Updated October 18, 2026: Added support checks, a default-settings baseline and realistic limits for what a completed stress test proves.
On supported Radeon systems, AMD Software: Adrenalin Edition includes a GPU stress test in its performance-tuning area. It is useful for checking a tuning change or reproducing instability, but passing one test does not certify every game, the whole PC or the card’s remaining lifespan.
Check that your installation supports it
Open AMD Software and find Performance > Tuning, or use its search if the interface differs. The available controls depend on the GPU and software installation. A driver-only installation, an unsupported device or an OEM-managed laptop may not show the same tools.
Use the official AMD or computer-manufacturer driver for the system. Do not assume a newer desktop Radeon package is the right choice for every laptop or handheld.
Establish a baseline before tuning
- Save or record any existing custom settings.
- For diagnosis, return GPU tuning to its default state so you are not testing an unknown overclock or undervolt.
- Close unrelated heavy applications and make sure cooling vents are unobstructed.
- Record the driver version and the symptom you want to investigate.
Changing the GPU, CPU and memory settings together makes a failed test difficult to interpret. Keep those variables separate; CPU idle C-states should not be confused with GPU tuning.
Run and observe the test
Select the Stress Test control when available, review its duration/options and start it. Watch the metrics your card exposes, such as GPU temperature, junction temperature, clock speed, fan behavior and power. Stop if you see artifacts, a burning smell, abnormal overheating or other concerning behavior.
There is no single temperature number that diagnoses every Radeon card. Compare readings with the limits and expected behavior for the exact model. Also distinguish a reported GPU sensor from a hotter junction or hotspot sensor.
Interpret the result
| What happened | Useful next step |
|---|---|
| Test completed normally | Try the actual game or workload that caused the problem. One short pass is limited evidence. |
| Crash after custom tuning | Restore defaults and compare. AMD notes that a crash/reboot during its test can reset tuning to default. |
| Crash at defaults | Record the error, logs and temperatures. Investigate drivers, power, cooling and the wider system before blaming one component. |
| Only one application fails | Check that application’s settings and known issues as well as the graphics driver. |
Repeatedly running a failing test at more aggressive settings is not a repair method. Save reproducible evidence, such as GPU-Z sensor logs, and contact the appropriate hardware or system support team when default operation remains unstable.
Images from the original article
These original illustrations are retained for reference. Product examples and software screens may show older versions; use the model-specific instructions above.
