My computer randomly powers off while playing certain games, sometimes after only a few minutes and other times after several hours. It shuts down as if the power was suddenly cut—there is no warning, blue screen, or error message. A repair shop tested it four times but never found a problem. I have already changed BIOS settings, replaced the power supply, connected the system to a UPS, replaced my HDMI and DisplayPort cables, swapped my mouse and keyboard, and replaced my 240 Hz monitor. None of that helped. The affected games vary widely in age and performance requirements, including Crime Boss: Rockay City, NBA 2K26, and War Dogs. The problem seems less likely to happen when I disconnect my second monitor and use an older 60 Hz display, but I do not want to rely on that setup after upgrading my PC. While recording hardware statistics on the second monitor, I saw the CPU reach about 89°C while the GPU was only at 69% usage. I have not repasted the CPU recently. What else should I test?
3 Answers
Try logging the CPU temperature, CPU package power, GPU temperature, GPU hotspot, clock speeds, and any WHEA errors right up to the shutdown. The second-monitor difference may be a clue, but it could also simply change the GPU load or timing rather than prove the monitor is faulty. If the CPU temperature is reaching its limit, fix the cooler mounting and thermal paste first. If temperatures are normal, test the system with a different graphics driver, a single monitor, and a CPU or GPU stress test separately to identify which component causes the power loss.
A complete black-screen power-off usually points to either the PSU protection circuitry triggering or the CPU reaching its thermal limit. Since you already replaced the PSU, the CPU temperature is the stronger suspect, assuming the GPU is powered with the correct separate PCIe cables. Use HWiNFO or similar monitoring software and check the CPU package temperature, thermal throttling flags, and maximum temperature while playing. Many processors will shut down around their safety limit—often roughly 85–95°C on AMD systems and near 100°C on Intel systems. The 89°C reading you captured is high enough to investigate seriously, especially if it keeps climbing or briefly spikes higher.
The repair shop may not have reproduced the problem because their testing was too short or they tested with the case panel removed. Heat can build up gradually inside the case, then a sudden CPU workload spike pushes the processor over its limit. Check that the CPU cooler is mounted evenly, the thermal paste is still good, and the fan and pump—if it is liquid-cooled—are actually working. Also verify that intake and exhaust fans are oriented correctly and that the case is not restricting airflow. If temperatures remain high, repasting or using a larger cooler may help, although loosening the mounting screws can make contact worse if the cooler is no longer firmly seated.
I moved the computer onto the desk to improve airflow and loosened the four CPU-cooler screws by about three turns, but War Dogs still caused a shutdown after around 20 minutes.