My PC randomly powers off and immediately restarts while gaming or using GPU-heavy applications. It happens roughly once or twice a day, usually when the GPU is near 99% usage, although it is not completely consistent. The problem began after replacing my graphics card and power supply.
Current system:
- Gigabyte B760M DS3H AX DDR4 motherboard
- Intel Core i7-12700 processor
- NVIDIA GeForce RTX 5060 Ti graphics card (new)
- Corsair Vengeance LPX 3200 32 GB memory
- Samsung 980 Pro 1 TB NVMe SSD
- Corsair CX650 power supply (new; previously used a CX550)
The PC also has occasional blue screens when XMP is enabled, including BAD_POOL_CALLER and Kernel Mode Heap Corruption. My two memory modules have different revision numbers, so I suspect the RAM or XMP instability may be a separate issue.
The PC is connected to a UPS that makes a clicking sound when the shutdown and restart occur. The UPS is about three years old. Could the UPS be overloaded or failing, or should I suspect the power supply, GPU temperatures, or another hardware problem? I am especially concerned because the shutdowns happen while working in After Effects and unsaved projects are lost.
3 Answers
The UPS may be the first thing to test. Some UPS units click when switching modes, and an overloaded or aging unit can shut down when the GPU suddenly draws more power. Try running the computer directly from a properly grounded wall outlet or a basic surge-protected power strip for a while, without changing anything else. If the crashes stop, check the UPS's rated wattage rather than only its VA rating.
A sudden power-off can also be caused by a GPU temperature problem or a transient power spike that exposes a weak power connection or power supply. Use an overlay or monitoring tool to record GPU temperature, hotspot temperature, clocks, and power while testing. Reseat the graphics card and its power cable, avoid adapters if possible, and make sure every connector is fully inserted. The CX650 may be sufficient on paper, but testing with a known-good, higher-quality unit is useful if the UPS is not responsible.
The blue screens point to a separate stability issue that should be dealt with independently. Turn XMP off and test the system at default memory settings first; mixed memory revisions or marginal modules can cause BAD_POOL_CALLER and Kernel Mode Heap Corruption. Check Windows minidumps for the blue-screen crashes, run a proper memory test, and disable automatic restarting if you need to capture the error. Until the system is stable, avoid assuming every failure is caused by the new GPU.
I will test it without the UPS, but I want to make sure the other possibilities are checked too before relying on that result.