Since the 1.2.1 update for 007 First Light, my entire PC sometimes powers off instantly when Path Tracing is enabled. This is not a normal crash or crash-to-desktop: there is no blue screen, freeze, or warning, and Windows only records Kernel-Power Event ID 41 with BugcheckCode 0 after reboot.
My system has an RTX 5080, Ryzen 9 7950X3D, MSI X870E Carbon WiFi motherboard, 32 GB DDR5-6000, MSI MPG A1000G PCIe5 1000W PSU, Windows 11, NVIDIA driver 616.92, and a 3440x1440 display. The game is running through Steam on version 1.2.1.
The shutdowns occurred with both High and Medium Path Tracing. High usually lasted 20–40 minutes, while Medium lasted about an hour before shutting down. Reducing the GPU power limit to 80%, reinstalling the graphics driver with DDU, and capping the game at 60 FPS did not fix it. In one 60 FPS test, the system powered off after roughly 24 minutes even though GPU power was only around 240 W.
Before one shutdown, the GPU was at about 313 W, 74.8°C, with an 88.6°C hotspot and 76°C VRAM. During another test it was closer to 240 W, 73°C, and an 86°C hotspot. The CPU reached around 88°C, but there were no thermal limiters, WHEA errors, or PCIe errors in the logs. Other demanding games, including Cyberpunk 2077 with heavy Path Tracing, remain stable for hours on this system.
The problem seems more likely in scenes with many reflections, glass, marble, indirect lighting, and multiple light sources. I have sent HWiNFO logs, Event Viewer data, DxDiag, and NVIDIA crash dumps to the game developer, and the issue has been forwarded to their technical team. Has anyone else experienced complete power-offs with Path Tracing on an RTX 5080 or 5090, especially after version 1.2.1? I am looking for possible workarounds besides disabling Path Tracing entirely.
3 Answers
The CPU temperature around 88°C is worth checking, although it does not by itself explain why other demanding games are stable. Verify what cooler is installed, check average CPU usage, and confirm that the CPU and GPU are not hitting thermal or power protection limits. The lack of WHEA and thermal-limit entries makes the PSU or a game/driver-specific GPU fault more suspicious than ordinary overheating.
A clean graphics-driver reinstall using DDU in Safe Mode is worth trying, followed by the newest available NVIDIA driver. It may not address a hard power loss, but it can rule out a corrupted driver installation.
Several people pointed to the power supply as the first thing to investigate. A system that instantly powers off under gaming load can be reacting to a transient power event even when average GPU wattage and temperatures look normal. Power-limit reductions and frame caps do not necessarily eliminate very short spikes. If possible, test with a known-good replacement PSU and check that the GPU power connector is fully seated.
An aging PSU can develop capacitor issues and lose its ability to handle transient spikes, causing sudden shutdowns or restarts. Testing another unit is more useful than relying only on the reported wattage.