Device Error Code Diagnostics
Focused procedures for chasing runtime and device failures live in this
directory. Start with Device Error Codes to identify
the reported code or watchdog signature, then use the matching guide below.
Capacity and progress failures
Document
What it covers
Capacity Codes Diagnosing codes 1, 2, 3, 4, and 11 across HBG fixed capacity and TRB reclaim
Stalls Diagnosing scheduler timeout code 100 and tensor-wait code 8 by ordering watchdogs and locating the stuck task
Core faults and defensive codes
Document
What it covers
AICore Faults Separating kernel addressing faults from control-flow corruption using device logs and static inspection
Codes Without End-to-End Tests Why codes 10, 11, 103 and stall classes S4, S5, and unknown cannot be triggered through the public API
Document
What it covers
Local Runtime Timeouts Default watchdog ordering and local overrides used to expose the most useful failure code
Debug a Failed Run User-facing first-response workflow for collecting and classifying failure evidence
Core Swimlane Profiling Intra-core task timing for kernels that run but are unexpectedly slow