GPU Artifacts Already on the BIOS / UEFI Screen? How Pre-OS Symptoms Rule Out the Driver and Point to Hardware

| HK GPU Fix

GPU Artifacts Already on the BIOS / UEFI Screen? How Pre-OS Symptoms Rule Out the Driver and Point to Hardware

Key Takeaways

If artefacts are already visible on the BIOS or UEFI screen before Windows loads, the Windows graphics driver cannot be the cause, because it has not loaded yet. This guide explains how to use the monitor's own menu, cable and port swaps, and reseating checks, in order, to separate a cable, monitor, firmware, or graphics-card hardware fault before seeking inspection.

What This Article Covers, and What It Does Not

This article is for one specific situation: you switch the computer on, enter the motherboard firmware setup screen or see the power-on self-test screen, and the picture is already wrong. There may be speckles, blocks of wrong colour, horizontal or vertical lines, missing colours, or ghosting. The picture is present, but it is not clean. That combination matters, because it changes what can be ruled out and what still needs to be tested.

This article does not cover a completely black screen or a "no signal" message. A screen with no picture at all is a different diagnostic problem, with different first steps. It also does not cover artefacts that appear only after Windows has loaded, only in a game, or only in one application. Those symptoms point in a different direction, and they are discussed in other guides on this site. The value of this article is narrow on purpose: when the fault is already visible before the operating system starts, timing itself is evidence, and using that evidence correctly can save a great deal of wasted effort on driver changes that cannot affect the firmware screen.

A useful habit runs through the whole article: change one thing at a time, look at the firmware screen again, and write down what changed. If two things are changed at once and the picture improves, it is impossible to know which change mattered. That uncertainty matters when the fault returns later, or when the card is taken in for inspection and the history has to be described accurately.

Why Timing Rules Out the Windows Driver

When a computer starts, the motherboard firmware and the graphics card firmware draw the early screens using a basic frame buffer. This is a deliberately simple way of putting a picture on the screen before the full operating system exists. At this stage, the Windows graphics driver supplied by the graphics card manufacturer has not been loaded. Game shaders have not been loaded. Overlay software has not been loaded. Driver settings, driver versions, and driver profiles have not come into play.

That is why a firmware-screen artefact cannot be caused by a Windows driver setting, a game shader, overlay software, or a driver version. None of those components are running yet. Reinstalling a driver, changing a driver setting, or cleaning out a driver with a driver-removal utility cannot change what the firmware screen looks like, because the firmware screen does not use that driver. This is not a claim that drivers never cause display problems. Drivers very often cause problems after Windows loads. It is a claim about timing: a fault that is already present before the driver loads cannot be blamed on the driver.

It is equally important not to jump from "not the driver" to "the graphics card core has failed." Ruling out the driver narrows the field, but it does not finish the diagnosis. Several layers are still in play, and they must be separated in order: the cable and connector, the monitor itself, the graphics card firmware, the graphics card output path, and the graphics card memory and core hardware. Each of those layers can produce a dirty picture on the firmware screen. The rest of this article works through them in an order that starts with the cheapest and least invasive checks and ends at the point where home testing should stop.

The Most Decisive Test: The Monitor's Own Menu

The monitor's own on-screen display menu, usually called the OSD, is drawn by the monitor itself. It does not pass through the graphics card rendering path. That single fact makes the OSD test the most decisive step in this whole guide, and it should be done before any cable is swapped or any component is reseated.

Press the monitor's own physical buttons to bring up its menu. Look carefully at the menu text, icons, and borders. There are two possible outcomes, and they point in opposite directions.

If the OSD menu itself is dirty — if the menu text shows the same speckles, lines, or colour errors as the firmware screen — then the fault is on the monitor panel side. The monitor is producing a distorted picture even when it draws its own menu, without any help from the graphics card. In that case, further work on the graphics card is going around in circles. The monitor side needs attention, and the graphics card steps in this article no longer apply.

If the OSD menu is clean — sharp text, correct colours, straight borders — while the firmware screen behind it is dirty, then the fault is on the graphics card, cable, or connector side. The monitor can draw a clean picture when it draws for itself, so the panel is not the source of the distortion. The distortion is entering the picture before it reaches the monitor, or it is being introduced in the connection between the card and the monitor. That result sends the diagnosis on to the cable and connector checks.

This test costs nothing, takes a minute, and prevents the most common waste of effort in this situation: treating a monitor fault as a graphics card fault, or the reverse. Do it first, and trust its result. If the result is unclear because the OSD menu is small or the distortion is subtle, take a photograph of the screen with a phone, look at the photograph enlarged, and repeat the test. Do not rely on memory to describe a subtle distortion later.

How the Firmware Picture Is Built, in Plain Terms

Understanding the firmware picture in broad terms helps explain why the checks are ordered the way they are. At power-on, the motherboard firmware initialises the system and the graphics card firmware provides a basic display output. Together they write a picture into a frame buffer, which is a region of memory holding the image to be shown. The display output path then sends that image through a connector and a cable to the monitor, which displays it.

A fault anywhere along that chain can distort the picture. A damaged cable or a loose connector can corrupt the signal on its way to the monitor. A faulty monitor panel can distort a clean incoming signal. A problem in the graphics card firmware can produce a wrong picture from the start. A problem in the graphics card output path can corrupt the picture as it leaves the card. A problem in the graphics card memory or core hardware can corrupt the picture as it is being formed. All of these produce the same broad symptom — a dirty firmware screen — which is why the symptom alone cannot identify the layer. Only ordered elimination can.

One more distinction is worth making. The firmware screen has no screenshot function. A screenshot taken inside Windows captures the rendered image before it is sent to the monitor, so a screenshot fault points toward the rendering side. On the firmware screen, there is no such tool. The only record is what the eye sees and what a phone photograph captures. That limitation is not a minor detail. It means that descriptions given later, for example when seeking inspection, should be based on photographs taken at the time, not on a remembered impression of "lines" or "dots" that may have changed since.

The Diagnostic Sequence, in Order

Work through the steps below in order. After each step, return to the firmware screen and look again. Record the result before moving on. If a step produces a clean firmware screen, the layer changed in that step is where the fault sits, and the later steps are no longer needed for diagnosis, although the faulty item itself may still need replacement or inspection.

Step 1: Record When and How the Artefacts Appear

Before changing anything, record the timing and the form of the fault. Is the distortion present only on the firmware screen, or does it also appear in Windows and in games? Is it fixed in one place on the screen, or does it move as the picture changes? Is it present from the first moment the screen lights up, or does it appear partway through the firmware screens? Take photographs of the firmware screen with a phone. Photograph the whole screen, and, if the distortion is localised, a closer photograph of the affected area as well.

This record serves two purposes. First, it creates a baseline. Later steps can only be judged against a baseline: without it, a small improvement may be missed or an unchanged fault may be mistaken for an improvement. Second, it gives an accurate history if the card is later taken in for inspection. A photograph is more useful than a verbal description, because it shows the exact pattern, position, and severity at the time it was observed.

Step 2: The Monitor OSD Test

Bring up the monitor's own menu as described above. If the menu is dirty, the diagnosis points to the monitor panel side, and the graphics card steps that follow do not apply. If the menu is clean and the firmware screen is dirty, continue to Step 3.

If the monitor has no accessible menu, or the menu cannot be brought up, note that limitation and move on, but do not skip the next step. The cable and monitor swap in Step 3 provides a second way to separate the monitor from the card.

Step 3: Cable, Connector, and Monitor Cross-Checks

Change one item at a time: first try a different cable, then a different output connector on the graphics card if one is available, then a different monitor if one is available. After each single change, check the firmware screen again.

If the firmware screen becomes clean after a particular change, the item just replaced is where the fault sits: the old cable, the old connector, or the old monitor. That conclusion follows only because one thing was changed at a time. If the firmware screen remains dirty after every swap — a different cable, a different connector, and a different monitor have all been tried — then the fault remains on the graphics card side, and the diagnosis moves to Step 4.

A note on cables is worth making here. Swapping a cable is a test, and the result of the test is the evidence. The price of the replacement cable is not evidence. An expensive cable that produces the same dirty picture has not ruled anything in or out beyond what a standard cable would have shown. Use a known-good cable for the test, and judge by the picture, not by the price tag.

Step 4: Reseat the Graphics Card and Check Power Connectors

With the power disconnected, reseat the graphics card in its slot and confirm that the power connectors are fully inserted and latched at both ends where applicable. This step addresses the contact layer: a card that is not fully seated, or a power connector that is not fully latched, can produce unstable behaviour. Reseating is a limited, defined action. It does not involve cleaning internal components, modifying the card, or opening the card itself.

After reseating, check the firmware screen again. If it is clean, the fault was at the contact layer. If it remains dirty, the diagnosis has now exhausted the home checks that this article recommends, and the next step is inspection rather than further home testing.

Step 5: The Inspection Boundary

If all four steps above have been completed and the firmware screen is still dirty, stop home testing. The remaining candidate layers — graphics card firmware, output path, memory, and core hardware — cannot be separated at home without specialist equipment and without risking further damage. In particular, do not attempt to rewrite graphics card firmware at home when a hardware fault is suspected, as explained in the boundaries section below.

A cross-check with another computer is sometimes suggested at this point. That test is only useful if another suitable computer is already available. Do not buy hardware for the sake of this test. If another computer is available and the card shows the same dirty firmware screen in it, that result strengthens the conclusion that the fault is on the card side. If the card is clean in another computer, the fault may sit in the original computer's slot, power delivery, or configuration, and that information is useful for inspection. Either way, the result should be recorded with photographs, not from memory.

Comparing Firmware and Windows Symptoms

The combination of firmware symptoms and Windows symptoms carries diagnostic meaning, and it is worth recording both halves before seeking inspection.

Firmware screenWindows screenWhat the combination suggests
DirtyDirtyThe fault is at the hardware, cable, or monitor layer, because it is present before and after the driver loads
DirtyCleanThe rendering path already shows a problem at firmware level, even though Windows currently looks normal; this still warrants inspection, as explained below
CleanDirtyThe fault is at the driver or system layer, because it appears only after the driver loads
CleanClean, but the fault appears only in one game or applicationThe fault is more likely at the application, settings, or load-specific layer, not at the firmware or general driver layer

The second row surprises many readers, and it deserves emphasis. A firmware screen that is dirty while Windows looks completely normal is not a reason to ignore the problem. It means the rendering path already shows a fault at the simplest level of operation, before any load is applied. Windows looking normal does not cancel that evidence. It may mean the fault has not yet shown itself under normal desktop use, or that Windows is compensating in a way the firmware screen does not. Either way, the firmware evidence stands, and inspection is warranted rather than waiting for the fault to appear under load.

The third row is the mirror image, and it is the reason this article keeps pointing readers away from driver changes for firmware faults. When the firmware screen is clean and the fault appears only after Windows loads, the driver and system layers are genuinely in play, and driver-related steps become relevant. That is a different article, and its steps should not be imported into a firmware-screen problem.

Boundaries and Stop Rules

These rules are not optional extras. They mark the point where further home testing stops helping and starts risking damage or wasting money.

First, if a firmware-screen fault is accompanied by a burning smell, visible scorch marks, or discoloured connectors, stop using the card immediately and seek inspection. Do not power the system on again to "check whether it is still there." The smell or mark is itself the evidence, and repeated power-on attempts can only add damage.

Second, do not rewrite graphics card firmware at home when a hardware fault is suspected. A firmware rewrite that is interrupted, or applied to a card that already has a hardware fault, can leave the card unusable. Graphics card firmware problems are a real category, but they are a specialist matter, and this article deliberately does not provide rewrite steps. If firmware is suspected, that suspicion should be taken to inspection, where it can be checked with proper tools.

Third, as noted above, a dirty firmware screen with a normal Windows desktop still warrants inspection. Do not adopt the rule "Windows looks fine, so there is nothing wrong." The firmware screen is a simpler, earlier test of the same rendering path, and a fault there is meaningful even when Windows currently hides it.

Fourth, do not use driver updates as the solution for this symptom. Updating a driver cannot change the firmware screen, because the firmware screen does not use that driver. A driver update performed for other reasons is a separate matter, but it should not be presented as a fix for a firmware-screen fault, and a failed driver update should not be mistaken for evidence about the hardware.

Fifth, do not open the graphics card, modify its cooling, or attempt component-level work at home on the basis of this article. The steps here are deliberately limited to observation, menu tests, cable and monitor swaps, and reseating. Anything beyond that belongs to inspection.

Common Misconceptions

"A fault before Windows loads is the same as a fault after Windows loads"

Timing is itself diagnostic evidence. Before Windows loads, the driver is not running, so a fault at that stage cannot be a driver fault. After Windows loads, the driver is running, so a fault at that stage may be a driver fault. Treating both stages as identical discards the most useful piece of information available, which is when the fault first appears.

"A dirty firmware screen means the graphics card core has failed"

This jumps past several layers that must be excluded first: the cable, the connector, the monitor, and the firmware. Each of those can produce a dirty firmware screen. A hardware conclusion is only reasonable after the monitor menu test and the cable, connector, and monitor swaps have been completed and the fault remains. Even then, the conclusion is "the fault remains on the graphics card side," which is narrower and more honest than naming a specific failed component.

"A screenshot shows the fault, so it must be the graphics card"

A screenshot taken inside Windows captures the image before it reaches the monitor, so a screenshot fault does point toward the rendering side. But the firmware screen has no screenshot function at all. A photograph of the firmware screen is not a screenshot, and it does not carry the same diagnostic meaning. On the firmware screen, the only valid record is direct observation and a phone photograph, and the diagnostic weight comes from the OSD test and the swap tests, not from the existence of a photograph.

"Reinstalling the driver or cleaning it out will fix it"

Driver-removal utilities clean the Windows driver. The firmware screen does not use the Windows driver. A driver clean may be the right step for a Windows-only fault, but it cannot change a firmware screen, and performing it for a firmware fault wastes time and creates a false sense of having "tried everything" when the relevant layers have not been tested at all.

"A more expensive cable will fix it"

A cable swap is a test, and the test result is what matters. If a different cable produces a clean firmware screen, the old cable was at fault, regardless of price. If a different cable produces the same dirty screen, the cable layer has been excluded, regardless of price. Price is not a diagnostic result, and buying progressively more expensive cables without recording the result of each swap is not a diagnostic sequence.

"The pattern of the artefacts tells you which chip has failed"

Pattern-based identification — for example, attributing a particular speckle or block pattern to memory or to the core — is a separate subject with its own guide on this site. This article deliberately does not make that attribution. Its method is timing and elimination, not pattern reading. Readers who want pattern-based discussion should read that guide separately, and should not import its conclusions into a timing-based diagnosis without checking that both methods agree.

What to Bring to Inspection

If the diagnostic sequence ends at the inspection boundary, the following record makes inspection faster and more accurate: photographs of the dirty firmware screen, photographs of the clean or dirty monitor menu, a note of which cable, connector, and monitor swaps were tried and what each one showed, a note of whether reseating changed anything, and a note of whether Windows is clean or dirty alongside the firmware screen. Also note any burning smell, scorch mark, or connector discolouration, and do not power the card on again if any of those are present.

Inspection charges follow the site's published three-tier range of HK$400–2,800, with the actual quote given after inspection. No other price is stated in this article, because the correct charge depends on what inspection finds, and stating a figure in advance would be guessing.

Frequently Asked Questions

If the firmware screen is dirty but Windows is fine, can I just keep using the computer?

It is understandable to want to keep using a computer that looks normal in Windows, but the firmware result should not be ignored. The firmware screen tests the rendering path at its simplest level. A fault there means the path already shows a problem before any load is applied. Inspection is warranted to establish which layer is at fault, rather than waiting for the fault to appear under load, when it may be harder to diagnose or may coincide with other symptoms.

Does a dirty firmware screen always mean the graphics card needs replacing?

No. The cable, connector, and monitor layers must be excluded first, and the monitor menu test may point to the monitor rather than the card. Even when the fault remains on the card side after all home steps, "card side" is not the same as "card must be replaced." Inspection may identify a firmware, output-path, or contact issue, or may confirm a hardware fault. Replacement is a conclusion that follows inspection, not a conclusion that follows the symptom alone.

Can I fix this by updating the graphics driver?

No, not for a firmware-screen fault. The firmware screen is drawn before the Windows driver loads, so a driver update cannot change it. If a driver update is performed for another reason, it should not be counted as a test of this fault, and its result should not be used to argue for or against a hardware cause.

Why can't I take a screenshot of the firmware screen?

The firmware environment does not provide a screenshot function. Screenshots are an operating-system feature that captures the rendered image inside the system. Before the operating system loads, that feature does not exist. A phone photograph is the practical record for a firmware screen, and it should be taken at the time the fault is observed, not reconstructed later from memory.

My monitor menu looks clean but the firmware screen is dirty. What does that prove?

It proves the monitor can draw a clean picture for itself, so the panel is not the source of the distortion. The fault therefore sits on the graphics card, cable, or connector side. It does not yet say which of those three is at fault. That separation comes from the cable, connector, and monitor swaps in the next step.

My monitor menu is dirty too. Should I still test the graphics card?

No, not as the next step. If the monitor's own menu is distorted, the monitor panel side is implicated, because the menu does not pass through the graphics card. Further graphics card testing would be working on the wrong side of the problem. The monitor side needs attention first, and graphics card conclusions should wait until a known-good monitor shows whether the firmware screen is clean or dirty.

I changed the cable and the connector at the same time and the picture improved. Is that enough?

It shows the fault was in one of the two things changed, but not which one. If the fault returns, the history will be ambiguous, and a replacement may be bought for the wrong item. When time allows, repeat the test changing one item at a time, so the faulty item is identified. At minimum, record that both were changed together, so the ambiguity is known rather than hidden.

Is it safe to rewrite the graphics card firmware myself if I suspect a firmware problem?

No, this article recommends against it when a hardware fault is suspected. A rewrite performed on a card that already has a hardware fault, or a rewrite that is interrupted, can leave the card unusable. Firmware suspicion is a reason to seek inspection with proper tools, not a reason to attempt a rewrite at home. This article deliberately provides no rewrite steps.

The artefacts only appear on the firmware screen for a few seconds and then disappear. Does that still count?

Yes, it still counts as a firmware-screen fault, provided it is genuinely observed on the firmware screen and not confused with a later Windows screen. Record it with a photograph if possible, note how long it lasts and whether it appears on every start or only sometimes, and apply the same sequence. An intermittent fault is still a fault, and intermittency is itself useful information for inspection.

What if I have no spare cable, connector, or monitor to swap with?

Then the swap step cannot be completed at home, and that limitation should be stated plainly rather than worked around by guessing. The monitor menu test and the recording step can still be done, and reseating can still be done with care. Beyond that, the diagnosis stops at the same inspection boundary, with a clear note of which steps could not be performed. Inspection can then perform the swaps with known-good parts.

Summary

A dirty BIOS or UEFI screen is useful evidence precisely because of when it appears. It rules out the Windows driver, but it does not by itself condemn the graphics card. The monitor menu test separates the monitor from the card side. Cable, connector, and monitor swaps separate the connection from the card. Reseating separates the contact layer from the rest. If the fault survives all of those steps, or if there is any sign of burning or scorching, home testing stops and inspection takes over. Working in that order, changing one thing at a time, and photographing what is actually seen, gives a diagnosis that is honest about what has been excluded and what has not.

For inspection, HK GPU Fix is at Room 1209, Dali Plaza, Shek Mun. The telephone number is 2186 8010 and WhatsApp is 6558 6806. No home-visit service is offered. Inspection charges follow the site's published three-tier range of HK$400–2,800, with the actual quote given after inspection.

Related articles

Related services

If you need Display output repair, see these service pages: