My RX 580 suddenly stopped working. I had it checked by a technician, but they couldn't determine what went wrong. A few days earlier, I noticed my PC seemed hotter than usual, although there were no strange noises or burning smells. Could excessive heat have damaged the graphics card?
The computer was purchased prebuilt, and I've been replacing parts gradually as they fail or need upgrading. I was considering buying additional reverse-flow fans and a CPU water cooler so I could better control temperatures before installing another GPU. I don't want to spend money on a replacement card only for the same thing to happen again. Would improving the cooling help, or is there something else I should check first?
The system has an Intel Core i5-2500 and an RX 580. The rest of the hardware is also fairly old.
4 Answers
Running components hot for long periods can shorten the life of things like capacitors, even if the main processor protects itself from immediate overheating. That said, there is no guaranteed service life for a CPU or GPU, and old hardware can fail unexpectedly even when temperatures look acceptable. Before replacing the RX 580, check the power supply, display cable, monitor, PCIe slot, and whether the card's fans spin.
Those parts are around 9 to 15 years old, depending on the exact model, so ordinary age-related failure is a very plausible explanation. Better cooling may help extend the life of working hardware, but I wouldn't invest heavily in a water cooler for this platform. Put that money toward a newer system or a quality replacement PSU instead.
The power supply is worth checking before buying another graphics card. If it is the original unit from the prebuilt system, it may be low quality or simply old, and an unstable PSU can damage components or cause symptoms that look like a dead GPU. I would test with a known-good power supply if possible.
The GPU chip itself usually protects against overheating by reducing its clock speed and eventually shutting the system down. A graphics card can still fail from heat, but supporting parts on the board—such as capacitors, MOSFETs, or fuses—are often more likely to fail than the actual GPU silicon. In many cases, diagnosing and repairing one of those components costs more effort than replacing the card.
The GPU die itself doesn't contain the capacitors being discussed, and modern CPUs and GPUs are generally designed to throttle or shut down before heat causes immediate damage. The surrounding board components can still age or fail, but this doesn't mean the GPU was necessarily killed by one hot day.