• Hardware Platform: GPU (NVIDIA A2)
• DeepStream Version: 5.1
• NVIDIA GPU Driver Version: 495.46
• Issue Type: bugs
• Issue: Pipeline containing accelerated GPU decoding element (nvv4l2decoder) fails to run on an NVIDIA A2 GPU.
• How to reproduce the issue:
I’m using the official docker image:
nvcr.io/nvidia/deepstream:5.1-21.02-triton
After starting a container with the following command:
docker run -it --gpus all nvcr.io/nvidia/deepstream:5.1-21.02-triton /bin/bash
And trying to run the following pipeline:
gst-launch-1.0 -e filesrc location="/opt/nvidia/deepstream/deepstream/samples/streams/sample_1080p_h264.mp4" ! qtdemux ! h264parse ! nvv4l2decoder cudadec-memtype=0 ! fakesink
I get this output that contains an error:
Setting pipeline to PAUSED ...
Pipeline is PREROLLING ...
ERROR: from element /GstPipeline:pipeline0/nvv4l2decoder:nvv4l2decoder0: Device '/dev/nvidia0' failed during initialization
Additional debug info:
gstv4l2object.c(4116): gst_v4l2_object_set_format_full (): /GstPipeline:pipeline0/nvv4l2decoder:nvv4l2decoder0:
Call to S_FMT failed for H264 @ 1920x1080: Unknown error -1
ERROR: pipeline doesn't want to preroll.
Setting pipeline to NULL ...
Freeing pipeline ...
The same exact pipeline works without issues on other GPUs.
Is the nvv4l2decoder not supported on the A2 GPU?