How it works
HWiNFO sensors-only, 1s polling, reproduce the problem. Add RTSS for framerate.
Overview flags the crash signature, rail sags, and why clocks dropped.
Charts and Baselines show the exact moment and what "normal under load" looks like.
The Validate tab walks you through stress tests, Xid codes, and RMA evidence.
No log handy? Try a real one
These are trimmed from actual HWiNFO crash logs (sensor columns only) so they load instantly and work offline. Each shows a different failure signature.
Insights
Automatic read of the log: crash signature, PCIe stability, power delivery, and thermals.
Key numbers
Peaks and lows for the sensors that matter most. Colored borders flag values worth a second look.
Time series
Each line is normalized to its own range (0–100% of that sensor's min–max), so different units overlay cleanly. Hover for exact values.
Load distribution
Load zones computed from GPU utilization.
Baseline by load zone
Average / peak for each sensor, split by how hard the GPU was working. Use it to tell "normal under load" from "something's off."
| Sensor | Idle | Low | Medium | High |
|---|
Translate the metrics
What every sensor actually means, what's normal, and what should worry you. Load a log and matching rows get tagged detected with your values.
| Sensor | Plain meaning | Normal & red flags |
|---|
Tool tutorials
Short, no-fluff guides for the tools that produce and confirm these logs. Each ends with the one thing to look for. For the full validation workflow, see the Validate tab.
The validation toolkit
A phased workflow for testing and validating your PC — from capturing the log to an RMA-grade evidence pack. Every tool states what it actually proves, so you know when you're done.
NVIDIA Xid codes worth knowing
When the driver reports a hardware error it logs an Xid code — find them in
Windows Event Viewer (source nvlddmkm) or journalctl -k on Linux.
These are the ones that matter for a crash investigation.
| Code | Meaning | Points to | Notes |
|---|