simaaiprocesscvu

The simaaiprocesscvu GStreamer plugin processes buffers by executing a SiMa.ai graph on the CVU (e.g., EV74). It acts as an aggregator, collecting input buffers from multiple sources and producing a single output buffer. The plugin utilizes SiMa.ai’s memory management and dispatcher for efficient execution on the hardware.

Properties

Property	Type	Default Value	Description
config	string	/mnt/host/evxx_pre_proc.json	Path to the JSON configuration file for the SiMa.ai graph.
num-buffers	uint	5	Number of output buffers to allocate in the buffer pool.
dump-data	boolean	FALSE	If TRUE, saves output buffers as binary dumps in /tmp/{node-name}-%03d.out.
silent	boolean	TRUE	If TRUE, suppresses verbose output messages.
transmit	boolean	FALSE	If TRUE, transmits KPI messages to the GstBus.

Usage

The simaaiprocesscvu plugin integrates into a GStreamer pipeline as a transformer/aggregator element. It requires input buffers with SiMa.ai metadata. Here’s an example gst-launch-1.0 command:

simasrc location="input" node-name="allegrodec" blocksize="size_of_input" \
! simaaiprocesscvu name=cvu config="/path/to/your/config.json" source-node-name="input_image_name" buffers-list="a65-topk,test" \
! fakesink

• Replace /path/to/your/config.json with the actual path to your JSON configuration file.

• source-node-name specifies the name of the source node in your SiMa.ai graph.

• buffers-list is a comma-separated list of buffer names that serve as input for the graph. This should match names defined in the config file.

• The name property assigns a name to the plugin instance within the pipeline, used for debugging and metadata.

Installation

To install the simaaiprocesscvu plugin, copy the plugin files (located in /usr/local/simaai/plugin_zoo/gst-simaai-plugins-base/gst/) into your project’s plugins/ directory. Ensure that the necessary SiMa.ai libraries are correctly installed and configured on your system. The plugin will then be automatically discovered by GStreamer.