Is My NVIDIA RTX 3070 Failing? Investigating Repeated LiveKernelEvent 193 (VIDEO_DXGKRNL_LIVEDUMP)

Background

I’m troubleshooting repeated Windows Reliability Monitor reports involving LiveKernelEvent 193, specifically VIDEO_DXGKRNL_LIVEDUMP.

I’m trying to determine whether my NVIDIA GeForce RTX 3070 is beginning to fail, or whether these reports are related to the NVIDIA driver, Windows graphics subsystem, or another system component.

I recently performed a fresh installation of Windows, so there are currently only a limited number of dump files available.

Important: I am not currently experiencing any noticeable symptoms. There are no black screens, display flickering, freezes, unexpected restarts, application crashes, OBS encoder errors, or obvious performance problems.

My concern is whether these events could indicate an early hardware problem before symptoms become apparent.

---

System Specifications

---

Primary Issue

Windows Reliability Monitor reports:

• Event: LiveKernelEvent
• Code: 193
• Description: VIDEO_DXGKRNL_LIVEDUMP
• Reported reason/argument: 0x80e
• Additional text: 100 Internal

Dump file:

C:WindowsLiveKernelReportsWATCHDOGWATCHDOG-20261001-1503.dmp

Multiple Reliability Monitor reports appeared between approximately 3:03 PM and 3:41 PM on October 1, 2026, referencing the same diagnostic dump.

I would like to understand whether these are separate failures or repeated reports associated with the same underlying event.

---

WinDbg Analysis

I opened the dump using WinDbg version:

10.0.29617.1000

The debugger reported the following bucket:

LKD_0x193_dxgkrnl!DxgCreateLiveDumpWithWdLogs2

Relevant Stack Entries

dxgkrnl!DxgCreateLiveDumpWithWdLogs2
dxgkrnl!DxgCreateLiveDumpWithWdLogs
NtDxgkPinResources

Additional Findings

• Process associated with the dump: dwm.exe
• System uptime at the time of the dump: approximately 3 minutes and 8 seconds
• Dump type: Live kernel triage dump
• Dump contains limited register and stack information
• NVIDIA kernel driver nvlddmkm.sys was loaded
• The dump does not provide enough information to establish a definitive hardware failure

The presence of dwm.exe and the graphics kernel functions has made me question whether this is a GPU problem or an issue involving Windows Desktop Window Manager or the DirectX graphics subsystem.

I understand that the stack alone does not establish which component initiated the underlying problem.

---

NVIDIA Driver and Device Verification

I checked the installed NVIDIA driver packages.

The Driver Store contains three NVIDIA display driver packages:

Using PowerShell and Windows device information, I confirmed that the RTX 3070 is currently associated with oem52.inf.

The device reports:

• Driver signed: Yes
• Device status: OK
• NVIDIA driver version: 32.0.16.1714
• Hardware ID matches the RTX 3070

The presence of older driver packages in the Driver Store has been observed, but I have not established that they are causing a conflict.

---

NVIDIA-SMI Monitoring

I checked the GPU using nvidia-smi.

Initial Readings

Approximately 58 Seconds of Monitoring

The GPU responded normally to NVIDIA-SMI queries during this monitoring period.

No abnormal temperature or obvious failure was observed during this limited, relatively low-load monitoring.

---

Windows Event Viewer Checks

WHEA Hardware Errors

Checked the WHEA-Logger provider over the preceding seven days.

Result: No WHEA events were returned.

Display Driver Recovery

Checked for Event ID 4101 over the preceding seven days.

Result: No records were returned.

NVIDIA Driver Events

Checked the nvlddmkm provider over the preceding seven days.

Result: No records were returned.

DirectX Graphics Kernel Logs

Checked the following logs:

• Microsoft-Windows-DxgKrnl/Operational
• Microsoft-Windows-DxgKrnl/Admin

Both were enabled, but their reported record counts were zero.

Kernel-Power Events

Found informational events including IDs 109, 577, and 172.

The available event descriptions for IDs 109 and 577 indicate system-initiated restart activity through the Windows Kernel API.

I have not found evidence connecting these informational events to a GPU hardware failure or an unexpected power loss.

Kernel-PnP Event ID 219

Repeated events were found involving:

HIDVID_1B1C&PID_0A71&MI_03&Col03

The reported status was:

0xC0000365

The message indicated that WUDFRd failed to load.

I have not established a relationship between this HID device event and the graphics dump.

---

What Has Not Been Observed

At this point, I have not observed:

• GPU artifacts
• Black screens
• Display signal loss
• Display driver recovery notifications
• System freezes
• Unexpected GPU-related restarts
• OBS NVENC failures
• Abnormal GPU temperatures
• WHEA hardware errors
• A confirmed PCIe error
• A confirmed GPU memory error

I have also not yet completed a dedicated GPU stress test or VRAM test.

---

What I Am Trying to Determine

My primary concern is whether the RTX 3070 could be experiencing an early hardware failure despite appearing to operate normally.

I would appreciate technical insight into the following:

1. Does VIDEO_DXGKRNL_LIVEDUMP (193) with reason 0x80e indicate a specific type of graphics hardware failure, or is it a more general graphics kernel diagnostic event?

2. Does the WinDbg bucket LKD_0x193_dxgkrnl!DxgCreateLiveDumpWithWdLogs2 provide any meaningful indication of a failing GPU?

3. Does the presence of NtDxgkPinResources and dwm.exe suggest a particular graphics resource-management issue?

4. Could this be related to the NVIDIA driver, Windows Desktop Window Manager, DirectX, or the Windows graphics kernel rather than physical GPU failure?

5. What additional logs, dump analysis, or diagnostic tests would help distinguish a failing GPU from a software issue?

6. Are there specific NVIDIA, Windows, PCIe, or hardware monitoring logs that I should collect before considering hardware replacement?

7. Given that the GPU currently reports normal temperatures, utilization, power, and device status, is there any evidence here that specifically points toward hardware degradation?

---

My Goal

I want to establish whether my RTX 3070 is actually failing before spending money replacing it unnecessarily.

At present, I have a recurring graphics-related LiveKernelEvent report but no observable GPU symptoms.

I am looking for an evidence-based way to determine whether this is a hardware concern, a driver issue, or a Windows graphics subsystem event.