Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
To check whether a GPU is working, first confirm that Windows detects it and loads a driver; then verify that a real graphics workload uses it, run a repeatable benchmark, and—if needed—perform a supervised stability test. These steps answer different questions: detection does not prove rendering ability, a benchmark measures performance rather than overall health, and a stress-test pass cannot rule out every intermittent fault.
A quick, safe GPU check
- Check detection: Open Device Manager and confirm the expected GPU appears under Display adapters without a device error.
- Check driver information: Press Win+R, enter
dxdiag, and review the Display or Render tab. - Check activity: Open Task Manager with Ctrl+Shift+Esc, select Performance, and watch the relevant GPU while running a game or other graphics application.
- Check rendering and performance: Run a game’s built-in benchmark or a repeatable graphics benchmark with recorded settings.
- Check stability only if needed: If the system crashes, shows artifacts, or has recently been overclocked or upgraded, run a supervised test at stock settings.
Do not begin with an extreme stress test. If the GPU is missing, the driver is failing, or the card has a power or cooling problem, a heavy workload adds risk without answering the first diagnostic questions.
What “working” means
- Detected: Windows can identify the device. This is a basic check, not proof that it can render correctly.
- Rendering: The GPU can draw a desktop, video, or 3D workload without immediate failure.
- Stable under a test: It completes a particular workload without artifacts, driver resets, crashes, freezes, or shutdowns.
- Performing normally: Its result is reasonably close to comparable systems using the same GPU class and similar settings.
- Fully healthy: This is a much stronger conclusion. No single benchmark can rule out intermittent faults, marginal memory, connector trouble, or failures that appear only when the card is hot or under a particular workload.
Before testing: check the connection, power, and GPU selection
For a desktop with a discrete graphics card, connect the monitor to the ports on the graphics card, not the motherboard video output. Check that the card is fully seated in its PCIe slot and that every required GPU power connector is attached. On a modular power supply, use the correct cables for that PSU; do not substitute questionable adapters. Make sure the fans are unobstructed and inspect the card for visible damage. For a new build, also check the monitor’s selected input and try a known-good display cable.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →On a laptop, connect AC power and choose an appropriate performance mode before benchmarking. Many laptops have both integrated and discrete graphics, and Windows may assign different applications to different GPUs. Activity on the integrated GPU alone does not mean the discrete GPU has failed. See Microsoft’s explanation of integrated and discrete GPUs.
#1 Best Overall
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5080
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Check whether Windows detects the GPU
Device Manager
- Right-click Start and choose Device Manager.
- Expand Display adapters.
- Confirm that the expected GPU model appears. Open it and check Properties > Device status.
A correct model with no warning is encouraging, but it establishes detection only. A Microsoft Basic Display Adapter entry generally means Windows is using a generic display driver; install a suitable driver from the GPU maker or, where appropriate, the PC or laptop manufacturer.
A yellow warning icon or Code 43 means Windows has reported a problem with the device. It can be caused by software or a driver issue as well as hardware trouble; it does not, by itself, prove that the GPU is dead. Microsoft’s Code 43 guidance includes driver troubleshooting.
If the GPU is absent, investigate power, seating, the PCIe slot, BIOS/UEFI configuration, and whether the device has been disabled before assuming hardware failure.
DirectX Diagnostic Tool
Press Win+R, enter dxdiag, and press Enter. Inspect the Display or Render tabs; on a system with multiple adapters, the relevant GPU may be listed on a different tab. Choose Save All Information to create a report containing hardware, driver, and DirectX details. NVIDIA also provides instructions for generating a dxdiag report.
dxdiag is an inventory and compatibility check, not a benchmark or load test. A report that lists the GPU does not establish that it is stable under 3D load.
Confirm that the application is using the intended GPU
- Press Ctrl+Shift+Esc to open Task Manager.
- Select Performance, then choose the GPU you want to check.
- Start a game or GPU-accelerated application and watch the GPU engines and memory use.
Task Manager may show separate engines such as 3D, Copy, Video Decode, or Video Processing. A video playback task may use the decode engine without driving 3D utilization. Windows’ GPU scheduling and memory reporting are described in the DirectX team’s Task Manager overview.
Low utilization is not automatically a fault. A game may be capped at a frame rate, limited by the CPU, waiting on storage, or using a different GPU. If the computer has more than one GPU, check Windows graphics settings and the application’s own settings to confirm which adapter it is assigned to. A laptop’s battery or quiet mode can also restrict performance.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Check the driver before judging a result
Use a driver appropriate for the GPU and system: NVIDIA for GeForce and professional NVIDIA cards, AMD for Radeon, and Intel for supported Intel graphics products. Laptop owners should also check the laptop maker’s support page. Customized system features or validation requirements can make an OEM driver preferable; AMD notes this possibility for laptops and all-in-one systems in its driver compatibility guidance.
After installing or changing a driver, restart Windows before testing. Return overclocks and undervolts to default, and temporarily disable overlays if a benchmark crashes. If trouble began immediately after a driver update, trying a known-good earlier driver can help isolate the change. A clean reinstall is one possible step, not a requirement for every problem.
Run a real-world graphics test
A game or application you actually use is a useful first rendering test. Prefer a built-in benchmark because it can replay a defined scene more consistently than ordinary gameplay.
- Choose a documented resolution and graphics preset.
- Record whether ray tracing, upscaling, and frame generation are enabled, along with any frame-rate cap.
- Confirm that the intended GPU is selected.
- Run the test at least twice and note average FPS and, if available, minimum FPS or one-percent lows.
- Record the game version, driver version, resolution, preset, and relevant settings so you can repeat the test.
For a performance comparison, use the same benchmark scene and settings. Microsoft’s Windows testing methodology uses repeatable 3DMark runs and in-game benchmarks while emphasizing that test procedures and versions can change.
A game test can expose a problem that a synthetic benchmark misses, or vice versa. A crash limited to one title may reflect a game, driver, API, or shader issue rather than a defective GPU.
Run a repeatable synthetic benchmark
3DMark is one option for measuring graphics performance and checking whether a system completes a repeatable workload. Its official benchmark page lists available tests and versions; availability changes, so check the current page rather than relying on an old list of tests.
Choose a workload the GPU supports. A modern rasterization test is a reasonable general performance check; use a ray-tracing test only if the GPU supports its required features. For integrated graphics, laptops, or thermally constrained systems, a lighter test may be more appropriate.
Rank #3
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Compare scores fairly
There is no universal score that separates a healthy GPU from a faulty one. Compare results against the same GPU model and, where possible, similar CPU, driver, benchmark version, resolution, preset, power limit, and cooling conditions. Laptop GPUs with similar names to desktop models can have very different power limits and performance.
Free tools Windows power users keep installed
One-click scans. No signup required.
A result far below comparable runs is a reason to investigate, not a diagnosis. Check for the wrong GPU, a power-saving profile, background work, thermal throttling, low clocks or power limits, CPU limitation, and driver or tuning problems. A normal score is reassuring for that workload, but does not establish that every GPU function is healthy.
Monitor behavior during a workload
Use the GPU vendor’s software or a suitable monitoring utility to observe values the hardware exposes. Record:
- GPU temperature and, if available, hotspot or junction temperature.
- Utilization and dedicated/shared memory use.
- Core and memory clocks, board power, and fan speed.
- Frame rate and frame-time consistency.
- Artifacts, application crashes, driver resets, freezes, or unexpected shutdowns.
Do not apply one temperature cutoff to every GPU. Limits and normal behavior vary by model, sensor, firmware, cooler, laptop design, and ambient conditions. Consult the manufacturer’s specifications for that exact product and watch whether temperatures coincide with falling clocks or instability.
Run a stability test only when appropriate
A benchmark asks, “How fast was this GPU in this workload?” A stress test asks whether it can sustain a demanding workload without failing. Use a stability test after basic checks—particularly if crashes or artifacts occur under load, after a hardware change, or while validating an overclock. Supervise it and use stock settings first.
AMD Radeon: Adrenalin stress test
AMD’s documented procedure uses the tuning controls in AMD Software: Adrenalin Edition. Open the performance or tuning section, select the GPU if the system has more than one, choose a duration, and start the stress test. Watch temperatures, clocks, power, and stability. The referenced AMD procedure documents a 60-second default; interface labels and behavior can vary by software release. AMD says a crash or reboot during its procedure resets tuning settings to defaults. See AMD’s stress-test instructions.
A short pass is a quick stability indication, not a comprehensive burn-in or VRAM validation.
Rank #4
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- Phase-change GPU thermal pad helps ensure optimal heat transfer, lowering GPU temperatures for enhanced performance and reliability
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
- 0dB technology lets you enjoy light gaming in relative silence
FurMark and other unusually heavy tests
FurMark can impose an unusually high power draw compared with many real applications. NVIDIA warns that it may drive a GeForce GPU beyond typical real-world loads and trigger thermal or over-current protection; see its guidance on FurMark and stress tests.
Do not use an extreme test as the first check on a system with unknown power cabling or suspected cooling trouble. Keep any test supervised. Stop if you see severe artifacts, smoke, an unusual electrical smell, abnormal fan behavior, or rapidly escalating temperatures. A FurMark failure is not proof of a defective GPU—it may expose a power or cooling limit—and a pass does not guarantee stability in every game or compute workload.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Investigate artifacts and possible memory faults
Colored blocks, flashing polygons, corrupted textures, or flickering output can point to a graphics-memory or core-stability problem, but they can also come from an unstable overclock, display cable, monitor, driver, or application. Try another cable and display, return the GPU to stock settings, and see whether the issue occurs in more than one application or API. If available for the GPU and system, use an appropriate memory-focused diagnostic rather than treating an ordinary FPS benchmark as a VRAM test.
Note where the artifacts occur. Problems visible in BIOS or before Windows starts are less likely to be caused by a Windows game setting and warrant checking the display path, card, and firmware. Artifacts limited to one game may instead be game-, shader-, or driver-specific. A photo or screenshot can help establish whether the issue is visible in the rendered image or only on the physical display.
Interpret common symptoms
| Symptom | Possible causes | Useful next check |
|---|---|---|
| GPU missing from Device Manager | Power or seating issue, BIOS/PCIe configuration, slot, motherboard, or GPU fault | Shut down and inspect power and seating; check BIOS/UEFI; try another compatible slot or system if available. |
| Code 43 or another device error | Driver/software problem or hardware/configuration issue | Update or roll back the driver, restart, and retest before drawing a hardware conclusion. |
| Desktop works, but games crash | Driver, temperature, power delivery, VRAM, unstable tuning, or game-specific fault | Return to stock settings; test more than one workload while monitoring temperature and clocks. |
| Colored blocks or corrupted textures | Memory or core instability, overclock, cable/display, driver, or application | Try another display path, stock settings, another application, and a memory-focused test if available. |
| Black screen under load | Power or thermal protection, driver timeout, display path, or GPU fault | Check cabling, cooling, and monitored behavior; retry at stock settings with a less extreme workload. |
| Benchmark score is very low | Wrong GPU, power-saving mode, thermal throttling, CPU limit, background load, or driver/tuning issue | Verify adapter, settings, clocks, temperatures, and power mode; compare only with equivalent runs. |
| High utilization but low FPS | GPU limit, low clocks, thermal or power restriction, or demanding settings | Check clocks, temperature, power, and frame times during the workload. |
| Low utilization and low FPS | CPU limit, frame cap, wrong GPU, engine limit, or waiting on another component | Check CPU load, caps, and application GPU assignment. |
| Stress test fails but games work | Extreme test load, power or cooling limit, unstable tuning, or test-specific issue | Return to stock, check cooling and power, and compare with a less extreme real-world workload. |
| Benchmark passes, but one game fails | Game-, driver-, API-, or shader-specific issue; workload-specific memory fault remains possible | Test another game and driver/API path; do not treat one benchmark pass as a universal clearance. |
If a test fails: a sensible recovery sequence
If the GPU is not detected
- Shut down and disconnect system power before touching internal components.
- Reseat the card and reconnect its required power cables.
- If available, try another PCIe slot and check relevant BIOS/UEFI settings.
- Use integrated graphics, if available, to help isolate display and system problems.
- Test the GPU in another compatible system, or try a known-good GPU in the original system, if practical.
If the driver or device status is failing
- Record the current driver version and the exact device status or error.
- Install a suitable current vendor or OEM driver and restart. If the problem began after an update, test a known-good earlier version.
- Remove unstable overclock or undervolt profiles.
- Repeat the original workload and check Device Manager and
dxdiagagain.
If the system crashes during a load test
- Stop testing and return clocks, voltage, and power settings to stock.
- Check fan operation, airflow, temperature behavior, PSU cabling, and power supply suitability.
- Try a less extreme game benchmark and another suitable driver.
- Review Windows events for display-driver or system errors around the crash.
- If the problem persists at stock settings across workloads, isolate the card in a known-good system or contact the PC/GPU vendor.
One crash, one low score, or Code 43 alone is not enough to declare a GPU defective. Repeated failures across drivers and workloads, after eliminating configuration, power, cooling, and system causes, are stronger evidence.
When the GPU is probably working normally
Confidence is reasonable when Windows detects the right device, the appropriate driver loads without an error, a real 3D workload uses the intended GPU without artifacts or crashes, temperatures and clocks behave plausibly for that model, and repeatable benchmark results are in range for comparable systems. A supervised stability test at stock settings can add evidence, especially when investigating load-related faults. For an intermittent issue, test more than one workload and for long enough to reproduce the conditions that trigger it.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesFor supported NVIDIA workstations, servers, and compute systems, NVIDIA’s Data Center GPU Manager (DCGM) offers more targeted diagnostics than a consumer stress utility. Its tools can include software deployment, compute, memory, PCIe, bandwidth, and stress checks, depending on GPU support, DCGM release, installed components, permissions, and plugins. Example commands from the documentation include:
dcgmi diag --run 1dcgmi diag -r 3 -p diagnostic.test_duration=300.0These are not ordinary Windows gaming-PC instructions; use them only in an appropriate supported environment and consult NVIDIA’s DCGM diagnostics guide and command reference. A diagnostic failure identifies a failed check or condition, not automatically a defective GPU; software setup and execution problems can also cause failures.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
