• Hardware Platform: GPU
• DeepStream Version: 6.2
• TensorRT Version: 184.108.40.206
• NVIDIA GPU Driver Version: 525.85.12
• Issue Type: Bug
We have an RTSP source where the nvv4l2decoder hangs after receiving the first buffer. This problem occurs on A100 only. We have tested with several different GPU:s and identical configuration with the decoder handling the stream without any issues. I have attached an mp4 example which reproduces this issue. The H264 video in the file is dumped from the RTSP source.
Thank you for reporting the issue, we will look into it.
I’ve tried with A100 machine, no decoder issue found with your video. Can you use the following command to dump the output?
gst-launch-1.0 filesrc location=./a100-repro.mp4 ! qtdemux ! h264parse ! nvv4l2decoder ! nvvideoconvert ! 'video/x-raw,format=RGB' ! filesink location=out_a100.rgb
Thanks for testing! At first we got the same problem with the command you sent, but after further investigation we realized that the reason for this problem is that we have the wrong version of libnvidia-encode installed in our docker image. After changing the docker file to not install libnvidia-encode the decoding works well on A100.
I’m sorry for reporting this as an error as the reason was actually that we built an incorrect image! We got confused by the image working well for other streams and this specific stream working for other GPU:s.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.