Whether you’ve just bought one of the best graphics cards or you’re curious how your older GPU is faring, knowing your GPU’s safe temperature range is important.
To assess your graphics card, it’s crucial that you know what its normal GPU temperatures should be.
If your GPU’s temperature is too high, you’ll need to know how to lower your graphics card’s temps.
There are several ways to do this, and most are pretty simple, so don’t worry if you notice your GPU’s running a little hot.
This guide will tell you what GPU temperatures you should be aiming for, how to see whether your graphics card is operating within this safe range, and what to do if it isn’t.
The Quick Answer – What is a Safe GPU Temperature?
Modern NVIDIA GPUs should stay below 85C under full load to be safe, although many can exceed this by a few degrees before hitting their max temperatures.
Modern AMD GPUs should stay below 100C under full load to be safe, although the most recent AMD GPUs hit their max temperature at 110C.
To check your GPU temps, load a graphically intensive benchmark like Unigine Valley while monitoring your GPU’s temperature with software like MSI Afterburner. If it’s running too hot, remove any overclocks and make its fan curve more aggressive.
If this doesn’t work, check everything is running properly inside the PC, especially the graphics card’s fans. And if all else fails, you can install an aftermarket GPU cooler or RMA your graphics card if it seems so hot as to be defective.
GPU Temperature Limit – Normal GPU Temperatures
GPU manufacturers put safety measures in place to throttle its performance or shut it down when its temperature hits a certain maximum threshold – this is the temperature limit.
Normal temperatures should be as far below this limit as possible. Not only will this ensure that performance isn’t throttled and your games don’t crash, but it will also reduce long-term damage to your card and increase its longevity.
NVIDIA Max Temps
To check your NVIDIA card’s maximum temperature, you can look at its specifications as listed by NVIDIA. NVIDIA RTX 30-series cards have a maximum temp of 93C, meaning they can run this hot without throttling speeds or shutting down.
Older generation cards have similar but not identical limits. NVIDIA GTX 10-series cards, for example, have a maximum temp of 94C, and RTX 20-series cards have a max temp of only 88C.
For all three generations, a good baseline is to try to keep your graphics card below 80C when gaming, because this gives you a little leeway.
AMD Max Temps
An AMD blog post tells us that AMD 5700 GPUs have a maximum temperature of 110C before throttling. The same is true of current-gen AMD cards like the AMD RX 6700 XT. If the GPU sensors detect a ‘hotspot’ or ‘junction’ temperature (‘Tj-max’) of 110C or above, it will begin to throttle performance or, if necessary, shut down.
So, AMD cards should be kept below 110C. But remember, this is a maximum limit, so modern AMD GPUs should be kept well below this in practice. While you shouldn’t think it’s the end of the world if your modern AMD GPU is hitting 100C, it might be time to try some of the steps below to try and reduce its temperature.
Why is my GPU Hot?
Any electrical component will generate heat, because energy and power conversions are never completely efficient, and energy is often lost in the form of heat. But the GPUs in graphics cards can become damaged if they get too hot, which is why graphics card manufacturers spend so much time designing different ways to cool them down.
If your graphics card is getting too hot and it isn’t faulty – a faulty card can be the culprit, sometimes – then this might be because of a bad fan curve setup, ambient room temperatures, your case’s airflow, your graphics card’s overclocked speeds, or your cooling setup.
How to Check your GPU Temp
Checking idle temperatures can be useful, but to ensure your graphics card stays within a safe temperature range, you should make sure it’s staying cool enough when under 100% load.
GPU Temperature Monitoring Software
There are many different programs you can use to check your GPU’s temperature.
The easiest way is to load up Task Manager (press Ctrl + Shift + Esc), select the ‘Performance’ tab, and select your GPU from the list on the left – it should give you its temperature at the bottom.
You can also use all-purpose hardware monitoring software such as HWMonitor, which should have your GPU’s temperature listed in a separate sub-section.
My favourite GPU temperature monitoring program, however, is MSI Afterburner. This software doesn’t just monitor your GPU’s temps, but also lets you adjust its core clock, memory clock, and its fan speed.
What makes Afterburner great for monitoring your GPU’s temperature is its graph, which can be brought up by pressing the bottom-left icon. This will give you a simple picture of how your GPU’s temperature relates to its load levels over time. As you can see, my GPU consistently maxes out at 82-83C under 100% load while running Unigine Valley benchmark.
Given that my NVIDIA GTX 1070 is rated for a maximum temperature of 94C, I can rest assured that it is operating within a safe temperature range.
How to Test GPU Temps
You should test your GPU by putting it under load. If you don’t want to download software to do this, you can simply load up and run the most graphically demanding games that you own. Monitoring software should tell you whether your GPU is running under full (or close to full) load while playing these games.
Apart from running games, however, you can install benchmarking or stress testing software like Unigine Valley to put your GPU under load.
Install and run the benchmark, set its settings to high enough that it puts your GPU under full load, and run it for a few minutes.
While it’s running, load your GPU temperature monitoring software and keep your eye on your GPU’s load and temperature. If you use MSI Afterburner, you can open its graph utility and leave this and the benchmark running for a few minutes while you do something else. Then, close the benchmark and look at Afterburner’s graph, noting your GPU’s temperature under different loads.
How to Lower your GPU Temp
If you’ve tested your GPU’s temperature and found it to be too hot, there are several things you can do to fix it:
- Remove GPU overclocks
- Increase your graphics card’s fan speed
- Check your cooling works
- Clean your graphics card
- Improve your PC’s cable management
- Reapply thermal paste
- Install a third-party cooling solution
If none of these solutions work and your GPU is still too hot, you might have a faulty card and should contact the manufacturer.
Remove GPU Overclocks
If you’ve previously overclocked your GPU, the first thing you should try is resetting its core clock, memory clock, and voltage to default values. MSI Afterburner should allow you to display the offset in clock and voltage values, which you can then reset to zero. Or you can select the ‘reset’ icon at the bottom to reset all values to their factory defaults.
You can even ‘under-volt’ your GPU to lower temperatures, but this is only recommended if you’re experienced, and shouldn’t be necessary to keep a GPU in its safe temperature range.
Increase your graphics card’s Fan Speed
If your GPU’s running at factory settings and is still too hot, the next thing you should try is making its fan curve more aggressive – in other words, set the fan speed to increase at lower temperatures and top out at a higher speed.
If you don’t want to configure the curve in detail, use MSI Afterburner to lower the GPU’s ‘temp limit’. This should make the graphics card automatically increase its fan speed to prevent the GPU from surpassing the limit.
My temp limit was originally set to 83C. To reduce my GPU speed, I simply lowered this limit to 77C and hit apply.
As you can see, this lowered my GPU’s temperature from 83C max to 77C max when I ran Valley benchmark again.
If you want more control over how much and when your graphics card increases its fan speed, Afterburner and other fan speed programs should allow you to configure this in more detail.
Check your Cooling Works
If the above doesn’t work, the only option left is to dive into your PC’s innards.
If your GPU is running very hot, there might be a problem with its cooling system. You should boot up your PC with its side-panel removed, load a graphically intensive game or benchmark, and watch to ensure all the graphics card’s fans are spinning.
If they’re not, once your PC is turned off, try reseating the graphics card in its PCIe slot and reconnecting its power cables. If they still don’t spin, contact the manufacturer.
Clean your Graphics Card
If your graphics card’s fans are spinning but it’s still too hot, you could try cleaning it. Once your PC is turned off, unplug your graphics card and brush between its fan blades to dislodge any dust. Use compressed air from a few inches away to blow away surface dust, and blow inside the graphics card in the direction of airflow to dislodge any dust inside.
Remember to use anti-static equipment if you’re worried about ESD (electrostatic discharge). An anti-static brush and grounding wrist strap should be sufficient.
Improve your PC’s Cable Management
Improving your PC’s cable management can sometimes help reduce your GPU’s temperature. While airflow isn’t as important for component temperatures as many people think, it can’t hurt. As we know, the air temperature surrounding the graphics card can affect how hot it gets, and ensuring hot air leaves your PC case quickly can help.
To sort your cable management, unplug all the PSU’s cables from the various components inside your case and begin afresh. Try to utilise the rubber grommets or pass-throughs to pack cables behind the backplate. Also route cables together if they’re on a similar path and use zip- or Velcro-ties to keep them pressed to the case.
Reapply thermal paste
If none of the above has worked and your GPU’s temps are dangerously high, it might be time to RMA your graphics card. But if you want to give it one last shot on your own, you can reapply your GPU’s thermal paste.
Thermal paste is what sits on top of your GPU to help transfer heat to its cooling system. Over time, this thermal paste can degrade and lose its conductivity, making it harder for your GPU to offload its heat.
If you’re in the DIY mood, you can take apart your graphics card, remove its thermal paste, and reapply it. But you should only do this if you know your way around your hardware and are comfortable prising open a graphics card.
Install Third Party Cooling Solution
If all else fails, you can throw the towel in with your graphics card’s stock cooling solution and opt to buy an aftermarket one. This will mean removing the manufacturer’s cooling system – fans, heatsink, and all – and installing your own one on top of the PCB and GPU.
It goes without saying that you should only do this if you’re comfortable taking your graphics card apart and piecing it back together.
Aftermarket GPU coolers are often very good at what they do. You can even get liquid or hybrid liquid-air GPU coolers, which should not only reduce the GPU’s temperature but also do so very quietly.