Video Card Stability Test Checklist: Identify Crashes, Artifacts, and Overheating

Stress Testing Your GPU: The Best Video Card Stability Test Methods

What a GPU stress test checks

• Stability: detects system crashes, driver failures, or application hangs under sustained load.
• Thermals: reveals overheating and cooling adequacy (throttling behavior).
• Power delivery: shows VRM or PSU limits causing instability.
• Artifacts: visual corruption indicating memory or core errors.
• Performance consistency: measures whether boost clocks and frame rates hold over time.

When to run a stress test

• After overclocking or undervolting.
• When diagnosing crashes, freezes, driver crashes, or visual glitches.
• After installing a new GPU, drivers, or PSU.
• Before long sessions of gaming or GPU compute work.

Preparation (quick checklist)

1. Update drivers to the latest stable release.
2. Monitor temps/power/voltages with tools like HWInfo, GPU-Z, or manufacturer software.
3. Close background apps and set power plan to high performance.
4. Ensure adequate cooling (case fans, airflow).
5. Record baseline temps and clock behavior during light use.

Recommended stress-test tools (what they target)

• FurMark — extreme GPU shader stress; good at provoking thermal/power limits but can be unrealistically harsh.
• Unigine Superposition / Heaven — GPU rendering workloads that balance realism and load; useful for artifact checks and sustained load.
• 3DMark (Time Spy/Fire Strike) — benchmark suites with stressful, repeatable tests and score-based results.
• OCCT (GPU:3D or Power) — offers error checking, VRAM tests, and logging; good for diagnosing instability sources.
• MemTestG80 / MemTestCL — VRAM-specific tests for memory errors (useful for GPGPU or suspected VRAM faults).
• Games or long-play loops — real-world stability under target workload.

How to run an effective test (step-by-step)

1. Start with a moderately long run: 30–60 minutes using Unigine or 3DMark to observe temps and clocks.
2. Watch for artifacts, driver resets, or crashes. Note time and conditions of failures.
3. If stable, run a longer session (3–6 hours) or loop a game/benchmark to confirm endurance.
4. Use FurMark or OCCT for targeted thermal/power stress only if you need to find limits — stop early if temps exceed safe limits.
5. If artifacts or errors appear, reduce clocks/raise voltages (or do the reverse if overheating) and retest.
6. For VRAM concerns, run MemTestCL or dedicated VRAM tests for several passes.

Interpreting results

• Short crashes or driver resets often indicate power delivery or driver issues.
• Visual artifacts (streaks, blocks, colour corruption) point to GPU core or VRAM faults.
• Rapid thermal throttling or high sustained temps mean cooling is insufficient.
• Consistent performance drops over time suggest thermal throttling or power/VRM overheating.

Safety and limits

• Keep GPU temps below manufacturer-recommended maximum (typically <85–95°C depending on model).
• Avoid prolonged FurMark runs on laptops or poorly cooled systems.
• Monitor PSU capacities; stress tests draw near-peak power.

Quick troubleshooting actions

• Update or roll back drivers.
• Increase fan curve or improve case airflow.
• Lower core/memory clocks or reduce voltage.
• Test with a known-good PSU or another system.
• RMA the card if persistent artifacts occur after exhaustive testing.

Summary checklist

• Run baseline 30–60 min Unigine/3DMark.
• Monitor temps/clocks/artifacts continuously.
• Use OCCT/FurMark for targeted power/thermal stress if needed.
• Test VRAM separately when memory errors suspected.
• Adjust cooling, power, or clocking based on findings; repeat until stable.