SHM Consumers

Overview

The SHM transport writes raw BGRA8 frames into a named shared-memory ring buffer for same-machine consumers. A reader attaches by stream name, pulls frames, and gets a zero-copy view directly into the shared region — no CUDA involved on the consumer side.

Multiple readers can attach to one producer concurrently. Readers can come and go independently of the producer.

The unified ovstream.Client / ovstream_create_client is backend-agnostic — the transport is chosen by ClientType. This skill covers ClientType.SHM; the same Client also speaks ClientType.CUDASHM (GPU-resident, see the cudashm-consumers skill) and ClientType.NATIVE (network, see protocol-selection).

Python: ovstream.Client(ovstream.ClientType.SHM)

Source: examples/python/local_stream/main.py snippet shm-consumer

The minimal pattern: