How to Test a Failing GPU & Diagnose Graphics Card Artifacts or Black Screens
This guide provides comprehensive, multi-stage methods to isolate whether graphical artifacts, flickering, or black screens are caused by the GPU hardware, drivers, or software conflicts.
Direct Diagnostic & Remediation Summary
- Action 1: Driver Clean Reinstallation: Use DDU in Safe Mode to completely wipe old drivers before installing the latest version from NVIDIA/AMD.
- Action 2: Stress Testing (GPU): Run a demanding benchmark like FurMark or 3DMark for an extended period while monitoring core temperatures and clock speeds via HWMonitor.
- Action 3: Physical Check: Reseat the GPU in its PCIe slot, ensuring all power connectors are firmly attached to both the card and the PSU cables.
Initial Triage & Environment Setup
Before assuming hardware failure, we must eliminate driver corruption and basic connection issues. This step focuses on the software environment.
First, download Display Driver Uninstaller (DDU) onto a separate partition or USB drive. Running DDU in Safe Mode is critical for ensuring no residual registry entries interfere with testing.
Procedure:
- Boot into Safe Mode: Use Windows' advanced startup options to enter Safe Mode.
- Run DDU: Execute DDU and select the option to clean and restart for your specific GPU vendor (NVIDIA, AMD). Do not select 'Clean and Restart'; choose 'Clean and Shutdown' first if you are comfortable manually booting into a clean state later.
- Check Connections: With the system powered down, physically inspect the PCIe slot on the motherboard. Gently remove and reinsert the graphics card, ensuring it clicks securely. Reconnect all necessary power cables (6-pin/8-pin connectors) from the Power Supply Unit (PSU).
Diagnostic Scans & Hardware Isolation
Now that drivers are clean, we test stability under controlled loads. We must differentiate between driver instability and actual hardware failure (e.g., VRAM degradation or core overheating).
If the system passes this stage, the GPU is likely fine; if it fails consistently, the issue points toward the card itself.
Diagnostic Tools & Testing:
- Monitor Temperatures: Install HWMonitor and monitor GPU Temperature (Temp), Core Clock Speed, and Memory Junction Temps.
- Run Stress Test 1 (Synthetic Load): Use a demanding benchmark like FurMark or Unigine Heaven/Superposition for at least 20 minutes. Watch the temperatures closely; if they spike rapidly above 85°C under load, cooling is inadequate.
- Run Stress Test 2 (Real-World Load): Play a graphically intensive game you know works well on this hardware for 1-2 hours. Observe artifacts appearing only during specific complex scenes.
Troubleshooting Artifacts:
- If artifacts appear immediately upon booting into Windows, the problem is likely driver/OS related (revisit Step 3).
- If artifacts only appear after intense gaming sessions and eventually cause a crash, this strongly suggests hardware failure or overheating components on the card.
Targeted Remediation & Fixes
If the stress tests failed, we employ targeted fixes based on the suspected failure point.
- Driver Rollback/Reinstall: If artifacts appear randomly without extreme load, perform a clean reinstall using DDU again. Then, download the driver directly from the GPU manufacturer's official site (NVIDIA GeForce Experience or AMD Adrenalin) rather than relying on Windows Update.
- Power Management Check: Use the GPU control panel software to ensure that power saving modes are disabled during testing. The card must be allowed to draw maximum rated power.
(For monitoring current usage via command line).nvidia-smi -q -i 0 - BIOS/Firmware Check: Verify that your motherboard BIOS is running the latest stable version. Outdated BIOS versions can sometimes cause PCIe lane instability affecting GPU performance.
Pro Tip for Laptops: If this is a laptop, always check the manufacturer's utility software (e.g., Dell Power Manager) to ensure the discrete GPU profile is active and not being throttled by power-saving settings.
Verification, Results & Preventative Maintenance
After completing all fixes (driver clean, reseating, BIOS update), the final validation is crucial. The goal is stability under sustained maximum load without any visible artifacts or unexpected shutdowns.
- Final Stress Test: Re-run FurMark or a similar benchmark for at least 30 minutes. Monitor temperatures via HWMonitor; they should remain below 85°C (ideally lower).
- Artifact Check: Visually inspect the screen constantly during this final test. Any single pixel glitch, color bleeding, or pattern disruption indicates persistent hardware failure that requires replacement.
- Preventative Maintenance Checklist:
- Keep PSU cables securely connected at all times.
- Periodically clean dust from the GPU fans and heatsinks using compressed air (never vacuum cleaners).
- Never run a system with known unstable or outdated drivers for prolonged periods.
nvidia-smi -q -i 0
Frequently Asked Questions
What is the best way to test if my GPU overheats?
Use a benchmark like FurMark for an extended period (20+ minutes) while monitoring core temperatures with HWMonitor. Excessive sustained heat (>85°C) indicates cooling failure or throttling.
Are artifacts always hardware related?
Not necessarily. They can be caused by corrupted display drivers, faulty monitor cables (try swapping HDMI/DisplayPort), or even unstable motherboard BIOS settings.
How do I know if the PSU is failing?
A failing PSU often causes random shutdowns under load across all components, not just the GPU. Test stability with minimal peripherals attached to isolate the PSU's role.
Need help implementing this solution?
Our cybersecurity and infrastructure engineers can audit, harden, or configure your systems directly.
Community Discussion & Confirmations (3)
Technical questions, hardware findings, and verified fix reports.