VIDEO_MEMORY_MANAGEMENT_INTERNAL (0x10E) means Windows’ video memory manager hit a condition it couldn’t recover from, and Microsoft says a misbehaving video driver usually causes it. Install Windows updates and the current display driver, or roll back the one that preceded the crashes, then reset GPU core, memory and power tuning to stock. Test hardware only if it keeps returning.
Contents
At a glance
- Applies to
- Windows 11 and Windows 10, stop code 0x10E
- Microsoft's fix
- Check Windows Update for an updated video driver
- Find Parameter 1
- Event Viewer, Kernel-Power event 41, Details tab
- Don't
- Buy a GPU or replace dxgmms2.sys over the stop-code name
Provenance
- Oct 4, 2026Published
- Nov 18, 2026Last substantive update
VIDEO_MEMORY_MANAGEMENT_INTERNAL is bug check 0x0000010E. It means Windows’ video memory manager reached a condition it could not recover from. It does not mean “your graphics card ran out of VRAM,” and it does not condemn the memory chips on the card.
Microsoft’s 0x10E reference says the bug check is usually caused by a video driver behaving improperly, and its resolution is to check Windows Update for an updated video driver. Everything after that driver step below is an order for changing one thing at a time; Microsoft’s page doesn’t prescribe it.
1. Confirm the stop code and find Parameter 1
Windows has several graphics and memory blue screens whose names blur together. The number separates them and prevents a long detour through the wrong tests.
| Stop code | What Windows caught | Start here |
|---|---|---|
0x10E VIDEO_MEMORY_MANAGEMENT_INTERNAL | The video memory manager could not recover from an internal condition | This guide: display driver, stock GPU settings, Parameter 1 |
0x116 VIDEO_TDR_FAILURE | Windows tried to reset the display driver and the recovery failed | GPU driver crash and TDR guide |
0x1A MEMORY_MANAGEMENT | Windows found a severe error in its general memory-management structures | MEMORY_MANAGEMENT diagnosis |
A game error that reads DXGI_ERROR_DEVICE_REMOVED is an error code a DXGI function returns to the game, not this bug check; the DXGI device-removed guide covers it.
You don’t need to have caught the values on the blue screen, and you don’t need extra software. Windows keeps them:
- Open Event Viewer from Start search and go to Windows Logs > System.
- Find the Kernel-Power event with ID 41 logged when the PC restarted after the crash.
- On its Details tab,
BugcheckCodeis the stop code in decimal:270is0x10E.BugcheckParameter1is Parameter 1, already in hex.
Microsoft’s Event ID 41 page documents those fields, and its stop-code data page confirms the System log lists all four parameters. If you use WinDbg, !analyze -v on the newest file in C:\Windows\Minidump shows the same values. If BugcheckCode reads 0, Windows recorded no stop code for that restart. Microsoft says that can mean the power was interrupted or the power supply is underpowered or faulty, or that no dump file was configured (look for a volmgr event 46 and check the page file), so check those first. If the PC restarts with no blue screen at all, the random-restart guide starts from the same event.
Microsoft says Parameter 1 describes the type of video-memory error, and that a value missing from its table needs individual examination rather than a guess.
Use Parameter 1 to choose the branch
As of November 18, 2026, Microsoft’s 0x10E table lists 68 Parameter 1 values, from 0x1 through 0x44. Nine of those rows explicitly name a driver error or a display-driver-interface call (0xB, 0xF, 0x10, 0x12, 0x13, 0x16, 0x19, 0x2D and 0x2E), and the phrases “memory corruption” and “bad hardware” do not appear. Those are counts from the published table, not data on how often any cause occurs on real PCs. The table supports starting with the driver, but it neither identifies nor excludes a hardware fault. I would not buy or RMA a card on the stop-code name alone.
You do not need to translate every row. The groups below collect the ones useful for routing. The middle column is Microsoft’s; the right-hand column is a suggested next test, not something Microsoft prescribes per value.
| Parameter 1 | What Microsoft’s row says | Next test |
|---|---|---|
0xB, 0xF, 0x10, 0x12, 0x13, 0x16, 0x19, 0x2D, 0x2E | A driver returned an invalid result, overflowed a buffer, broke a contract, or a DDI call failed | Step 2: driver update or rollback first |
0x25, 0x30 | A GPU aperture write or GPU virtual-address operation failed | After the driver step, remove tuning variables (Step 3) and look for a repeatable result |
0x37 | VidMm failed to lock allocation pages during TDR | Compare the TDR/0x116 fix path after the driver step |
0x3B, 0x3C | Memory remained rotated, allocated, or mapped during a D3 power transition | Record whether the same power transition repeatedly precedes the crash |
0x41, 0x42 | A cross-adapter allocation had an invalid residency or adapter count | Step 4: compare supported single- and multi-adapter modes |
0x43 | A reserved-frame-buffer save or restore could not map a page | Record whether a power-state transition repeatedly precedes the crash |
If your value is not in that short list, read its exact row on Microsoft’s page. For example, 0x25 says the GPU attempted to write over an undefined aperture area; that does not tell you whether the initiating fault was firmware, a driver, unstable tuning, or failing hardware.
2. Update Windows and the display driver, or roll it back
This is Microsoft’s own resolution for persistent 0x10E crashes. Microsoft’s driver-update instructions add that a manual driver should come from the device manufacturer’s site, and document rollback when trouble begins after an update.
- In Settings > Windows Update, install the normal pending Windows updates and restart.
- Install the current display package intended for your model. On a laptop or branded desktop, check the PC maker’s support page and choose only a package listed for that exact model and Windows version.
- If the crash began immediately after a display-driver update, roll back to the last version that was stable on this machine instead of repeatedly installing the same package.
- If ordinary update or rollback does not change the result, use the clean-install procedure in the GPU driver crash guide.
Change one layer at a time, then repeat the workload that triggered the crash. A simultaneous BIOS update, driver update, Windows repair and hardware swap may stop the crash, but it leaves you with no idea which change mattered, and no useful evidence if the blue screen returns next week.
3. Put the GPU and video memory back to stock
Microsoft’s 0x10E page does not name tuning as a cause. Removing it is still a cheap isolation step that leaves a repeat test with fewer variables.
- Reset GPU core offset, memory offset, voltage curve, and power-limit changes.
- Close tuning utilities so they do not reapply a profile at sign-in.
- If you recently switched hardware-accelerated GPU scheduling on or off, put it back the way it was; the HAGS guide shows where the switch is.
- If the card has a physical dual-BIOS switch, shut down before moving it and document the original position. Use the card maker’s instructions.
- Retest the workload that normally triggers the crash.
If it stays stable at stock, stop there. A clean run suggests one of the removed settings correlated with the crash; repeat the same trigger before deciding which one.
4. Treat hybrid graphics as one stack
Laptop and multi-GPU systems can involve an integrated GPU, a discrete GPU, and cross-adapter allocations in the same session. Microsoft’s table specifically includes cross-adapter states at 0x41 and 0x42.
Start with the PC maker’s current packages for the exact model. If the firmware or graphics control panel offers a supported integrated-only or discrete-only mode, repeated tests in each mode can compare the two paths. Do not disable the only display adapter on a machine with no safe way back.
For a desktop with two add-in cards, remove the secondary card for a controlled test if doing so is physically safe and the primary card can drive the display. A difference between configurations is evidence about the path; it does not prove that one card is defective.
5. Repair Windows’ copy of the graphics components
Run this step only when Windows servicing or System File Checker has already reported corruption; Microsoft does not say 0x10E implies damaged Windows files. Microsoft’s system-file repair instructions put DISM before System File Checker. Open Terminal as administrator and run:
DISM.exe /Online /Cleanup-Image /RestoreHealth
sfc /scannow
Restart after both commands finish. This step is for damaged or mismatched operating-system files. It cannot stabilize an overclock, repair a failing memory chip, or correct a defective power supply.
Avoid sites offering individual DLL or SYS downloads. Microsoft’s VidMm documentation says the relevant kernel graphics files ship only as part of the operating system and that graphics-driver developers must not mix versions of them.
6. Test hardware only after the software path is clean
Start hardware isolation when 0x10E keeps returning with a current or known-stable driver and the GPU at stock. If the display also shows persistent blocks, colored flashes, or broken geometry, the GPU artifact guide covers that evidence; if the card drops out of Device Manager entirely, use the graphics-card-not-detected guide.
- Log the trigger. Note the application, display configuration, temperature, and whether the crash followed resume, sign-in, or a monitor change.
- Test one display path. Disconnect optional monitors and adapters. A clean run narrows the trigger but does not prove the disconnected monitor is defective.
- Test the card in a known-good system, or a known-good card in this PC. A failure that repeatedly follows one card is a stronger isolation signal than the filename in one dump, though still not proof by itself.
- Test system RAM when the evidence points beyond graphics. Multiple unrelated stop codes, memory-profile instability, or errors outside graphics workloads justify the 0x1A memory test path. A lone 0x10E does not.
Simple VRAM pressure is not a verdict either. Microsoft’s GPU-segments page explains how VidMm manages segment types and allocations; it does not say capacity exhaustion causes 0x10E, so a VRAM meter alone cannot explain the crash.
What 0x10E is reporting
Windows’ video memory manager is called VidMm. Microsoft’s VidMm and GPU scheduling documentation says it sits inside the DirectX graphics kernel, manages GPU memory, and works with the GPU scheduler. The Windows components that implement this work include dxgkrnl.sys, dxgmms1.sys, and dxgmms2.sys.
When a crash report names dxgmms2.sys, it identifies a Windows component in the path where the failure surfaced, which is a location rather than a culprit. Microsoft says these files are part of the operating-system installation, are not separately downloadable, and are designed to work with the other OS files that accompany them. Do not replace one with a copy from a download site or another PC.
What not to do
- Do not buy a GPU from the stop-code name. The official cause section points first to the video driver.
- Do not replace
dxgmms2.sysordxgkrnl.sysby hand. If Windows integrity checks report corruption, use the conditional DISM and SFC step above. - Do not edit TDR registry values as a 0x10E fix. Microsoft’s TDR registry documentation says end users should not manipulate those keys; they are for targeted testing and debugging during driver development.
- Do not use 0x1A advice for a 0x10E crash. System-memory testing may become relevant later, but the first branch is different.
- Do not treat Microsoft’s 68 table rows as cause statistics. They are error types, not a tally of user crashes.
Other stop codes read their first parameter differently: the WHEA_UNCORRECTABLE_ERROR guide decodes a hardware error record, and the KERNEL_SECURITY_CHECK_FAILURE guide covers 0x139.
Sources
- Microsoft Learn: Bug Check 0x10E: VIDEO_MEMORY_MANAGEMENT_INTERNAL — definition, Parameter 1 table, cause, and resolution; accessed November 18, 2026.
- Microsoft Learn: Event ID 41 and Analyze bug check data — where the four bug-check parameters appear in the System log; accessed November 18, 2026.
- Microsoft Learn: Video memory management and GPU scheduling — VidMm, VidSch, and the Windows graphics-kernel files; accessed November 18, 2026.
- Microsoft Learn: GPU segments — memory segments and aperture behavior; accessed November 18, 2026.
- Microsoft Support: Update drivers through Device Manager in Windows — Windows Update, manufacturer downloads, reinstall, and rollback paths; accessed November 18, 2026.
- Microsoft Support: Use the System File Checker tool to repair missing or corrupted system files — DISM and SFC order and commands; accessed November 18, 2026.
- Microsoft Learn: Testing and debugging TDR during driver development — TDR registry keys are for driver-development testing and debugging, not end-user manipulation; accessed November 18, 2026.
Frequently asked questions
What does VIDEO_MEMORY_MANAGEMENT_INTERNAL mean?
How do I fix VIDEO_MEMORY_MANAGEMENT_INTERNAL in Windows 11?
Does VIDEO_MEMORY_MANAGEMENT_INTERNAL mean my graphics card’s VRAM is failing?
Is VIDEO_MEMORY_MANAGEMENT_INTERNAL caused by overclocking?
Can bad system RAM cause a 0x10E blue screen?
Did a Windows update cause my 0x10E crashes?
Is 0x10E the same as the MEMORY_MANAGEMENT 0x1A blue screen?
Sources and corrections
- Last updated
- Methodology
- See our methodology for research and review standards. It draws on 5 cited sources, listed below, each checked against the original page on the date above.
- Sources
- Microsoft Learn: Bug Check 0x10E VIDEO_MEMORY_MANAGEMENT_INTERNAL accessed 2026-11-18
- Microsoft Learn: Event ID 41 when the system restarts unexpectedly accessed 2026-11-18
- Microsoft Learn: Analyze bug check (stop code error) data accessed 2026-11-18
- Microsoft Support: Update drivers through Device Manager in Windows accessed 2026-11-18
- Microsoft Learn: Video memory management and GPU scheduling accessed 2026-11-18
- Update log
- 2026-11-18 — Added how to find Parameter 1 in Event Viewer without extra software, led the answer with the driver and GPU-tuning fixes, numbered the steps and cut repeated hedging.
- Corrections
- Spotted an error? Email a correction with the affected line and what you found. The draft includes this page’s address. Verified changes are recorded here. Contact details.