iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
A GPU is probably working properly when Windows detects the correct model and manufacturer driver, applications use the intended GPU, images render cleanly, the system remains stable under a real workload, and temperature, clocks, power, and memory behavior are reasonable for that exact model.
There is no single GPU health test that proves a graphics card is perfect. A stress test can miss faults in a particular game, API, memory region, output port, or power state. Likewise, a failed test does not automatically mean the GPU chip is defective: drivers, overclocking, cooling, the PSU, RAM, motherboard, PCIe connection, and even the monitor can produce similar symptoms.
Quick answer: the five-minute GPU check
- Right-click Start, open Device Manager, expand Display adapters, and confirm that the expected GPU model appears.
- Open the GPU’s properties and check Device status. Look for a yellow warning icon or an error such as Code 43.
- Press Win + R, enter
dxdiag.exe, and inspect every Display tab for the GPU name, driver, display memory, and errors. - Open Task Manager > Performance, start a known 3D application, and confirm that the intended GPU’s 3D or compute activity rises.
- Run a game or other demanding application at stock settings. Watch for artifacts, black screens, crashes, driver recovery, extreme temperatures, or clocks and power that fall unexpectedly.
- If the problem repeats, inspect Event Viewer > Windows Logs > System for display-driver recovery events, WHEA-Logger entries, PCIe errors, or NVIDIA Xid messages.
This sequence separates basic detection and configuration problems from problems that occur only under load. It is more useful than judging the card by one temperature, one benchmark score, or one utilization percentage.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →What a healthy GPU should do
A healthy result is observable rather than absolute. You should normally see most or all of the following:
#1 Best Overall
- Powered by Radeon RX 9070 XT
- WINDFORCE Cooling System
- Hawk Fan
- Server-grade Thermal Conductive Gel
- RGB Lighting
- The operating system identifies the intended GPU model, rather than only a generic display adapter.
- The appropriate AMD, NVIDIA, or Intel display driver is installed and active.
- Device Manager has no persistent warning or error code.
- The desktop, video playback, and ordinary 2D acceleration work without flicker or corruption.
- A 3D application renders a clean image at the selected resolution and refresh rate.
- GPU activity increases when an uncapped, GPU-demanding workload is running.
- Temperature, fan behavior, clocks, power, and VRAM usage respond plausibly for that card and its cooling design.
- The system remains stable at default GPU settings.
- There are no recurring display-driver resets, WHEA hardware errors, NVIDIA Xid errors, or equivalent hardware-path errors.
A healthy GPU does not have to run at 100% utilization, reach a particular clock speed, remain below a universal temperature number, or produce a particular online benchmark score. Those values depend on the game, resolution, CPU, driver, BIOS, power limit, ambient temperature, cooling design, and system configuration.
Signs that a GPU may be failing
Symptoms are clues, not diagnoses. The most convincing evidence is a problem that is repeatable at stock settings, occurs in more than one application, and follows the graphics card when it is tested in another known-good system.
| Symptom | Possible causes | What it proves |
|---|---|---|
| Colored blocks, checkerboards, flashing polygons, or corrupted textures | GPU or VRAM fault, unstable overclock, overheating, driver problem, cable, or monitor | A strong warning when reproducible at stock settings across applications, but not proof by itself |
| Black screen during a game or stress test | Driver reset, PSU or power-connector problem, overheating, cable or output fault, PCIe connection, or GPU failure | Requires isolation; do not immediately replace the card |
| Display driver stopped responding and recovered | Driver, overclocking, cooling, power, compatibility, or defective hardware | Evidence of a graphics-stack timeout, not automatic proof of a dead GPU |
| Game crashes only in one title | Game bug, shader cache, graphics API, driver conflict, overlay, or application setting | Weak evidence of a GPU fault unless the symptom appears elsewhere |
| Sudden frame-rate drops | Thermal or power throttling, CPU bottleneck, frame cap, background process, VRAM pressure, driver, or game update | Needs telemetry and comparison before a hardware diagnosis |
| GPU is absent from Device Manager | Power, seating, PCIe slot, BIOS, driver, motherboard, or failed card | A detection problem, not necessarily a failed GPU |
| Microsoft Basic Display Adapter is listed | Manufacturer driver is missing or not active | Confirms that Windows is using a generic driver; it does not prove the card is dead |
| Fans do not spin at idle | Normal zero-RPM behavior on many modern cards | Not a failure unless the fans also fail to respond when the GPU becomes hot or heavily loaded |
| Coil whine | Electrical noise at particular frame rates or power states | Usually not evidence of malfunction, although a sudden or unusually loud change should be documented |
| Low GPU utilization | CPU bottleneck, V-Sync, frame cap, low settings, lightweight workload, wrong GPU selection, or power-saving mode | Not evidence that the GPU is broken |
Windows uses Timeout Detection and Recovery, or TDR, to reset a GPU when a task exceeds the default two-second timeout. The screen may flicker and Windows may report that the display driver stopped responding and recovered. Microsoft lists driver problems, overclocking, cooling, power, compatibility, and defective components as possible causes. See Microsoft’s TDR documentation and VIDEO_TDR_TIMEOUT_DETECTED guidance.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallWhen recurring Windows errors remain after the hardware checks, Outbyte PC Repair can be an optional tool for looking for broader system issues before you blame the GPU.
Check whether Windows detects the GPU
Use Device Manager
- Right-click Start and select Device Manager.
- Expand Display adapters.
- Confirm that the expected GPU model appears.
- Double-click the entry and inspect the General tab and Device status.
- Look for a yellow warning icon or an error code.
Common codes include:
- Code 43: A driver controlling the device reported that the device failed in some way. Microsoft recommends uninstalling the device, using Action > Scan for hardware changes, and consulting the hardware manufacturer if the problem remains.
- Code 45: Windows believes the device is no longer connected. On a desktop, check seating, power, the PCIe slot, and the motherboard. On a laptop, a firmware, driver, or hardware service issue may be involved.
- Code 28: Drivers are not installed.
- Code 31: Windows cannot load the required drivers.
Microsoft maintains a list of Device Manager error codes and their meanings. The message This device is working properly is useful, but limited: it means Device Manager has no current error. It does not test VRAM, output quality, cooling, power delivery, or stability under a 3D workload.
Understand Microsoft Basic Display Adapter
If Device Manager shows Microsoft Basic Display Adapter instead of the AMD, NVIDIA, or Intel model, Windows is using its generic display driver. The physical GPU may still be functional, but manufacturer-specific acceleration, performance features, power management, and some display capabilities may not be available.
When the manufacturer driver is missing or incompatible, Outbyte Driver Updater can be an optional way to identify and update Windows drivers before deeper GPU hardware testing.
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 glitchesInstall the correct driver from the GPU manufacturer. For laptops, check the laptop manufacturer first because its driver may include custom display switching, power-management, or firmware integration. Microsoft explains this situation in its guide to the Microsoft Basic Display Adapter.
Collect a DirectX Diagnostic report
- Press Win + R.
- Enter
dxdiag.exeand press Enter. - Inspect each Display tab.
- Record the GPU name, driver model, driver version and date, display-memory information, and any notes or errors.
- Select Save All Information… to create a report for a support request or troubleshooting record.
Multiple Display tabs are normal on systems with integrated and discrete graphics. Microsoft’s dxdiag instructions explain how to create the report. A clean report confirms that Windows and DirectX can describe the adapter; it is not a complete stability test.
Check whether the correct GPU is being used
Use Windows Task Manager
- Press Ctrl + Shift + Esc to open Task Manager.
- Choose Performance.
- Select each GPU entry and identify which is integrated and which is discrete.
- Start a known GPU-heavy game, renderer, or other 3D application.
- Watch the GPU’s 3D, Compute, Copy, Video Decode, or Video Processing activity, along with dedicated and shared GPU memory.
On the Processes tab, right-click a column heading and enable GPU and GPU engine if they are not visible. The GPU engine column can indicate which adapter and engine an application is using.
Do not treat GPU usage as one universal measurement. Windows reports separate GPU engines and represents overall GPU use using the busiest engine rather than averaging every engine. Dedicated GPU memory and shared GPU memory also mean different things. Shared memory is expected on integrated graphics; using some shared memory does not indicate a failing discrete card. Microsoft explains these details in GPUs in Task Manager.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Interpret low utilization correctly
Low utilization with low frame rates can mean that the CPU, rather than the GPU, is limiting the workload. Other possibilities include V-Sync, a frame-rate cap, low graphics settings, a lightweight scene, background activity, a laptop’s battery-saving mode, or the application using the integrated GPU.
For a useful test, temporarily use a known GPU-demanding workload with a frame rate that is not capped, then compare GPU activity with CPU activity, frame time, clocks, temperature, and power. Microsoft’s explanation of CPU- and GPU-bounded workloads describes why a powerful GPU can remain underutilized without being faulty.
Check hybrid graphics on NVIDIA laptops
A hybrid laptop can use the integrated GPU to drive the desktop while the discrete NVIDIA GPU renders a game. That arrangement is normal and can make Task Manager’s entries confusing.
To see which programs and displays are using the NVIDIA GPU:
Recommended Free Tools
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
- Open NVIDIA Control Panel.
- Open the Desktop menu.
- Enable Display GPU Activity Icon in Notification Area.
- Click the tray icon and inspect the active programs and displays.
Windows Graphics settings and the laptop manufacturer’s performance mode can also influence GPU assignment. NVIDIA documents the activity icon and hybrid behavior in its Desktop menu documentation and Optimus guidance.
Monitor temperature, clocks, power, and VRAM
Telemetry helps explain what happens during a failure. Ideally, record the values before, during, and after the symptom rather than looking only at an idle temperature.
Available monitoring options
- Windows Task Manager: Useful for utilization, engine activity, dedicated memory, and shared memory. Temperature reporting depends on the hardware and Windows version.
- AMD Software: Adrenalin Edition: Its performance metrics can include GPU utilization, clock speed, board power, GPU temperature, junction temperature, fan speed, GPU-memory utilization, frame rate, and frame time. AMD documents the default performance-overlay shortcut as
Ctrl + Shift + O, and its software can log metrics where supported. See AMD performance metrics documentation. - NVIDIA: The
nvidia-smiutility can show detected GPUs, driver state, utilization, temperature, power, clocks, and memory on supported installations. - Third-party utilities: HWiNFO, GPU-Z, and similar tools can expose additional sensors. Treat readings carefully because sensor names and availability vary by GPU model, driver, firmware, and utility.
NVIDIA monitoring commands
Open Command Prompt or PowerShell on a system with a working NVIDIA driver and run:
nvidia-smiThis provides a one-time summary of detected NVIDIA GPUs and driver state.
nvidia-smi -qThis requests detailed information for all detected GPUs.
nvidia-smi -q -d UTILIZATION,TEMPERATURE,POWER,CLOCKThis focuses the report on utilization, temperature, power, and clock sections where those fields are supported.
nvidia-smi -l 1This repeats the report every second until you press Ctrl + C. NVIDIA notes that unsupported fields can appear as N/A. The full option and field reference is in the NVIDIA nvidia-smi documentation.
There is no universal safe GPU temperature
Do not use a rule such as anything above 85°C means the GPU is failing. Thermal limits differ by exact GPU model, cooler, BIOS, workload, ambient temperature, and sensor. NVIDIA distinguishes values such as current, slowdown, maximum operating, target, shutdown, and memory temperature, and not every product reports every sensor. AMD software may expose both core and junction or hotspot temperature, but availability varies by product.
Use this process instead:
- Identify the exact GPU model.
- Check the manufacturer’s product documentation for its published thermal limits and sensor definitions.
- Record the maximum temperature during a representative workload.
- Look for clock reduction, fan saturation, crashes, black screens, or repeated throttling.
- Compare core, junction or hotspot, memory temperature, and ambient temperature when those readings are available.
A high temperature without instability may be normal for that model. Conversely, a temperature below a generic threshold does not prove that the card is healthy. NVIDIA’s temperature-field documentation and AMD’s GPU tuning guidance explain why model-specific readings matter.
Understand fans, power, and VRAM
Many graphics cards intentionally stop their fans at idle. This zero-RPM mode is normal when the card is cool; what matters is whether the fans start and respond as temperature and load rise. AMD documents this behavior in its Radeon fan and temperature guidance.
High VRAM usage is not automatically an error. Games may use available memory for textures and caches. However, heavy use of shared system memory can indicate that graphics data is spilling into system RAM, which may reduce performance. Watch for errors, stutter, or crashes rather than treating a high memory percentage alone as proof of failure.
Test the GPU under load safely
Use a staged test. Start with a real workload that reproduces the problem instead of immediately running the most aggressive synthetic stress test.
Stage A: Restore a known baseline
Before testing:
- Disable GPU overclocking and undervolting.
- Restore vendor tuning to default.
- Remove experimental fan curves.
- Close unnecessary overlays and monitoring tools.
- Record the display-driver version and current BIOS or firmware state.
- Ensure that the case or laptop has unobstructed airflow.
Test at stock settings first. AMD’s system-stability guidance also recommends checking the card in a stock configuration, verifying cooling, installing the driver cleanly, confirming seating and direct PSU connections, and removing PCIe risers during diagnosis.
Rank #3
- 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
Stage B: Run a real game or application
- Choose a demanding game, renderer, compute workload, or application that normally causes the problem.
- Use known graphics settings and resolution.
- Use an uncapped frame rate if the goal is to create sustained GPU load.
- Monitor GPU usage, engine activity, frame time, temperature, clocks, power, and VRAM.
- Run long enough to reproduce the original symptom, but stop if the system shows dangerous or severe behavior.
- Repeat with a second application, preferably one using a different graphics API.
Useful interpretations include:
- High GPU activity and stable, clean rendering: evidence that the GPU is doing work normally in that workload.
- Low GPU activity and poor frame rate: investigate CPU limitation, frame caps, V-Sync, low settings, power mode, wrong-GPU assignment, or background processes.
- Load followed by a black screen or driver recovery: investigate the driver, power delivery, temperature, overclocking, PCIe connection, and hardware.
- Artifacts in every 3D application at stock settings: substantially stronger evidence of a GPU-core or VRAM problem.
Stage C: Use a dedicated 3D or VRAM test
Run one controlled test at a time and record the settings and result.
AMD Adrenalin stress test
In AMD Software, search for Tuning, open Performance Tuning, and use the built-in stress test where it is available. AMD documents that if the test causes a crash or reboot, the software resets GPU tuning settings to default. A pass is useful evidence, but it does not cover every game, API, memory region, or output path. See AMD’s Performance Tuning documentation.
OCCT 3D and VRAM tests
OCCT provides separate tests for different failure patterns. Use 3D for core and rendering stability and VRAM for memory-focused testing. Its current v17 release was announced as stable on June 30, 2026; future releases may change the interface. The test options are described in OCCT’s stability-testing documentation and v17 release announcement.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →A repeated VRAM-test failure is strong evidence, especially if it occurs at stock settings, but it still should be confirmed after checking the driver, GPU tuning, cooling, PSU, and system memory. A synthetic 3D test can pass while a particular game fails, because different workloads exercise different engines, APIs, memory paths, and power states.
Benchmarks such as 3DMark
A repeatable benchmark is useful for comparing the same system before and after a change. Do not treat one score as a health certificate or apply a universal rule such as being within a fixed percentage of an online average. Compare systems with the same GPU model, CPU, driver branch, BIOS, power limit, resolution, game or benchmark version, memory configuration, cooling, and background tasks.
When to stop testing
Stop a test if you see a burning smell, severe visual corruption, repeated black screens, a system reboot, abnormal mechanical noise, or a temperature approaching the exact model’s documented limit. Do not keep repeating a crash to make the diagnosis more certain; record what happened and move to isolation steps.
Check Windows logs for GPU errors
Event Viewer and WHEA
- Press Win + R.
- Enter
eventvwr.msc. - Open Windows Logs > System.
- Inspect events at the time of the crash or black screen.
- Look for display-driver recovery events, WHEA-Logger events, PCI Express errors, unexpected shutdowns, and driver or kernel errors.
WHEA records hardware-error events in the Windows system event log and can include PCI Express error records. A WHEA entry is evidence of a hardware-path problem, but it does not identify the GPU as the only possible cause. The PCIe slot, motherboard, PSU, another PCIe device, or a connection can also be involved. Microsoft documents WHEA hardware events and WHEA error records.
Free tools Windows power users keep installed
One-click scans. No signup required.
NVIDIA Xid errors
NVIDIA Xid messages appear in the operating system’s kernel or event log. NVIDIA warns that an Xid can indicate a driver, application, hardware, PCIe, or memory problem. Treat the Xid number as a starting point for diagnosis, not as a definitive statement that the GPU chip has failed. See NVIDIA’s introduction to Xid errors.
Use logs to establish a pattern
One isolated event after a forced shutdown is less useful than the same display recovery or hardware error appearing every time a particular load starts. Note the application, driver version, GPU temperature, power behavior, tuning state, and whether the error follows the card or remains with the computer.
Fixes, in the right order
1. Check the display chain first
- Confirm the monitor is set to the correct input.
- On a desktop, connect the display cable to the graphics card rather than the motherboard video output, unless you are deliberately testing integrated graphics.
- Try another cable, another GPU output, and another monitor.
- Remove adapters, docks, and KVM switches temporarily.
- Check the selected resolution and refresh rate.
A monitor, cable, adapter, input selection, or refresh-rate problem can imitate a GPU fault. Microsoft recommends trying another cable, output, monitor, or system when isolating external-display problems; see its external-monitor troubleshooting guide.
2. Reset to default GPU settings
Undo overclocking, undervolting, custom power limits, and aggressive fan curves in the vendor utility. If the card becomes stable at stock settings, the immediate problem is configuration or operating margin rather than proof of permanent GPU failure. Do not use an overclocked result to judge a used or newly installed card.
3. Update, roll back, or reinstall the driver
If the problem began immediately after a driver update, try Device Manager > Display adapters > GPU > Properties > Driver > Roll Back Driver when the option is available. If rollback is unavailable, uninstall and reinstall the display driver using the appropriate vendor package. A clean installation option may be available in the installer.
For a laptop, prefer the OEM driver when the manufacturer states that its customized power management or display switching depends on it. If only one application flickers, also test with overlays, browser hardware acceleration, and game launchers disabled.
Rank #4
- 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
Microsoft recommends checking whether Task Manager flickers along with the rest of the display: if Task Manager flickers too, a display-driver problem is more likely; if Task Manager remains stable while another application flickers, an incompatible application is more likely. See Microsoft’s screen-flicker troubleshooting steps.
4. Improve cooling
- Restore stock settings.
- Confirm that all fans respond under load.
- Clean dust from filters, heatsink fins, and case intakes.
- Improve case airflow and verify that the laptop is not on a surface blocking its vents.
- Check whether a laptop is in a quiet, battery-saving, or low-power mode.
- Compare core, hotspot, and memory readings with the model-specific documentation.
Reapplying thermal paste may be appropriate for some out-of-warranty cards, but it is not a routine first step. Do not replace thermal pads without exact model-specific thickness measurements; incorrect pads can worsen cooling or damage the card. Avoid baking, reflowing, or other unsafe repair methods.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →5. Check seating, PCIe, and GPU power
Shut the computer down completely and disconnect power before opening a desktop case. Check that:
- The GPU is fully seated in the primary PCIe slot.
- Every required GPU power connector is inserted securely.
- There is no visible connector or cable damage.
- Independent PSU cables are used where appropriate instead of improperly sharing one cable or using an unsuitable adapter.
- A PCIe riser cable is removed temporarily during testing.
AMD recommends confirming card seating, using independent PSU cables where applicable, removing a riser during diagnosis, and performing a swap test with a known-good GPU when possible. Its common boot-failure guidance covers these checks.
6. Investigate the PSU and other components
A card can work at the desktop but fail when a game raises its power demand. Investigate PSU wattage and quality, loose connectors, the motherboard BIOS and PCIe configuration, RAM stability, CPU stability, and temperatures. Faulty RAM, motherboard, CPU, storage, and system-memory problems can crash graphics workloads and look like GPU failure; NVIDIA lists these possibilities in its graphics-card troubleshooting guidance.
If the symptoms suggest inadequate power, test with a known-good PSU of equal or greater appropriate wattage. AMD lists insufficient or faulty PSUs as possible causes of crashes, hangs, reboots, display corruption, and boot failures and discusses known-good PSU testing in its PSU troubleshooting guide.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
7. Isolate an application or overlay
If only one game fails, update or repair that game, clear or rebuild its shader cache if the game provides that option, test another graphics API when available, and disable overlays. A single game crash is weak evidence of a GPU fault. If the same artifacts or crashes occur in several unrelated applications, the evidence becomes much stronger.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to tell whether the GPU itself is defective
Use repeatability and cross-testing rather than one symptom. The following decision path is a practical way to classify the result:
- Is the GPU missing or showing a warning? Start with power, seating, PCIe slot, BIOS, cable, driver, and motherboard checks. Do not call the card defective yet.
- Is Microsoft Basic Display Adapter listed? Install the correct manufacturer or OEM driver first.
- Does the intended application use the GPU? Check Task Manager’s GPU engine and the laptop’s hybrid-graphics settings before diagnosing underuse.
- Does the image remain clean at stock settings? Test another cable, display, output, and application if artifacts or flicker appear.
- Does the problem occur only at high temperature, high power, or an overclock? Investigate cooling, PSU capacity, connectors, and tuning before blaming the GPU silicon.
- Does the problem repeat in several applications and tests? Run a controlled 3D test and, if memory corruption is suspected, a VRAM-focused test.
- Does the failure follow the card? Test the suspect GPU in another known-good system, or test a known-good GPU in the original system. This is among the strongest practical forms of evidence.
Evidence categories
| Evidence strength | Examples | Likely conclusion |
|---|---|---|
| Strong | Artifacts in multiple applications at stock settings; repeated VRAM-test failures; intermittent disappearance from the PCIe bus; the failure follows the card to another known-good system; a known-good GPU works in the original system; persistent Code 43 after driver and physical checks | Probable GPU hardware fault or a fault on the card. RMA or replacement is reasonable after documenting the tests. |
| Moderate | Crashes only under high power or temperature; repeated TDR events; black screens that disappear after reducing clocks or power; substantially poor performance after controlling for system differences | Needs further isolation of cooling, PSU, driver, tuning, motherboard, RAM, and application causes. |
| Weak | One low benchmark score; utilization below 100%; fans stopped at idle; coil whine; one game crash; a generic temperature number without the exact model and sensor type | Not enough evidence to diagnose a defective GPU. |
Do not claim certainty from a single stress-test pass or failure. Different tests exercise different rendering engines, memory regions, APIs, power states, and driver paths. A card may pass a synthetic test and fail a specific game, or fail a test because of unstable system RAM or a PSU problem.
Linux checks
For an NVIDIA Linux system with the proprietary driver installed, run:
nvidia-smiA successful result should identify the GPU and show driver-related information. NVIDIA documents nvidia-smi as a management and monitoring tool for supported Linux distributions and 64-bit Windows systems.
For an AMD system using a supported ROCm installation, run:
rocminfoThis enumerates the ROCm platform and available GPU agents. You can also use:
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
amd-smiwhere supported to monitor AMD GPU health, temperature, and utilization.
Free tools Windows power users keep installed
One-click scans. No signup required.
If ROCm commands fail, check whether the amdgpu driver is loaded, whether /dev/kfd and /dev/dri exist, and whether the user has the required render and video group permissions. These are ROCm-stack checks and do not automatically diagnose every consumer Radeon gaming installation. Consult the ROCm verification guide and rocminfo documentation.
Important edge cases
Integrated graphics
Integrated GPUs normally use system RAM. They may show little or no dedicated GPU memory and some shared GPU memory. That is expected and should not be interpreted as missing VRAM or a defective adapter.
Hybrid laptops
The integrated GPU may drive the display while the discrete GPU renders selected applications. Check both adapters and use Task Manager’s GPU engine information or the vendor’s activity tool.
External GPU enclosures
With an eGPU, the enclosure, Thunderbolt or USB4 link, cable, power adapter, firmware, and PCIe-over-the-link connection are additional failure points. Test the enclosure and cable separately where possible.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC 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 & 11Laptop GPUs
Laptop GPUs are often soldered to the motherboard and cannot be reseated or replaced like desktop cards. Include the OEM driver, firmware, power mode, cooling system, and display-switching behavior in the diagnosis.
Multi-GPU systems
GPU 0 and GPU 1 are operating-system numbering conventions, not quality rankings. Identify each adapter by its model and watch the engine assigned to the application.
Windows version
These instructions are written primarily for supported Windows 11 systems. Many steps remain similar on Windows 10, but Microsoft ended Windows 10 support on October 14, 2025. Labels and available driver options can also vary by Windows version, GPU generation, laptop vendor, and driver package.
Common mistakes when checking a GPU
- Using a universal temperature limit: thermal limits and sensor meanings are model-specific.
- Assuming 100% GPU usage means healthy: utilization shows that an engine is busy; it does not validate VRAM integrity, output stability, cooling, or long-term reliability.
- Assuming low utilization means the GPU is broken: CPU limits, V-Sync, frame caps, low settings, hybrid graphics, and lightweight workloads are common explanations.
- Treating Device Manager as a health certificate: no Device Manager error does not test the card under load.
- Ignoring the display chain: cables, monitor inputs, adapters, refresh rates, and outputs can imitate GPU faults.
- Running an extreme stress test immediately: a staged real-world test is safer and often more diagnostic.
- Using one benchmark as proof: one benchmark exercises only some engines and memory paths and is sensitive to system configuration.
- Blaming every crash on the GPU: PSU, RAM, motherboard, CPU, driver, storage, and application problems can produce similar symptoms.
- Ignoring hybrid graphics: the integrated GPU may be used intentionally for the display or for a power-saving application.
- Opening the card casually: thermal pads are model-specific, and unsafe repair methods such as baking or reflowing can cause further damage.
Final diagnosis checklist
Classify the result into one of these categories:
- Normal: correct detection, correct driver, clean output, plausible telemetry, and stable workloads.
- Configuration or driver problem: generic adapter, Device Manager warning, wrong GPU assignment, or a problem that began after a driver or tuning change.
- Cooling problem: temperature or hotspot rises abnormally for the model, clocks fall, fans do not respond under load, or stability returns after cooling improvements.
- Power or connection problem: failures occur under load, connectors or risers are suspect, or the issue disappears with a known-good PSU or corrected seating.
- Application-specific problem: one title or application fails while other applications and tests remain stable.
- Probable hardware failure: repeatable artifacts, VRAM errors, bus disappearance, or crashes across applications and systems at stock settings, especially when the failure follows the card in a swap test.
Only the last category justifies treating the graphics card as the primary replacement or RMA candidate. Until then, document the exact GPU, driver, settings, temperatures, error logs, applications tested, cables and displays used, and results from any known-good hardware.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Frequently Asked Questions
Can a GPU pass a stress test and still be faulty?
Yes. A stress test covers only the engines, memory paths, API behavior, and power states that it exercises. A GPU can pass one synthetic test but fail in a particular game, output port, driver version, or workload. Repeat the test with a real application and, where appropriate, a separate VRAM-focused test.
Does 100% GPU usage prove that the graphics card is healthy?
No. It only shows that one GPU engine is busy. It does not prove that the GPU is rendering cleanly, that VRAM is error-free, that the temperature is appropriate, or that the card will remain stable under other workloads.
Is a GPU temperature above 85°C automatically dangerous?
No. Temperature limits vary by exact GPU model, cooler, BIOS, sensor, workload, and ambient temperature. Check the manufacturer’s specifications and watch for throttling, fan saturation, crashes, or black screens instead of applying a universal threshold.
Why does my GPU show low usage while a game runs slowly?
The CPU may be limiting the game, or V-Sync, a frame cap, low graphics settings, a laptop power mode, background activity, or incorrect GPU assignment may be responsible. Check Task Manager’s GPU engine, CPU usage, frame time, power mode, and application settings.
When should I RMA or replace the GPU?
Consider an RMA or replacement when the fault is repeatable at stock settings, persists across clean driver states, applications, cables, and displays, and especially when the failure follows the card into another known-good system or a known-good GPU works normally in the original system.
The Bottom Line
The best practical GPU check is a chain of evidence: detect the correct model, load the correct driver, verify application assignment, inspect the rendered image, monitor behavior under a real workload, review logs, and repeat the problem under controlled conditions. Persistent artifacts, repeated VRAM errors, PCIe disappearance, or failures that follow the card are strong evidence of a hardware fault. A single low benchmark score, high temperature without context, stopped idle fans, or low GPU utilization is not.
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.
