I'm running an RTX 4080 16GB with a Ryzen 7 9800X3D, 32GB of DDR5 memory, and a 750W Corsair 80+ Gold power supply that is about three and a half years old. My other hardware includes a Focusrite 2i2 and microphone, a keyboard and mouse, and a 480Hz OLED monitor.
The GPU reaches around 98% utilization in some games. At the default 100% power limit, games sometimes cause the monitor to go black and the system to crash. Valorant may run for several matches before failing, while GTA V with graphics mods recently began crashing after roughly ten minutes. After several crashes in a row, I lowered the GPU power limit to 70% in MSI Afterburner, and the crashes stopped.
Could the power supply be struggling with the GPU's power draw or transient spikes? Or is this more likely a temperature or thermal-limit problem? I also checked HWiNFO while the game was running before lowering the power limit, and it appeared that the card may have been reaching a thermal limit.
3 Answers
Don’t assume the PSU is definitely at fault yet. A lower power limit also reduces core and hotspot temperatures, and the HWiNFO readings may indicate the card was thermal throttling. Monitor GPU temperature, hotspot temperature, memory temperature, clock speeds, and the reason for thermal or power-limit throttling while reproducing the crash. Clean the card and case filters, verify that the GPU fans are working, and check that the card has adequate airflow.
Black screens under heavy GPU load can also come from an unstable driver, factory overclock, bad cable connection, or an unstable overclock. Try a clean graphics-driver installation, return the GPU and memory settings to stock, reseat the GPU power connector, and avoid sharp bends near the connector. If the system remains stable at 70% but crashes at 100% after those checks, the PSU or the graphics card itself becomes much more suspicious.
A 750W unit can be borderline for an RTX 4080 system, especially if it is an older model or connected through an adapter. Reducing the power limit lowers both heat and transient power demand, so the improvement does point toward either power delivery or temperatures. Check the exact PSU model, GPU temperatures, hotspot temperature, and whether the power cables are fully seated. If possible, testing with a known-good, newer 850W ATX 3.x PSU would help isolate the problem.
The fact that it happens in more than one game makes me think it is not just a GTA V mod problem. I’ll compare the GPU core and hotspot temperatures and test with another power supply if I can.