Module runner 

Per-shape TensorRT warmup runner.

Extracted from trt_warmup.rs so the parent module can stay at facade length while keeping all per-shape logging, cache-hit classification, and engine-count snapshotting in one focused unit.

run_warmup_shape is invoked once per (batch, seq) shape by trt_prewarm (both during the dimensional-extreme coverage check and during the full compile path). ShapeRunResult is the per-call summary the caller aggregates into the worker-scoped PrewarmStats.

Structs

ShapeRunResult 🔒

Result of a single run_warmup_shape call.

Functions

run_warmup_shape 🔒

Runs session.run() for a single (batch, seq) shape, measures wall time, classifies the result as a cache hit or fresh compile, and — on success — fsyncs the engine cache directory for durability.