in Windows and GPU-Z.

(As a side note, even though my model has three fans, only two of them are showing up in GPU-Z or MSI Afterburner. It seems that "Fan 1" is the first fan, and "Fan 2" jointly represents the second and third, i.e., the two closest to the HDMI ports. Not sure if this is a problem, but it's relevant to the problem at hand.)

(Anothe side note, my machine has a 5800X CPU 64G of RAM and a WD Black NVME SSD.)

In order to test the GPU's stability, I decided to run Furmark for a few minutes, just to make sure nothing too crazy was happening. I ran FurMark at 2560 x 1080 8x MSAA and everything went well, until all of a sudden, Fan 1 went crazy, ramped up to 100% (3400 RPM) and stayed there. Fan 2 seemed to actually slow down as a result.

After a minute or so I stopped FurMark, and the GPU started gradually cooling down. However, Fan 1 was stuck at max speed. Moreover, in both GPU-Z as well as MSI Afterburner Fan 1 speed % is shown as 0%, even though its RPM/tachometer is 3400. Fan 2 is at 0% 0 RPM, as the GPU is not under load.

If I look at my GPU now, it seems one fan is running at 100% speed, while the other two are stopped!!
GPU temp: 28C, Hot spot: 38C, Memory: 36C. Fan 1 going crazy! And loud...

This happened to me earlier today, and after rebooting a couple of times it went away. Then, as soon as I ran FurMark again, the fan eventually ramped up, then got stuck at 100%, and here I am...

I shut down the computer for a day, and then that fan was stuck 100% again as soon as I hit the power button. Now it won't stop unless the computer is off.

Things I tried and did not work:
  • downgrade windows NVIDIA driver to 457.51
  • uninstall MSI Dragon Center, MSI Afterburner, and reboot, nothing
  • reboot into Linux - nope, Fan 1 maxes out as soon as I boot, before I can even enter the bios...
  • force fan speed from Linux - nope
  • force fan speed from MSI Afterburner in Windows - nope (I can control Fan 2, but Fan 1 does not respond...)
  • tried the new MSI Center (non-Dragon) - it did not help and seems like a step back from Dragon Center IMO. Feels very limited.
Why am I worried? The loudness is not a huge thing, but I'm worried this could be indicative of a deeper problem with the GPU. Furthermore, with a fan at 100% even when the GPU is idle, I wouldn't be surprised if it dies soon...

In your experience, is there anything I could try to solve this? Is this definitely a hardware issue? Should I RMA the GPU again?

Thank you for reading this long post, and thanks in advance for any help.


Mar 11, 2021
Absolutely rma it. There's no reason why you have to try to fix something that is in rma. I would not rma through msi, but through the shop you've bought it. MSI is well known for doing jacks**t when it comes to rma. To them, unless something doesn't power up, it doesn't have issues, so they basically don't do anything.

I had a similar problem with my old 1070 i put on my girlfriend's pc before (for that reason) swapping it with a 5700xt (since i got lucky and i got a tuf 6900 for almost msrp). Nothing i did worked. I even tried to completely format her rig (like not a simple reset, but a whole format) and even that didn't work. As soon as i would install the nvidia drivers and launch a stress bench/game the same thing would happen, even when hwinfo and aorus engine would read the temps fine and the curve supposedly to 0db. Only solution i could find was to remove the nvidia driver (but ofc that pretty much renders the card useless, only good for browsing). Even reboots wouldn't fix it, as soon as the pc would reboot the fans would keep ramping up to 100% at the start of windows. As i said, only way to stop em, was to not install nvidia drivers.

Only thing i didn't do was trying to flash the vbios. You could try that. Altho i would've been doing it on an old 1070, you'd be doing that with a 3090, so yeah. Unless it has a bios switch, don't.

What i ended up doing was to declare the gpu "dead"... or better, i am assuming there was a problem on the pcb sensors reading, making the fan spin to 100% even tho the temps were idle temps (35 degrees both on aorus and on hwinfo as i said). Managed to sell it for 300 eu on some finnish ebay like website (my gf is finnish). MIND YOU, clearly stating the issue (so it's not like i scammed anyone). But even then it luckily sold lol.
Last edited: