Nothing's more annoying than a graphics error crashing your session mid-render or mid-game, so let's fix 0xC0262106 without the usual runaround.
The fix that works most of the time
The culprit here is almost always dxgkrnl.sys or your display driver hitting a TDR (Timeout Detection and Recovery) limit. Windows gives the GPU 2 seconds to respond, and when it doesn't, it kills the allocation and hands you 0xC0262106. Bump that timeout and update the driver properly.
- Boot into Safe Mode. Shift+click Restart, then Troubleshoot → Advanced → Startup Settings → Restart → press 4. Don't skip this; the driver is locked while Windows is running normally.
- Use DDU to remove the driver. Download Display Driver Uninstaller from Guru3D. Run it, pick "Clean and restart" for your GPU vendor. Don't use Device Manager's uninstall — it leaves junk behind that keeps causing allocations to fail.
- Install a known-good driver. If you're on NVIDIA 552.xx and hitting this on a 30-series card, drop to 551.86. If you're AMD on Adrenalin 24.3.1, roll back to 23.12.1. The newest driver is very often the broken one.
- Raise the TDR delay. Open regedit and navigate to:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDriversCreate a DWORD called TdrDelay and set it to 10 (decimal). If it already exists, change the value. Also create TdrDdiDelay and set it to 10.
- Reboot. Cold boot, not restart if you can help it. Power off, wait 30 seconds, power on.
That sequence clears this error in roughly 8 out of 10 cases I've handled. The other 2 are hardware or VRAM-related, and I'll get to those below.
Why this actually fixes it
0xC0262106 is a WDDM status code. It's raised by the GPU scheduler when a video memory allocation request gets rejected — either because the driver stalled past the 2-second TDR threshold, or because the requested allocation exceeded what the driver said the GPU could handle. Bumping TdrDelay from 2 to 10 seconds gives heavy workloads (Blender renders, Cyberpunk at 4K, Chrome with 40 tabs plus hardware acceleration) room to finish before Windows yanks the rug out.
DDU matters because stale registry keys and leftover driver files from a previous vendor or version can leave the GPU reporting the wrong VRAM size to the allocator. Windows then rejects valid allocations because the reported ceiling is wrong. Clean install rebuilds those entries from scratch.
If you skip DDU and just "update driver" from Device Manager, you're asking for the same error again in a week. I've seen it too many times.
When the standard fix doesn't do it
1. Integrated GPU + discrete GPU conflict
Laptops with Intel UHD and an NVIDIA dGPU share memory through the display adapter. If the app is trying to allocate on the wrong adapter, you'll get 0xC0262106 every time. Force the app to the discrete GPU:
Settings → System → Display → Graphics → Add desktop app → Options → High performanceOr via the registry, set GpuPreference=2 under the app's UserGpuPreferences key.
2. Hyper-V or WSL2 stealing VRAM
If you've got WSL2 running with GPU passthrough (wsl --install --enable-gpu), it holds a chunk of VRAM. On a card with 8 GB, that leaves too little for the primary workload and allocations start failing. Shut down WSL with wsl --shutdown and see if the error stops.
3. Failing VRAM module
If the above fix works for a week and then comes back, and you're seeing it under a specific card, suspect the VRAM itself. Run OCCT's VRAM test for 20 minutes. Errors there mean the card is dying, and no registry tweak will save you.
4. Overclocking / XMP instability
Unstable memory OC bleeds into GPU allocations because they share the PCIe path and system memory for staging. Reset BIOS to defaults, disable XMP/EXPO, test again. If the error vanishes, your RAM OC was the problem, not the GPU.
Preventing it from coming back
- Never use GeForce Experience or Adrenalin's "auto update" for drivers. Pick a version, keep it for 3-6 months, only move when a specific game or app requires it.
- Keep TdrDelay set to 8-10 permanently. It costs you nothing and saves you this headache.
- Watch GPU temps. Above 83°C sustained, drivers start missing deadlines. Clean the fans every 6 months if you're in a dusty environment.
- Disable hardware acceleration in Chrome, Discord, and Slack if you're on a 4 GB card. Those three alone can nibble 1.5 GB of VRAM just sitting there.
- Keep 15% of VRAM free. If a game needs 7.5 GB on an 8 GB card, you're going to fight 0xC0262106 on every alt-tab.
Do the DDU pass first, set TdrDelay, and stop chasing this thing every patch cycle. It's a scheduling problem, not a mystery.