hpx engines#

List all inference engines available to the profiler.

Synopsis#

hpx engines

Output#

  tflm
  helia-rt
  helia-aot
  executorch

• tflm — vanilla TFLM baseline through the nsx-tflite-micro port; supports reference and upstream cmsis_nn backends.
• helia-rt — AmbiqAI heliaRT, an optimized TFLM fork (HELIA / CMSIS-NN / reference backends).
• helia-aot — AmbiqAI heliaAOT, an ahead-of-time compiler (no interpreter at runtime).
• executorch — Cortex-M ExecuTorch programs (.pte) through nsx-executorch, with the arm or ns CMSIS-NN provider.

The first three consume LiteRT (.tflite) models; executorch consumes .pte programs. Use any of these names with --engine or engine.type: in YAML.

See also#

• Inference Engines — full description of each engine, when to use it, and engine-specific config.