[CVS] GPU metrics polling integration for inference validation suites#241

Open

atnair-amd wants to merge 7 commits into

atnair/dtni-gpu-api

Commits on Jun 24, 2026

feat(gpu): add GPU polling loop, metrics, and threshold gates
atnair-amd
committed

Commits on Jun 25, 2026

fix(vllm_single): add missing gpu_metrics_snap module-scope fixture
atnair-amd
committed
fix(gpu): use exec_on_head for amd-smi; mkdir gpu_poll.log parent dir
atnair-amd
committed
fix(vllm_single): write gpu_poll.log to tmp then copy to node via exec_on_head
atnair-amd
committed
fix(gpu): move deferred imports to module level; fix test_gpu_metric rank
atnair-amd
committed
docs(gpu): add gpu.py section to AGENTS.md and integration guide
atnair-amd
committed
fix(vllm_single): write gpu_poll.log to local report dir so it lands in zip bundle
atnair-amd
committed