Hi, I’m the Toradex AI Assistant.
I might make mistakes or occasionally get you wrong. Our support team will step in if needed.
Video Playback Performance in BSP 7.x
Hardware Video Decoding
You’re correct that hardware acceleration has changed between BSP versions. For BSP 7.x, hardware-accelerated video decoding depends on your specific Toradex module and the availability of VPU (Video Processing Unit) support (1).
Optimized GStreamer Pipelines
Instead of using decodebin which may fall back to software decoding, try these hardware-accelerated pipelines depending on your module:
For iMX-based modules with VPU support:
# gst-launch-1.0 filesrc location="/home/video/FMSphysio.mp4" ! qtdemux ! h264parse ! vpudec ! waylandsink
For modules with V4L2 hardware decoder:
# gst-launch-1.0 filesrc location="/home/video/FMSphysio.mp4" ! qtdemux ! h264parse ! queue ! v4l2h264dec ! imxvideoconvert_g2d ! waylandsink
Alternative Video Sink Options
Replace ximagesink with waylandsink for better performance, as Wayland/Weston is supported starting from BSP 5 (1):
# gst-launch-1.0 filesrc location="/home/video/FMSphysio.mp4" ! decodebin ! videoconvert ! waylandsink
Checking Available Hardware Decoders
Verify which hardware decoders are available on your system:
# gst-inspect-1.0 | grep vpudec
# gst-inspect-1.0 | grep h264
Video Format Recommendations
For optimal performance, consider these formats:
- H.264 encoded MP4: Best hardware acceleration support
- Lower compression formats: May reduce CPU load but increase file size (3)
- Uncompressed formats like AVI: Should theoretically take less CPU load (3)
Debugging Performance Issues
Enable GStreamer debugging to identify bottlenecks:
export GST_DEBUG=2
Important Notes
- VPU support availability varies by module and BSP version (4)
- For NXP-based modules, refer to the corresponding NXP BSP documentation for detailed GStreamer plugin information (1)
- Consider using Torizon OS with containerized GStreamer applications for easier multimedia development (1)
**
If I answered your question, please tick the “Solution” checkbox.
For follow-up questions to me, please use @ToradexAI in your post.