API Reference hexagon_tpu v#0.1.2

Elixir bindings for the Qualcomm Hexagon NPU (HTP) via the QNN (Qualcomm AI Engine Direct) runtime.

A loaded QNN context: a pre-compiled context binary (produced on the host with qnn-context-binary-generator) deserialized onto the backend.

An executable graph retrieved from a loaded context.

Metadata for one graph in a loaded context: its name and I/O tensors.

Periodic guard for the NIF layer. Add it to your supervision tree

Quantize/dequantize helpers driven by HexagonTpu.TensorInfo metadata.

A QNN backend runtime: the dlopen'd backend library plus its backend and device handles.

Observability for the NIF layer: native resource accounting (leak guard), dirty-scheduler utilization, and memory snapshots.

Metadata for one graph input/output tensor, extracted from the QNN context binary.