sodcalculator

10 Tech How-To

Fix GPU Overheating Crashes & Artifacts

Guide 10-45913 · Updated · Step by step

If your graphics card is overheating, you’ll likely see sudden crashes, screen freezes, or visual artifacts like flickering polygons, colored dots, or distorted textures. This guide walks you through the most common causes and fixes—from cleaning dust to adjusting fan curves and checking thermal paste—so you can get your GPU running stable and cool again. Before starting, make sure your PC is powered off and unplugged, and have a can of compressed air, a soft brush, and a screwdriver (usually a Phillips #0 or #1) on hand.

What You’ll Need

  • Compressed air can or an electric air blower
  • Soft anti-static brush or microfiber cloth
  • Isopropyl alcohol (90% or higher) and cotton swabs (if reapplying thermal paste)
  • New thermal paste (if you plan to replace it)
  • Screwdriver set (typically Phillips head)
  • Access to your PC’s BIOS/UEFI and GPU software (like MSI Afterburner or your GPU’s control panel)

Step-by-Step

  1. Confirm the problem is temperature-related. Download a GPU monitoring tool (like MSI Afterburner or HWiNFO) and run a stress test (e.g., FurMark or a demanding game). Watch the temperature and note when crashes/artifacts appear. If the GPU hits above 90°C (194°F) at load, overheating is likely the cause. Also check if the fan spins and if the card is dusty.
  2. Clean the GPU and its cooler. Power off and unplug the PC. Remove the GPU from the PCIe slot (press the latch, then gently pull). Use compressed air to blow out dust from the fan blades, heatsink fins, and the PCB. Hold the fan blades still with a finger while blowing to prevent overspinning. Use a soft brush to dislodge stubborn dust. Reinstall the GPU securely.
  3. Improve case airflow. Ensure your case has unobstructed intake (front/bottom) and exhaust (rear/top) fans. Move cables away from fans and GPU. If your case has a side panel, temporarily remove it to see if temperatures drop significantly—if they do, you need better airflow.
  4. Set a more aggressive fan curve. Open your GPU software (e.g., MSI Afterburner) and create a custom fan curve. For example, set fans to 50% at 60°C, 70% at 70°C, 90% at 80°C, and 100% at 85°C. Apply and test. This helps keep temps lower under load.
  5. Undervolt the GPU (optional but effective). In MSI Afterburner, lower the core voltage (usually by 50-100 mV) while keeping the clock speed stable. This reduces heat output without significant performance loss. Test with a stress test to ensure stability.
  6. Replace thermal paste (if cleaning and airflow don’t help). Remove the cooler from the GPU (unscrew the backplate and fan shroud). Clean old thermal paste from the GPU die and heatsink with isopropyl alcohol and a lint-free cloth. Apply a small pea-sized drop of new thermal paste on the die, then reattach the cooler evenly.
  7. Check the GPU’s power cables and seating. Make sure the PCIe power connectors are fully inserted and the card is properly seated. Loose connections can cause instability that mimics overheating issues.
  8. Update or reinstall GPU drivers. Use Display Driver Uninstaller (DDU) in Safe Mode to remove old drivers, then install the latest driver from the manufacturer’s website. Corrupted drivers can cause artifacts and crashes that are not temperature-related.
  9. Test with a different PCIe slot (if available). Sometimes a damaged slot can cause issues. Move the GPU to another slot and test.

Common Mistakes

  • Using a vacuum cleaner: Vacuums create static electricity that can damage components. Always use compressed air.
  • Spinning fans with compressed air: This can over-rev the fan and damage its bearings. Always hold the fan still.
  • Applying too much thermal paste: A pea-sized drop is enough. Too much can insulate the die and cause worse temps.
  • Ignoring ambient temperature: If your room is hot (above 30°C/86°F), your GPU will run hotter. Ensure adequate room cooling.
  • Overclocking while troubleshooting: If you have an overclock, reset to stock clocks first. Overclocks can cause artifacts and crashes even at normal temps.

Tips

  • Monitor your GPU temperature regularly (every few months) to catch rising temps early.
  • Set a custom fan curve that is more aggressive than the default—many cards run too passively out of the box.
  • If you’re comfortable, replace thermal pads on the VRAM and VRM when replacing paste; they also degrade over time.
  • Use a stress test that runs for at least 15 minutes to confirm stability.
  • Keep your case filters clean—they trap dust but need regular cleaning to maintain airflow.

Troubleshooting

If the steps above don’t resolve the issue:

  • Still crashing after cleaning and new paste? The GPU may have a hardware fault (e.g., failing VRAM or solder joints). Try underclocking the GPU memory and core significantly (e.g., -100 MHz core, -200 MHz memory) to see if it stabilizes. If it does, the card is degrading and may need professional repair or replacement.
  • Artifacts only in certain games? This could be a driver issue or a game-specific bug. Try a different game or run a benchmark like 3DMark to see if the issue is universal.
  • PC shuts down instantly under load? This could be a power supply issue. Check your PSU wattage and ensure it’s sufficient for your GPU. A failing PSU can cause shutdowns that mimic overheating.
  • Temps are fine but still crashing? Use a tool like MemTestG80 or OCCT to test GPU VRAM. If errors appear, the VRAM is failing and the card is likely beyond simple fixes.
  • Card is under warranty? Contact the manufacturer before attempting any physical modifications like repasting, as that may void the warranty.

Frequently Asked Questions

What temperature is too hot for a GPU?

Most GPUs are designed to run up to around 85-90°C (185-194°F) under load. Sustained temps above 90°C can cause thermal throttling and increase the risk of damage. If you see 95°C or higher, that’s critical.

Can artifacts be fixed with software?

Sometimes. If the artifacts are caused by a driver bug or overclock instability, updating drivers or resetting clocks can fix them. However, if artifacts are due to physical damage (like cracked solder joints), software won’t help.

How often should I clean my GPU?

Every 6-12 months depending on your environment. If you have pets or smoke, clean more often. Dust buildup is the most common cause of overheating.

Is it safe to use a vacuum cleaner on GPU?

No, vacuums generate static electricity that can damage sensitive electronics. Always use compressed air or an anti-static brush.

Will undervolting reduce performance?

Undervolting can slightly reduce performance (usually less than 5%), but in many cases you can maintain the same clock speeds at a lower voltage. It’s a great way to reduce heat and fan noise.

Should I replace thermal paste on a new GPU?

No, new GPUs have fresh paste. Only replace it if the card is older (2+ years) and you’re experiencing high temps after cleaning.

Related

General steps for current mainstream versions; menus can differ by device and version. See the disclaimer or report an error.

More in this bed

The whole Tech How-To bed →