TDR Debugging

Timeout detection and recovery (TDRs) and device removals occur as a result of a fault in the driver, Direct3D runtime, or hardware. A TDR or device removal causes a reset of the GPU adapter, at which point all device resources are lost. This is typically caused by invalid commands issued due to a bug in the application.

TDRs can be captured, analyzed, and debugged using PIX GPU captures. Note that in order to take a GPU capture for debugging a TDR, you must be able to reproduce the TDR while a GPU capture is in progress. If the TDR is caused by a race condition or other rare, hard to repro problem, it may not be possible to capture it using PIX.

Due to incompatibilities with current drivers, before using PIX to debug a TDR we recommend opening Settings and:

• Unchecking Enable GPU Plugins
• Checking Disable PIX HUD in applications running under GPU capture

Taking a GPU Capture of a TDR

In order to debug or analyze a TDR, a GPU capture must be taken of the frame or workload that produces the TDR. In order to capture a TDR it is recommended to use programmatic capture via the IDXGraphicsAnalysis interface. Programmatic capture allows an application to control when to begin and end GPU capture itself, which is typically the easiest way to capture the specific workload known to cause the TDR. For more information about programmatic capture, see here.

PIX GPU captures are robust against application-caused TDRs and crashes. If an application terminates unexpectedly during capture, PIX will produce a valid GPU capture containing every API call made prior to termination. A capture of an application that suffered a TDR during GPU capture can be identified by the “Device Removal (TDR) at capture” message in the Warnings pane when opening the file.