GPU Installed but Not Detected in Windows: PCIe Slot, Power, Firmware, Drivers, and Device Evidence

This workflow begins only after the PC completes POST and reaches a usable Windows desktop. Record whether the display is connected to motherboard graphics or another working adapter, whether the new card appears anywhere in Device Manager, and whether the symptom began immediately after installation, after a driver change, or after firmware or other hardware work.

If the machine cannot complete POST or establish a usable display at all, use the no-POST workflow. If Windows is running and the GPU is present but the monitor intermittently loses signal, use the display-signal-loss workflow instead. Keeping those boundaries separate prevents a Windows enumeration problem from being treated as a generic black-screen failure.

Open Device Manager and inspect Display adapters plus any relevant unknown or problem devices. Record the exact device name, status text, and problem code before changing drivers. Microsoft documents distinct Device Manager states such as disabled hardware, failed start, failed installation, device-not-present, and Code 43; the code is evidence about the Windows device state, not a universal diagnosis of the failed component.

If the GPU is completely absent, use Scan for hardware changes once and then continue with physical, slot, firmware, and platform checks. If it is present with a documented problem code, follow the meaning and resolution for that exact code rather than assuming every warning icon means the same thing.

Shut the PC down, disconnect external power as required by the system documentation, and inspect only user-serviceable connections. Confirm the card is fully seated and mechanically secured, and confirm every auxiliary power connection required by the exact graphics-card documentation uses the PSU manufacturer-approved cable path. Do not open the PSU, probe live connectors, repin cables, or substitute modular PSU cables by connector appearance.

Use the motherboard manual to identify the intended graphics slot. A full-length slot does not guarantee identical electrical wiring, firmware behavior, or resource sharing across every slot. If the manual documents slot-sharing or population restrictions, preserve those exact conditions instead of inventing a generic lane rule.

Consult the exact motherboard or system firmware documentation for PCIe device information, primary-display selection, slot enablement, bifurcation, CSM/UEFI behavior, or platform-specific graphics controls that actually exist on that model. ASUS documents platform-specific cases where firmware graphics settings can change which graphics path is used; that is evidence that firmware behavior is board-specific, not a reason to toggle unrelated settings blindly.

Restore documented defaults only when you understand the consequences for boot, storage, encryption, and other installed hardware. Do not flash firmware merely because a GPU is missing in Windows. A firmware update is justified only by the exact board/system support guidance, compatibility notes, or a known issue relevant to the hardware in question.

A working desktop through integrated graphics proves that Windows and one display path work; it does not prove the discrete GPU has enumerated correctly. Likewise, connecting the monitor to the motherboard can make diagnosis easier on a CPU/platform with usable integrated graphics, but it does not enable or repair the discrete card.

Record which adapter Windows reports and which physical output drives the monitor. On laptops, OEM systems, workstations, and hybrid-graphics platforms, display routing can be system-specific, so use the manufacturer documentation rather than applying desktop assumptions to every device.

When Device Manager lists the intended GPU, note its status and current driver provider/version before making changes. Use Windows Update, the system OEM where applicable, or the exact GPU vendor support path for a supported driver. NVIDIA and AMD both provide official driver installation/support paths; prefer those over third-party driver bundles.

For a documented Device Manager problem such as Code 43, Microsoft states that a driver has reported a device failure and recommends reinstalling the device. That does not by itself prove the GPU silicon is defective. Avoid broad driver-removal scripts, registry edits, and repeated forced installations before preserving the original status and testing a supported driver path.

If the motherboard manual supports another suitable PCIe slot, an alternate-slot test can help separate one slot/path from the card itself, but account for documented lane wiring, sharing, and physical constraints. A successful alternate-slot test narrows the path; it does not automatically prove the original slot is electrically defective.

A known-good compatible GPU in the same system, or the suspect GPU in another known-compatible system, can be stronger isolation evidence when available. Keep power, firmware, slot, and driver conditions documented. Do not buy replacement hardware solely because one Device Manager code or one failed boot appeared.

Preserve Device Manager status, Windows update/driver history, relevant system events, firmware version, motherboard model, GPU model, slot used, and power-cable arrangement. Device Manager problem codes describe configuration or device-stack states; they should be interpreted with the physical and firmware evidence rather than converted into failure probabilities.

If the GPU appears only after a particular reversible change—such as reseating with power removed, using a documented alternate slot, restoring a supported driver, or reverting an unsupported tuning change—record that correlation. Correlation narrows the diagnostic branch but is not a substitute for exact platform documentation or a reproducible known-good comparison.

Escalate with the exact GPU, motherboard/system, PSU, slot, auxiliary-power method, firmware version, Windows version, Device Manager presence/status/code, driver source/version, display path, and results from any documented alternate-slot or known-good comparison. That evidence is substantially more useful than reporting only that the GPU is “not detected.”

Stop if further testing would require live electrical probing, opening the PSU, unsupported adapters, firmware flashing without a documented reason, destructive Windows recovery, or hardware modification. If the GPU remains absent or in a documented failure state across supported configurations, use the board/system/GPU vendor support or a qualified repair path for deeper hardware diagnosis.