How to Investigate GPU Temperature and Suspected Throttling
Use a supported hardware monitor and repeatable workload to correlate temperature, clocks, and FPS without relying on simulated browser temperatures.
A slowing workload can have many causes. To investigate thermal throttling, collect temperature and clock readings from a monitoring tool that supports the exact device. This website cannot obtain those sensor readings through its WebGL tests.
Find a real sensor reading
Operating system and driver monitoring features vary by platform, device, and driver version. Some expose GPU temperature; others do not. If a built-in monitor lacks a reading, do not assume a browser can supply it. Consult the device vendor's documentation for a supported monitoring method.
Core, hotspot, and memory junction readings refer to different sensors. Preserve their labels in your notes. A single generic “temperature” field without its sensor identity is hard to interpret. Manufacturer limits are model-specific; this guide does not define a universal safe threshold.
Observe a fixed workload
Run the application in which the slowdown occurs, or use our shader stress test for a separate browser observation. Save work first. Start modestly, keep airflow clear, and stop for unexpected behavior. Fix window size and stress level during an observation so changing rendering demand does not masquerade as a temperature effect.
Record simultaneous observations:
Time in run:
Workload and settings:
FPS:
Core temperature and sensor source:
Hotspot or memory temperature, if available:
Clock frequency and power mode:
Other running applications:
This is a recording template, not a fabricated sample trace. Missing readings should stay missing.
Look for correlation before claiming a cause
A repeated FPS decline alongside a reported clock reduction and temperature limit is evidence worth investigating. FPS decline alone can also result from background tasks, battery policy, browser scheduling, or a scene change. Repeat the same conditions and consult the monitor's description of its reported limits.
Make one reversible change
Check external ventilation and close competing tasks. Compare the same workload after a cooling interval. Record power mode changes separately. Avoid increasing voltage or dismantling cooling hardware based solely on a browser FPS drop. Persistent shutdowns or sensor warnings need investigation through the manufacturer's support guidance.
Our methodology explains what browser FPS measures. The WebGL API documentation describes the graphics interface used by the site.