Performance measurements on the NVIDIA Jetson Platforms
This wiki presents some performance measurements of a typical GstRtspSink pipeline on the NVIDIA Platforms (Nano, TX2, Xavier, and Orin).
|
|
Problems running the pipelines shown on this page? Please see our GStreamer Debugging guide for help.
The pipeline ran without clients; after 1h, one client was connected to the stream.
GstRtspSink Pipeline
The following GstRtspSink pipeline generates video with a videotestsrc, encodes it using the NVIDIA HW accelerated H264 encoder, and streams it using GstRtspSink.
PORT=5000
MAPPING=/stream1
gst-launch-1.0 videotestsrc is-live=true ! video/x-raw, width=1920, height=1080, framerate=60/1 ! queue ! nvvidconv ! nvv4l2h264enc insert-sps-pps=true idrinterval=30 ! perf print-cpu-load=true ! video/x-h264, mapping=${MAPPING} ! rtspsink service=${PORT}
 | The perf element is optional. The perf element prints frame rate and CPU usage debug information. |
CPU, Framerate and bitrate usage
The following table summarize the CPU, Framerate and bitrate reported by perf element.
| Platform | Framerate | CPU | Bitrate |
|---|---|---|---|
| Jetson AGX Thor | 60.001 | 3 | 4155600 |
| Jetson AGX Orin | 59.929 | 9 | 4003336 |
| Jetson Xavier AGX | 59.996 | 44 | 6630958 |
| Jetson Nano | 52,141 | 32 | 3517957 |
Jetson TX2
Test run for 2:40 hours.
Framerate
The frame rate was measured using the perf element.
CPU Load
The CPU load was measured using perf element.
RAM Usage
The ram usage was measured using massif.
