Skip to main content

DPC_WATCHDOG_VIOLATION: what causes it, and how to fix it on Windows 11 and 10

Short answer

DPC_WATCHDOG_VIOLATION (0x133) means a driver kept the processor busy at top priority for too long. In 2025 and 2026 crash reports it was most often the graphics driver, then VPN, network and storage drivers. Read Parameter 1, find the driver in the dump and reinstall it cleanly, often a slightly older version.

On this page

DPC_WATCHDOG_VIOLATION (stop code 0x133) means a driver kept the processor busy at top priority for too long, so Windows’ watchdog timer decided the PC had hung and stopped it. In the 2025 and 2026 crash reports we read, the driver was most often the graphics driver, NVIDIA’s nvlddmkm.sys or AMD’s amdkmdag.sys. VPN, network and storage drivers came next. The fix is to read the first parameter of the crash, find the driver in the dump, and replace that driver with a clean, often slightly older, version.

How we put this together: this guide is research, not a test. It’s built from Microsoft’s bug check reference, Microsoft Q&A threads where advisors read real crash dumps, NVIDIA’s developer forums, Intel’s support note and Framework’s community forum, all listed under Sources. Steps apply to Windows 11 (24H2, 25H2 and 26H2) and Windows 10.

What does DPC_WATCHDOG_VIOLATION mean?

When a device needs attention, Windows runs short pieces of driver code at very high priority, and they’re meant to finish in microseconds. Microsoft’s guideline is under 100 microseconds for a deferred procedure call (DPC) and under 25 for an interrupt routine. If one runs far longer, or they hog the processor between them, the watchdog stops the PC. Microsoft’s reference says the cause is generally faulty driver code that doesn’t finish in time.

Parameter 1 tells you how hard this will be

The first number in the crash details splits this error into two kinds:

Parameter 1What happenedWhat the dump shows
0One driver routine ran past its limitThe stack usually names the driver
1Several routines together spent too long at high priorityThe dump may not stop in the guilty code, so look for a pattern across several dumps

Find it in Event Viewer, under Windows Logs, System: open the BugCheck entry for the crash and read its parameters. Or read it from the dump, as in step 4 below.

What causes it?

  1. The graphics driver. In Microsoft Q&A threads from November 2025 to May 2026, advisors who read the dumps named nvlddmkm.sys again and again. In one AMD case, a Radeon RX 9070 XT that crashed in Diablo 4, the owner only became stable after rolling back to Adrenalin 25.9.1, once 26.1.1, 26.2.2 and 26.3.1 had all crashed.
  2. VPN and network drivers, especially around sleep. A Framework laptop on Windows 11 24H2 crashed in sleep until its owner switched Mullvad’s VPN from WireGuard to OpenVPN. A Realtek network driver showed up in another thread’s dumps.
  3. Storage firmware and drivers. One Tom’s Hardware user with a Crucial P1 NVMe drive fixed it with a power setting from Crucial’s own guidance.
  4. Resuming from hibernate on new NVIDIA cards. RTX 5090 owners report the crash about two and a half minutes after waking from hibernate, and on 15 July 2026 NVIDIA staff said an internal bug is tracking it.
  5. Peripherals, overclocks and failing hardware are possible but less common. In the AMD thread, a new power supply didn’t help, and in another the RAM tested clean while the dumps kept naming drivers.

How to fix DPC_WATCHDOG_VIOLATION, step by step

1. Note what triggers it

Waking from sleep or hibernate, one game, a VPN connecting, plugging in a monitor, or a long video export: each points at a different driver. Intel’s own note on this error is about laptops crashing when an external HDMI monitor is connected.

2. Remove the variables

Unplug USB devices you don’t need and remove anything you added recently. Turn off overclocking, plus XMP, EXPO or PBO in the BIOS, while you test. If the crashes stop, add things back one at a time.

3. Update from the right places

Run Windows Update first. Then install the current graphics, chipset, network and storage drivers from your PC maker or the chip maker, and update the BIOS from your PC or motherboard maker. Don’t use a “driver updater” app.

4. Read the crash dump

Install Microsoft’s free WinDbg, open the newest file in C:\Windows\Minidump with File, Open Crash Dump, and run the analysis:

winget install Microsoft.WinDbg
.symfix C:\Symbols
.reload
!analyze -v

Check Arg1 (0 or 1) and IMAGE_NAME. If Arg1 is 0, the image name usually is the culprit. If it’s 1, run !dpcs to list the queued routines and compare several dumps. Treat “Probably caused by ACPI.sys” or ntoskrnl.exe as a location, not a verdict. In one Tom’s Hardware thread ACPI.sys showed up in most dumps, while the drivers that mattered were NVIDIA’s and Realtek’s.

Also run winver to confirm your real Windows version. One dump in a 2026 Microsoft Q&A thread reported “Windows 10” on a PC running Windows 11 24H2.

5. If it’s the graphics driver, reinstall it cleanly

This is the recipe two Microsoft Q&A advisors gave in 2026 threads. Remove the current driver completely with Display Driver Uninstaller (DDU), a free third-party tool, running it in Safe Mode. Restart, then install a slightly older stable driver from NVIDIA, AMD or Intel and test it for a few days.

Be ready to go back more than one release, as the AMD owner above did. If two older versions in a row still crash, move on to the next steps.

6. If it crashes on waking up, check VPNs and hibernate

Look at VPN and virtual network adapters first. Switching the VPN protocol, or updating or removing the VPN, fixed the Framework case. For the NVIDIA hibernate bug, owners turn hibernate off and use normal sleep until a fixed driver arrives. NVIDIA hasn’t confirmed that workaround. From an administrator Command Prompt:

powercfg /h off

7. Check your SSD

Update the SSD’s firmware with its maker’s tool, and the storage controller driver from your PC maker. If you own a Crucial drive, read Crucial’s own article on this error before anything else. Some older guides tell you to swap your storage driver for Windows’ standard AHCI driver. They disagree on which way round, so if you try it at all, make a restore point first.

8. Repair system files and test the hardware

These rarely fix a 0x133, but they’re cheap to rule out. Run DISM and then SFC from an administrator Command Prompt, then test your RAM with MemTest86, which runs 4 passes by default:

DISM.exe /Online /Cleanup-image /Restorehealth
sfc /scannow

Clean results don’t clear the drivers. One NVIDIA forum user with a clean SFC, DISM, chkdsk and drive health check still had four graphics-driver crashes under load.

9. Last resort

Try System Restore, then an in-place repair install that keeps your files and apps, then a reset. If the dumps point at a vendor bug, like the NVIDIA hibernate case, report it to the vendor and keep using the workaround until a fixed driver ships.

What not to do

Don’t install the old Windows Server 2012 hotfix KB2789962 that some pages link to: Microsoft wrote it for servers, and it doesn’t apply to Windows 10 or 11. Skip driver-updater and “PC repair” apps. And don’t uninstall your antivirus to test, as one popular guide suggests, without a dump that points at it.

Still on Windows 10?

The same steps apply. Windows 10’s free support ended on 14 October 2025, and paid Extended Security Updates to 12 October 2027 are security-only. Driver fixes still come from NVIDIA, AMD, Intel and your PC maker, so steps 3 and 5 matter most. On Windows 11, 24H2 Home and Pro stopped getting updates on 13 October 2026, so move to 25H2 or 26H2.

Frequently asked questions

Is DPC_WATCHDOG_VIOLATION a hardware problem?

Usually not. Microsoft points to driver code, and in recent threads, hardware swaps like a new power supply didn’t stop it. Test hardware, but don’t start there.

Is it caused by my graphics card?

Often by its driver. Most 2025 and 2026 Microsoft Q&A threads with read dumps named NVIDIA’s nvlddmkm.sys or AMD’s amdkmdag.sys.

Why does it happen after sleep or hibernate?

Drivers wake devices up on resume, and a slow one trips the watchdog. Known cases include a VPN driver and an NVIDIA bug on RTX 5090 cards that NVIDIA says it’s tracking.

Can an SSD cause it?

Sometimes, through its firmware or controller driver, as with one Crucial P1 drive. Update the firmware before changing any drivers.

What to do now

Open Event Viewer, find the last BugCheck in the System log and note Parameter 1. If it’s 0, the dump will very likely name the driver in step 4. If you can’t read dumps and you have a dedicated graphics card, try the clean graphics driver reinstall in step 5: a graphics driver change fixed three of the five solved cases we read.

Sources

Changelog

24 Oct 2026: rewritten from Microsoft’s documentation and 2025 to 2026 crash reports, with the Parameter 1 check and the graphics driver pattern. It replaces a guide first published on 25 March 2018 that covered Windows 10 only.