I am getting random gpu drops in games

Random GPU usage drops like those shown in the graph are typically related to how the game and drivers interact with the GPU rather than a power-supply or hardware-capacity issue, especially with high‑end hardware.

Use these steps to narrow it down on Windows 11:

1. Check for driver/TDR or LiveKernelEvent issues
   • Open Reliability Monitor (search "View reliability history").
   • Look for Hardware error, LiveKernelEvent, or display‑driver errors around the times the drops occur. These can indicate TDR (Timeout Detection and Recovery) or GPU driver faults, which are known to cause black screens, freezes, or sudden GPU usage changes under load.
   • If such errors appear repeatedly, cleanly reinstall or change GPU drivers. In similar gaming cases, using Display Driver Uninstaller (DDU) and then installing a different NVIDIA driver version (often a slightly older one) has resolved random crashes and TDR‑related issues.
2. Update or reinstall GPU drivers
   • Even if drivers are current, a corrupted install can cause instability. A clean reinstall using the GPU vendor’s full driver package (not only auto‑update tools) has restored proper multi‑GPU reporting and behavior in other Windows 11 systems after updates.
3. Monitor temperatures and stability
   • Use a monitoring tool while gaming to watch GPU temperature and clocks. Overheating or thermal protection can cause sudden usage drops, black screens, or watchdog/TDR events. In one case, dried‑out GPU thermal paste led to protection kicks at ~80°C and black screens under load; repasting and ensuring proper cooling fully stabilized GPU clocks and usage.
4. Check for system freezes or hangs
   • If the usage drops coincide with short freezes, check Event Viewer → Windows Logs → System for repeated display or kernel‑power warnings around those times. These can be related to TDR or watchdog timeouts that temporarily reset or stall the GPU.
5. Profile the game’s GPU usage
   • If developing or modifying games/apps, use Visual Studio’s GPU Usage tool (Debug → Performance and Diagnostics → GPU Usage) to see whether the app is CPU‑bound or GPU‑bound, and to correlate frame‑time spikes with GPU activity. This helps distinguish normal workload variation from driver or hardware problems.

If Reliability Monitor and Event Viewer show no hardware or LiveKernelEvent errors and temperatures are normal, short GPU usage dips—especially during loads, scene changes, or streaming in open‑world games—are often normal workload fluctuations rather than a fault.

References:

• GPU usage
• Use the GPU Usage tool
• Use the GPU Usage report
• Bug Check 0x117: VIDEO_TDR_TIMEOUT_DETECTED
• Use DRED to diagnose GPU faults
• My Windows 11 PC Hard Freezes for about 30 seconds to a minute when I play certain games. - Microsoft Q&A
• Recurring "LiveKernelEvent 1b8" hardware error (Watchdog Timeout) - Microsoft Q&A
• Minecraft for windows is crashing - Microsoft Q&A
• Task Manager not longer shows my GPU 1 after Windows Update 5/3/26, how do I fix this - Microsoft Q&A