One GPU fan increases speed to 3k RPM when idle, coordinating with other fans during high load
One GPU fan increases speed to 3k RPM when idle, coordinating with other fans during high load
In short, my GPU fans aren't working together when idle, but they synchronize during higher loads and heat up. The graph shows this pattern. For instance, when I started Elden Ring, temperatures increased and Fan2 rose while Fan1 kept pace for a while. After turning it off, the fan speeds dropped before Fan1 suddenly spiked again. I saved a chart for reference. I recorded the data using HWInfo, but rarely use fan control tools. MSI Afterburner is installed, yet the applied curves don’t seem to matter. During startup it runs at a normal RPM similar to the other fan, though I haven’t identified a clear reason. I don’t have a VBIOS to flash yet. There’s some talk online about thermal paste issues with Gainward cards—would that mean the fan stays at a steady speed, not dropping under load? Or does it behave normally when launching? Thanks for any advice! I’ve tried several troubleshooting steps so far.
It seems the PWM controller on the fan or the PWM module on the board is the issue. You might try connecting the faulty fan connector to the socket of the next fan in the GPU without opening the card. Observe if it functions similarly and what happens when you connect another fan to that socket. Please proceed at your own risk if you decide to test it.
@Autissima I have an ASUS DUAL RTX2060 and the fans don't always run at the same RPM. I think I saw/read that one is for the GPU and the other is for the VRAM. If that is the case then it isn't unreasonable that they may run at different speeds depending on what is being done. but I could be totally wrong...
someone else using a 20-series GPU is experiencing fan issues similar to yours, raising doubts about whether it's a driver problem or an incompatible update. Check your own setup here
Hey there, OP. I'm the other user @Amir F. who mentioned in his comment. I own a 2080 super KO (evga) card that faced a similar problem. I just reprinted my card, and although temperatures weren't out of the normal range, it quickly fixed the issue. My thought is there might be some built-in safety that kicks in to spin up the fans when certain conditions are met—this doesn't show up in the afterburner or anything. I also noticed my two fans would spin up separately or together before returning to normal after a restart. Repasting isn't a sure fix, but my temperatures were better before than now, so it might be worth giving it a shot. I bought the Artic MX-4 paste from Amazon for around 4 or 5 USD, which was the most affordable option for me after the software fixes.
Ah I didn't mean open it, just to try and see if the controller is accessible without opening it! The thing is that, as you said, my temps aren't that bad, and the weird part is, mine don't ever hit the max of the fan curve. I graphed some more today, directly started logging after booting my PC. First running furmark until my card heated up once (underlined in red), then played a bit of RDR2 (pink), and when I left RDR2, the fan began spinning out, until I applied a load again (ran furmark again, green). What concerns me is that the card spins far beyond max RPM, which seems to be at ~2.5k RPM, and it's not like it's intermittently, it's practically constantly running at that speed til I shut off the system... I wrote to Gainward and they said if the issue persists they could get me the stock VBIOS file for me to flash, if that doesn't help, safe for any new ideas, I guess I'll bring it to a repair shop to get it repasted by someone less clumsy than myself. Though as I write this (~45 minutes after I stopped the green furmark run), my GPU dropped below 40°C, and the fan also matched the other for a brief moment before returning to ~2.2-2.5k, really weird. Edit: And now 30 seconds later it's back beyond max RPM lol.
the speed isn<|pad|>, staying under max rpm is a common way to display it, like showing 150% for example. Did you orient the GPU correctly? If yes, thermal paste could have shifted due to heat and gravity over time. Use gpuz to check the hotspot when the fan spins faster—likely a paste issue. Others who had similar problems confirmed this. If you need, contact the manufacturer for a refill. Next time, make sure the graph is clearer so it’s easier to read. It was tough to understand what it said, but I realized Windows Magnifier helps a lot.