Published Written and reviewed by 秋和PC
Black Screens, Flickering or Glitched Graphics in Games: How to Find the Cause
The screen goes black for a moment, a driver error appears, or 3D models break up. How to tell whether the driver, power, connections or the graphics card itself is to blame, with a diagnosis we did on a high-end PC.
The screen goes black for a moment during a game, flickers, or textures and 3D models appear broken. These are common signs of a problem around the graphics card (GPU).
That doesn't always mean the card itself has failed, though. The driver, the power cables, or memory and CPU settings can cause the same symptoms. Work through the checks in order before buying new parts.
If the main symptom is a brief freeze every few seconds or audio dropouts, also see what to check when games freeze every few seconds.
Start by recording the symptoms
The most useful thing for diagnosis is a record of what happens:
- Which games and which scenes trigger it (one game only, or several)
- Whether it's a black screen, flickering, broken graphics or an error message
- When it started, and what changed just before (a driver update, a new part, etc.)
- Photos or video of the screen taken with your phone
In Windows "Event Viewer", open "Windows Logs" → "System". You may find display driver errors logged at the moment the screen went black. If events saying that the display driver stopped responding and has successfully recovered keep appearing, the problem is likely somewhere around the GPU.
1. Reinstall the driver
If it happens only in certain games, or started after a driver update, suspect the driver.
- Reinstall the previous stable driver, or the latest one, from the graphics card maker's website (NVIDIA, AMD, etc.)
- On a PC that has been through many driver updates, completely removing the old driver before reinstalling can help
- On laptops, the driver from the laptop manufacturer is sometimes more stable
2. Check power cables and connections
If the driver isn't the cause, check power and connections:
- Is the power connector fully inserted? The 16-pin 12V-2x6 (12VHPWR) connector in particular can cause poor contact and heat if it isn't pushed all the way in
- Are you splitting one cable into several connectors? For cards that need two or more 8-pin connectors, use separate cables from the power supply wherever possible
- Are you using a riser cable? With a vertically mounted card, the riser cable can cause instability
- Is the power supply big enough? If the screen goes off, or the whole PC shuts down, only under heavy load, the PSU may be underpowered or ageing
We cover choosing a power supply in detail in replacing or upgrading a power supply.
3. Check temperatures and overclocking
- Check the GPU temperature with monitoring software. Dust in the fans makes temperatures climb
- If the graphics card or memory is overclocked, set it back to stock
- Temporarily disable the memory overclock profile (XMP / EXPO) on the motherboard and see if anything changes. Unstable memory can look like a GPU problem
4. Lower the clock and watch
If nothing so far has helped, suspect the graphics card itself. One useful test is whether the card becomes stable when its clock speed is lowered slightly.
On one job, a high-end PC with an MSI X870E GODLIKE, an RTX 5090 SUPRIM LIQUID, 256 GB of DDR5 and a Noctua 1600 W power supply was showing black screens, flickering, driver errors and broken 3D models in several games.
During our visit, half a day of testing failed to reproduce the problem. Based on how the symptoms appeared, we suspected the card itself and asked the customer to lower its clock with MSI Afterburner and keep an eye on it. Later, when the customer set the clock back to normal, the problem returned, and the card was sent for repair under the manufacturer's warranty.
If the card is stable with a lower clock and the problem comes back at the normal clock, a hardware fault is likely.
5. Before using the warranty
- Keep your receipt and warranty card
- Keep photos and video of the symptoms, and the Event Viewer records
- Removing the cooler or taking the card apart yourself can void the warranty
If the problem only shows up occasionally, a repair center may return the card saying they couldn't reproduce it. A record of the symptoms makes the repair decision much easier.
What to tell us before booking
- Your PC's specs (CPU, graphics card, power supply, memory)
- Which games and scenes trigger it, and how often
- Photos or video of the screen
- What you've already tried (reinstalling the driver, removing overclocks, etc.)
Even when the problem is hard to reproduce, a good record lets us narrow down the likely causes before we diagnose.
References
- Timeout Detection and Recovery (TDR) (Microsoft Learn) (Opens in a new tab)
- Official NVIDIA drivers (NVIDIA) (Opens in a new tab)
- AMD drivers and support (AMD) (Opens in a new tab)
External pages may change; check the official sites for the latest information.
Related service cases
- On-site gaming GPU error diagnosis in KachidokiOn-site diagnosis of a gaming GPU error in Kachidoki, checking each likely cause before assuming the graphics card had failed.
- On-site ROG desktop power-supply replacement in KachidokiAn on-site power-supply replacement for an ASUS ROG desktop in Kachidoki, with connector, cabling, and startup checks.