feat: add per-pod GPU SM utilization metrics for time-slicing workloads#638

Closed

zbennett10 wants to merge 7 commits into

NVIDIA:mainfrom

zbennett10:feat/per-pod-gpu-util-time-slicing

Commits on Mar 2, 2026

feat: add per-pod GPU SM utilization metrics for time-slicing workloads
zbennett10
committed
ci: add unit-tests job for per-pod GPU collector
zbennett10
committed
ci: trigger workflow on feat/** branches for pre-PR validation
zbennett10
committed
fix: resolve os import conflict and podResourcesClient interface mismatch
zbennett10
committed
fix: remove //go:generate directives for unexported interfaces; scope CI tests
zbennett10
committed
fix(collector): use built-in counter default, no CSV required
zbennett10
committed
fix(collector): populate GPU device metadata labels in per-pod metric
zbennett10
committed