My PC works normally for browsing and other light tasks, but after playing games such as Counter-Strike or Rainbow Six Siege for a while, the game may freeze and the monitor suddenly reports no signal. The display sometimes switches between analog and HDMI even though the connection was working moments earlier. Restarting the graphics driver with Win+Ctrl+Shift+B and completely removing and reinstalling the drivers with DDU did not help.
The computer itself appears to remain powered on: the fans and lights stay active, and during some crashes I can still hear the game, control my character, listen to Spotify, and talk through Discord. Other times the game closes completely. The only reliable recovery is holding the power button to shut the PC down, waiting several minutes, and then starting it again. If I restart immediately, the display often still does not return.
The graphics card reached roughly 60°C with a 70–72°C hotspot before the last crash, so it does not appear to be overheating in the usual sense. I am using the original power-supply cables with a Gigabyte P650B and a GTX 1660. What should I test next, and is the graphics card or power supply more likely to be failing?
3 Answers
The power supply could still be involved even if the temperatures look fine. Try reducing the GPU power limit substantially—around 80%—and see whether the crashes stop. If the system becomes stable with a lower power limit, that points toward a power-delivery problem or a card that is no longer handling its normal load reliably. Also check that the GPU power connector is firmly seated and that the card is properly installed.
The symptoms sound more like the GTX 1660 entering an unrecoverable hardware failure than a normal driver crash. The fact that the PC continues running while the display disappears, the driver reset does nothing, and the card only works again after being completely powered down are all consistent with the GPU failing under load.
A GPU can have a normal core temperature and still have a hardware or power-delivery fault. Older cards can also develop bad solder or connection problems after years of repeated heating and cooling. In that situation, the card may work when cold, fail during gaming, and recover only after it has cooled and lost all power. Before buying anything, test with another known-good GPU if possible, or test this card in another computer. If the problem follows the card, replacement is usually more practical than repair. Do not try an oven or similar heating method; it can permanently damage the card.
That does sound very similar to what is happening here. Even though the GPU and hotspot temperatures were only around 60°C and 70–72°C, the need to wait several minutes before the display works again makes a hardware problem seem possible. I’ll try another graphics card or consider replacing the 1660.
Before settling on a diagnosis, check the Windows Event Viewer or Reliability Monitor after a crash for display-driver resets, hardware errors, or unexpected shutdowns. You can also try a different HDMI/DisplayPort cable and another monitor input, reseat the graphics card, and run a memory test. The most useful isolation test is swapping parts: try another GPU in this PC, or run the GTX 1660 in another system. A clean driver installation alone cannot rule out a failing card, and temperature readings do not rule out faults in the GPU’s memory, VRMs, or physical connections.
I tried setting the power limit to 90%, but it still crashed. I may test 80% as well, although I’m not sure it will make a difference.