P1-FEATURE-013: Profiling Integration — IHV Tools (QC / Intel / AMD)

Summary

Integrate ModelKit with IHV-specific profiling tools from Qualcomm, Intel, and AMD to enable operator-level performance analysis on hardware NPUs/GPUs.

Context

IHV profiling tools provide deep insights into hardware execution that generic ONNX Runtime profiling cannot — e.g., which NPU ops caused stalls, memory bandwidth bottlenecks, or kernel dispatch overhead. This integration is needed to make performance optimization actionable.

From plans/release/0501_release_plan/P0_CHECKLIST.md (P1-FEATURE-013). Builds on the base profiling work in #402.

Target tools:

• Qualcomm: QNN profiling API (HTP backend profiling output)
• Intel: OpenVINO Performance Analysis tool / VTune
• AMD: ROCm profiler / Ryzen AI profiler

Current State

• Base profiling implemented (#402) — measures E2E inference latency
• No IHV-specific profiling tool integration
• No operator-level profiling data from hardware backends

Desired State

• wmk perf --profile enables IHV profiling output for supported EPs
• Profiling data consumed from QNN/OpenVINO/AMD tools
• Bottleneck analysis: identify top-N slowest operators per EP
• Output: profiling summary in artifacts/profiling_report.json

Acceptance Criteria

• QNN profiling output integrated (EP session profiling via QNN SDK)
• OpenVINO profiling output integrated (per-layer timing via IE API)
• AMD profiling output integrated (best-effort — if SDK available)
• Bottleneck analysis: report top-N slowest ops per EP
• artifacts/profiling_report.json generated with operator-level timing
• Works on at least 5 P0 built-in models × 2 EPs (QNN + OpenVINO NPU)

Technical Notes

• QNN SDK profiling: enable via session option qnn_context_enable_graphs_profiling
• OpenVINO profiling: InferRequest.get_profiling_info() method
• AMD Ryzen AI profiling: available via Ryzen AI SDK; check access with hardware team
• Normalize profiling output to a common schema across all IHV tools

Related Files

• plans/release/0501_release_plan/P0_CHECKLIST.md — P1-FEATURE-013
• plans/release/0501_release_plan/feature-scale.md
• #402 — base profiling implementation