ADRT_SLAVE(1)

NAME

adrt_slave - ADRT distributed ray-tracing rendering worker

On this page

SYNOPSIS
DESCRIPTION
OPTIONS
EXAMPLES
SEE ALSO
AUTHOR
COPYRIGHT
BUG REPORTS

SYNOPSIS

adrt_slave [-v] [-h] [-p port] [-t threads] [host]

adrt_slave is a rendering worker for the Advanced Distributed Ray Tracer (ADRT). It receives geometry and per-tile work units over a network socket, ray-traces them with the librender TIE engine, and returns the rendered pixels (or shotline, component, and bounding-box results) to the dispatcher. Multiple slaves cooperating with a single dispatcher allow a single image to be rendered in parallel across many machines.

If a host argument is given, adrt_slave connects to the ADRT dispatcher running on that host (default port 1980). If no host is given, the slave runs as a daemon and listens for an incoming connection (default port 1981). The geometry to render is supplied by the dispatcher, adrt_master(1); the network rendering session is driven by isst(1).

OPTIONS

-p port: Use the given TCP port instead of the protocol default (1980 when connecting to a host, 1981 when running as a daemon).
-t threads: Launch threads rendering threads per slave. The value is clamped to the range 0-32; 0 lets the engine choose based on the available cores.
-v: Print the ADRT version and exit.
-h: Print a usage summary and exit.

EXAMPLES

Example 1. Run a slave as a daemon awaiting work

adrt_slave -t 8 -p 1981

Start a worker that listens on port 1981 using eight rendering threads. A dispatcher (for example, one launched from isst(1)) can then connect to this host and hand it tiles to render.

Example 2. Connect a slave to a running dispatcher

adrt_slave -t 16 dispatcher.example.com

Connect to the ADRT dispatcher on the host dispatcher.example.com (port 1980) and contribute sixteen rendering threads to the rendering pool. Launching this command on several machines pointed at the same dispatcher spreads the work of a single frame across the cluster.