• GPU RTX 2080 Ti
• DeepStream Version 6.1
• TensorRT Version 8.6
• NVIDIA GPU Driver Version 515
• Slow Performance
• Language C++
I’m encountering a performance discrepancy while working with GStreamer pipelines in C++. Specifically, I’m dealing with two pipelines that share the same plugins but handle different sources. One pipeline utilizes filesrc, h264parse, and nvv4l2decoder elements and runs smoothly at 60 fps on my x86 PC. The other pipeline, which incorporates v4l2src, capsfilter, videoconvert, nvvideoconvert, and another capsfilter, is only running at 5 fps.
While I understand that v4l2src might not be the most efficient source element, my ultimate goal is to handle a 3D camera stream or use the appsrc plugin. Despite the performance difference, I’m determined to optimize the second pipeline for better frame rates.
Could someone kindly help me understand the reasons behind this performance gap and suggest any optimizations I could apply to my C++ code? I greatly appreciate any insights or advice you can provide.
Thank you for your time and assistance!