Having trouble with fan noise? You’re not alone. Default fan settings often aren’t ideal, usually running too hot or too loud. And then there are those overly-sensitive fans that constantly speed up and slow down during everyday use. While there’s no perfect solution, this fan curve guide should help you strike a good balance between noise and cooling.
The idea is to create fan curves that reduce noise on the desktop while still offering adequate cooling when gaming. It’s not necessarily what everyone wants, but we feel it’s a good middle ground that should work for most users. So, if you’re struggling finding that balance, these fan curves might just be for you. Let’s get started.
How to Set Fan Curves
A quick note before we start. I’ve based this guide on my experience using and running air-cooled rigs in relatively airflow-friendly cases (such as the Fractal Meshify 2). You will have to experiment with temperature breakpoints and fan speeds yourself, depending on case airflow, fan setup, and ambient temperatures. So go through the step-by-step processes outlined to figure out the best temperatures and fan speeds for you.
Fan Control Software
First off, you’ll want to install fan control software so you can set your fan curves without having to keep restarting your system and going into the BIOS. We highly recommend Fan Control: it’s free and open-source, and the developer’s constantly updating it with bug fixes and new features.
I’ll be using Fan Control and MSI Afterburner for this custom fan curve guide. However, most of this advice will apply even if you use other solutions.
Setting a CPU Fan Curve
The most important piece of advice we can give you about CPU fan curves is this: don’t set a linear, constantly ascending graph. CPU temperatures fluctuate a lot when doing basic tasks such as opening a new Chrome tab or launching a program, and you’ll often see momentary spikes of up to 10 degrees Celsius (or more) during daily use.
If you set a simple, linear fan curve, your CPU fans will ramp up and down constantly every time you do something on the desktop. This constant ramping up and down gets very annoying very fast. Not only that, but it’s also of no use since your CPU will be running at safe temperatures anyway, even when it does spike.
To avoid that, we recommend a fan curve that looks like this:
Here, you can see that I have my CPU cooler fans to run at a static 45% up to 65 degrees Celsius. After that point, they ramp up to a maximum speed of 90% at 90 degrees Celsius. The idea with this curve is to keep CPU cooler fans at a constant speed during desktop loads so that they don’t keep speeding up and slowing down every time you do something basic.
These temperature settings work for me, but they probably aren’t ideal for you. So you’ll want to find the correct temperatures for your rig. Here’s a quick step-by-step guide to doing so:
- Decide on the idle fan speed you prefer. Set a fixed fan speed and increase/decrease it until you find a speed with an acceptable noise level.
- Monitor your CPU temperatures for 5 to 10 minutes while doing everyday tasks. My hardware monitor of choice is HWiNFO; if you use it, the reading to look out for is Tdie (or Tctl/Tdie), not Tctl on its own (if present).
- Check that the fan speed you’ve chosen is enough to keep your CPU at an acceptable average temperature (depending on your CPU, this should be around 40 to 50 degrees Celsius). Also, note the maximum temperature your CPU spikes to over the monitoring period.
- Create a fan curve. If you’re using Fan Control, set the temperature source to Tdie (or Tctl/Tdie), and use your idle fan speed as the first point on the curve.
- Add a few degrees Celsius to the max recorded temperature and use this as the second point on your CPU fan curve.
- After this point, you can start ramping up fan speeds under load situations. I prefer a linear increase from 45% at 65 degrees Celsius up to 90% at 90 degrees, but feel free to go higher or lower. You can also add more points to the ramp if you want, but it’s not essential.
Note that Modern AMD CPUs run hot by design, so there’s no need to go overboard and max your fans out at 70 degrees. Ninety degrees Celsius is “typical and by design” for these Zen 3 CPUs under full load.
Setting Case Fan Curves
I set my case fans the same way I set my CPU fans: static up to a specific temperature before a gradual linear increase. This way, I get the airflow under load that I want while maintaining stable noise levels when I’m just browsing the internet.
After all, it’s much easier to tune out a static noise floor than ignore fans that are constantly ramping up and slowing down according to the slightest load.
It’s a roughly similar process here:
- Experiment with static fan speeds to find one that you’re comfortable with for everyday use. This will be your first point.
- Monitor your system temps with this starting fan speed to ensure that there’s enough cooling for components like your chipset and VRM.
- Determine how much airflow you want under load. You can figure out the right fan speed based on noise level or component temperatures; it depends on your priorities. It’s not an exact science, so feel free to experiment. Once you find it, this temperature will be the third point.
Now that you’ve identified the first and third points, it’s time to find the second point where your curve will start ramping up. The second point depends on the temperature source for your case fan curve.
While there is no correct answer, many users use their GPU temperatures as the temperature source for their case fans. Software fan control has a definitive advantage over BIOS fan control here, as you can’t map fan curves to GPU temperatures in the latter. The closest you get is PCIe slot temperature, which often has little relation to your graphics card temperature.
I use one of Fan Control’s advanced custom sensors called “Mix.” It lets me choose two (or more) sources and assign a fan curve to the average, maximum, minimum, sum, or difference between the two.
I’ve selected CPU and GPU temps and chosen the “max” function. With this setting, my case fans ramp up based on the hotter of the two sensors, useful for those situations when a workload only stresses one of the two.
You want to take the same approach as when setting up your CPU fan curve with your chosen temperature source. Identify the maximum temperature your source hits during everyday desktop use and add a few degrees. This is to keep your case fans at a constant speed when not under load. That’ll be the second point in your case fan curve.
I run uniform fan curves for both intake and exhaust, but you can go crazy with individual fan curves or different curves for intake and exhaust. Just make sure you plug your fans into separate motherboard fan headers so that they’re detected individually.
Setting a GPU Fan Curve
More than the other curves we’ve discussed, figuring out your GPU fan curve is a case of thermals vs. noise. GPUs are quick to reduce clock speeds based on temperature, more so than CPUs even. But GPU fans are also tiny and get really noisy once you crank fan speeds up.
I hate noisy GPUs, so I tend to set my GPU fans based on noise. And, unless you don’t care about how loud your graphics card gets, it’s what I’d recommend everyone do. Here’s another quick step-by-step guide to figuring that out:
- Find the maximum fan speed you’re willing to tolerate. It tends to be around 60% with most of the cards I’ve owned, but your mileage may vary.
- Load up a reasonably demanding game and monitor temperatures. You can use Furmark for a worst-case scenario test, but a Triple-A game at high settings and native resolution should be good enough if you’re only planning to game. If you’re using HWiNFO, pay attention to GPU Temperature and not GPU Hot Spot Temperature.
- If the temperatures are acceptable (ideally in the mid-70s at most), then you’re good to go with this particular fan speed.
- If your GPU goes past 80 degrees Celsius and starts thermal throttling to compensate, increase the fan speed until it stays under the thermal limit.
Once you find the right speed, treat it as your GPU’s new maximum fan speed and create a curve accordingly. How you do this will depend on whether you like GPU fan stop.
If you, like me, appreciate GPU fan stop and want to retain that behavior:
- Open MSI Afterburner’s properties, go to the “Fan” tab, and tick the box next to “Enable user defined software automatic fan control.”
- Note the dotted white line on the graph (if present). That’s is the speed at which your fans shut off and will be the first point in your curve.
- Monitor your GPU temperatures with the fans off. Once you find the maximum, add a point a few degrees above that maximum temperature along the dotted line. This is the second point in the graph.
- Add another point at the maximum fan speed you identified earlier at around 70 degrees Celsius or slightly higher. Keep this at or below 80 degrees Celsius.
- If you’re using FanControl for your GPU fan curve, remember to select your GPU as the temperature source.
If you don’t care about fan stop, you’re free to go with whatever fan curve you prefer. But I still recommend checking the fan stop speed with MSI Afterburner, if only to identify how fast you need to set your fans to avoid them turning off.
Hysteresis is the amount that the temperature source must change over a given periodbefore the fan increases (or decreases) its speed according to the curve you’ve set. In more simple terms, Hysteresis adds a bit of “lag” to your fans so that they don’t react immediately to any little temperature change along your fan curve.
It’s helpful in conjunction with the fan curves we’ve set to stop the fans from immediately ramping up if one of our temperature sources crosses the threshold for a split second. It also prevents fans from ramping up and down with minute changes in GPU or CPU temperature under load.
Fan Control offers three hysteresis-related settings. The first setting, “hysteresis,” sets the amount of hysteresis. “Response time” sets the amount of time the fan curve takes to detect a temperature change. Finally, there’s a checkbox to specify whether hysteresis applies during temperature increases and decreases or only when temperatures decrease (MSI Afterburner’s standard behavior).
There aren’t any hard and fast rules here, but a hysteresis setting of around five degrees Celsius is a good place to start. Not every game or workload puts a constant strain on your components, and temperatures can fluctuate even when gaming. The idea is to set a large enough hysteresis so your fans don’t immediately slow down when the load (and temperature) reduces.
For example, I’ve seen component temps drop by two or three degrees Celsius during normal gameplay based on what’s going on in-game (or when I enter a menu). Suppose that happens at temperatures between your two points (during the “ramp”). In that case, your fans will start slowing down and speeding up constantly, putting us back at square one.
Hysteresis keeps your fans going at the same speed even when the temperature drops slightly. It might seem less efficient (why keep the fans running faster than they need to?), but it also means that your fans won’t have to speed back up when there’s more work to do.
Whether to enable hysteresis only on the way down (applicable only to Fan Control) isn’t clear-cut and will require some experimentation. I tend to enable it for my GPU so that it reacts quickly to any significant increases in 3D workloads, but feel free to test it out with other curves to see what works noise- and temperature-wise.
Finding the best custom fan curve settings for your rig is more of an art than a science, and it’s impossible to cover every single approach or philosophy behind fan curves. Some users don’t mind noise and max their fans out for the lowest possible temperatures. Others are even more noise-obsessed and run their fans as slowly as possible with less regard for temps.
We’ve outlined our favorite mid-point approach, aiming for unobtrusive fan noise on the desktop with enough airflow to keep things cool when you’re gaming or rendering. And while we think this is the overall best approach to take, you should experiment and use our guide as a jumping-off point for your own take on fan curves. Have fun!