Owen Bradley Few moments in PC building are as deflating as pressing the power button on a fresh install and finding that your graphics card is not detected. The fans spin, the RGB glows, and yet Windows insists you are running on integrated graphics, or the display stays stubbornly black. A GPU that is physically fine almost always comes back once you work through the causes in the right order. This guide walks through that order, from the cheapest checks to the ones needing a screwdriver, so the card shows up again in Device Manager as quickly as possible.
What “Not Detected” Actually Means
The phrase covers several symptoms, and identifying yours narrows the search immediately. If the monitor shows nothing at all, the system may not be posting, which points at power or seating. If you get a picture only from the motherboard video output, the card is present electrically but not initialising or not selected as the primary display. If Windows boots normally but Device Manager shows a generic basic display adapter or a yellow exclamation mark, the hardware is seen and the problem lives in drivers or firmware.
Note which of those three you are seeing before touching anything. It saves you disassembling a machine when a driver reinstall would have fixed it.
Tools and Preparation
You do not need much, but having it ready avoids stopping halfway through with the case open.
- A Phillips screwdriver, ideally magnetic, for expansion slot brackets and side panels.
- A spare known-good display cable, plus an adapter if your monitor and card use different connectors.
- Compressed air or a soft brush for clearing dust from the slot and contacts.
- An anti-static wrist strap, or at minimum the habit of touching bare case metal before handling components.
- A second machine or a phone with data, so you can download drivers if the affected PC has no working display.
Before opening anything, shut the system down fully, switch off the power supply at its rear rocker, unplug the mains cable, and hold the case power button for five seconds to drain residual charge. Modern boards keep standby power live even when the machine appears off, and installing a card into a live slot is a genuine way to kill it.
Step-by-Step Fixes in the Right Order
- Check the display cable and input first. Plug the monitor directly into the graphics card outputs, not the motherboard ones, and confirm the monitor is switched to that input. A surprising share of “dead GPU” reports are a monitor sitting on the wrong HDMI channel. Try a different cable too, since DisplayPort cables in particular fail more often than people expect.
- Confirm the PCIe power cables are fully seated. Most discrete cards need one or more supplementary connectors from the power supply. Push each one in until it clicks and the retention clip engages. Use separate cables from the PSU where possible rather than daisy-chaining two connectors off one lead, especially on higher-wattage cards. If your card uses a newer 12-pin style plug, seat it firmly and check that the connector sits flush with no visible gap.
- Reseat the card in its slot. Remove the retention screw, release the PCIe slot latch at the rear of the slot, lift the card straight out, then reinsert it firmly until the latch clicks shut. Cards can work loose during shipping or when a heavy cooler sags over time. While the card is out, blow dust from the slot and inspect the gold contacts for grime or corrosion.
- Try a different PCIe slot. If the board has a second full-length slot, move the card there. If it works, the primary slot or its lanes are faulty, which is worth knowing before you spend money. Note that secondary slots often run at reduced lane counts, so treat this as a diagnostic rather than a permanent home.
- Clear the CMOS. A bad overclock, a stale memory profile, or a corrupted firmware setting can stop a card initialising. Remove the coin cell battery for a minute with the system unplugged, or short the clear-CMOS jumper per your manual, then reboot. This resets the board to defaults and resolves a real percentage of detection failures on its own.
- Check the BIOS settings. Enter firmware setup and look for the primary display or initial display output option. Set it to PCIe or PEG rather than auto or integrated. While you are there, confirm the relevant slot is set to auto or Gen 4 rather than disabled, and if the board offers a Gen speed override, try dropping it one generation to rule out a signalling issue with a riser cable.
- Enable or disable Above 4G Decoding and Resizable BAR. These related settings control how much memory space the system maps for the card. Some older boards and some cards misbehave when they are on; others require them. Toggle them one at a time and reboot between changes so you know which one mattered.
- Update the motherboard BIOS. A board released before your card’s generation may simply not know how to initialise it. Download the newest firmware from the board maker using another machine, follow their flashing procedure exactly, and do not interrupt the process. Many boards now support flashing from a USB stick without a CPU or GPU installed, which is invaluable here.
- Clean install the graphics drivers. If Windows boots but shows a basic adapter, uninstall the existing driver completely, ideally in safe mode with a removal utility, restart, then install the latest package downloaded directly from the GPU vendor. Avoid layering a new driver over a broken one, since leftover files are a common cause of a card that appears with an error code.
- Test the card and slot separately. If it still fails, put the card in another PC, or put a known-working card in your PC. This single test tells you whether you are dealing with a dead card, a dead slot, or an inadequate power supply, and it turns guesswork into a decision.
When the Power Supply Is the Real Culprit
Detection failures under load, random black screens, and cards that appear only intermittently often trace back to the PSU rather than the GPU. An old, low quality or undersized unit can deliver enough power to post but not enough to keep the card stable once drivers load and clocks ramp up. Compare the wattage recommendation in your card’s documentation against your unit with headroom rather than exactly at the limit, since transient spikes briefly draw well above the rated figure and trip protection circuits on marginal supplies.
Cable quality matters as much as wattage. Never mix modular cables between different power supply models, since pinouts differ and the result can be a fried card. If you have been running a high-draw card on a unit bought years ago for a much smaller build, compare your options among the best PSUs for high-end GPUs before you assume the graphics card itself has failed.
Motherboard and Slot Problems
If the card works flawlessly in another machine, attention turns to the board. Bent pins in the CPU socket can disable PCIe lanes entirely, since those lanes come from the processor on most platforms, so inspect the socket under bright light. Slot damage, cracked solder joints from a heavy sagging card, or debris in the contacts are all worth ruling out visually.
Boards also differ in how they share lanes. Populating certain M.2 slots can disable or halve the lanes going to a full-length slot on some models, which occasionally causes a card to vanish after a storage upgrade. The block diagram in your manual will tell you. If you are shopping for a replacement with cleaner lane allocation and more robust slots, the best motherboards for gaming and the wider selection of best ATX motherboards both cover boards with reinforced primary slots.
Safety Notes Worth Repeating
Ground yourself before handling any card and hold it by the bracket and edges rather than the circuit board. Never force a card into a slot; if it will not seat with gentle firm pressure, something is misaligned. Never run a system with the GPU only partially inserted, as arcing across half-contacted pins damages both card and slot. Treat the power supply as a sealed unit and never open its casing, since capacitors inside hold a dangerous charge long after it is unplugged.
Common Mistakes That Waste Hours
The first is changing several things at once. Reseating the card, clearing CMOS, and flashing BIOS in a single pass means you learn nothing when it works and nothing when it does not. Change one variable, reboot, observe.
The second is trusting the monitor. Testing on a single display with a single cable leaves two untested failure points in the chain. The third is assuming a new card must be good. Dead-on-arrival hardware exists, and warranty windows are finite, so do the swap test early rather than after a week of firmware experiments.
The fourth is ignoring the case: some compact chassis stop a long card seating because of a drive cage or front radiator, so measure clearance first. If the card really has failed, the roundups of the best GPUs for gaming and the best RTX series GPUs are a sensible starting point for a replacement that matches your monitor and power supply.
Frequently Asked Questions
Why does my PC only use integrated graphics?
Either the monitor is plugged into the motherboard rather than the card, or firmware is set to prioritise the integrated display adapter. Move the cable to the card’s own output and set the primary display option in BIOS to PCIe.
Can a weak power supply cause a card to disappear?
Yes. Insufficient or unstable power commonly produces intermittent detection, black screens under load, and driver crashes that look identical to a hardware fault. Test with a stronger unit before writing the card off.
Does the PCIe slot latch really matter?
It does. The latch holds the card at the correct depth, and a card sitting a millimetre proud makes poor contact on some pins. Always confirm the latch clicked shut.
Should I flash the BIOS before trying anything else?
No. Flashing carries a small risk of bricking the board and should come after the free, reversible checks. Reserve it for cases where your card is newer than the board or the release notes mention support for your GPU generation.
My card is detected but shows Code 43. Is it dead?
Not necessarily. Code 43 usually indicates a driver or power delivery problem rather than a failed core. Do a clean driver reinstall, confirm the supplementary power cables are seated, and reset firmware to defaults first.
Final Thoughts
A graphics card that refuses to appear feels catastrophic, but the underlying cause is usually mundane: a cable on the wrong port, a latch that never clicked, a firmware setting inherited from a previous build, or drivers that half-installed. Work outward from the free checks, change one thing at a time, and use a known-good card or a second system to draw a firm line between component failure and configuration. Do that and you will either have your display back within the hour or a confident answer about which part needs replacing, which is worth just as much when you are deciding where to spend next.
