A problem about docker image of deepstream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I run docker images like:

sudo docker run -it --net=host --runtime nvidia  -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/deepstream-l4t:4.0.2-19.12-samples

I run some docker images in my the edge gateway,I want to allocate the specific gpu resource size for every docker images.
Can I allocate a gpu resource of the specified size to docker image.

2

Hi,

Sorry, you won’t be able to do that.
Please check this page for information:
https://github.com/NVIDIA/nvidia-docker/wiki#can-i-limit-the-gpu-resources-eg-bandwidth-memory-cuda-cores-taken-by-a-container
----------------------------------------------------------------------------------------------
Can I limit the GPU resources (e.g. bandwidth, memory, CUDA cores) taken by a container?
No. Your only option is to set the GPU clocks at a lower frequency before starting the container.
----------------------------------------------------------------------------------------------

Thanks.

3

HI AastaLLL

Thank you AastaLLL

Will you support this function in future releases?

4

Hi,

Sorry that this support is not in our future list yet.
Thanks.

5

Hi AastaLLL
when I run two docker container,one of both occur error follow:

NvxBaseWorkerFunction[2575] comp OMX.Nvidia.std.iv_renderer.overlay.yuv420 Error -2147479552
.
.
.

The progress of starting container is:

  1. export DISPLAY=:0
  2. xhost +
  3. sudo docker run -it --net=host --runtime nvidia -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/deepstream-l4t:4.0.2-19.12-iot
  4. deepstream-test5-app -c source12_1080p_dec_infer-resnet_tracker_tiled_display_fp16_tx2.txt

The other one is same as it.
when I try use other display,“export DISPLAY=:1”,it still don’t work.

Last question is that can I not use x11? In fact,I needn’t display in local screen.

7

Hi,

You can try to launch the container with this command:

$ sudo docker run -it --net=host --runtime nvidia  -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/deepstream-l4t:4.0.2-19.12-iot

We can run the step.4 command in the above environment without error.
Thanks.

8

Hi AastaLLL,

When I run first container,it is correct.But When I run second container,it occur the error.
Both programs, “deepstream-test5-app” ,are running at the same time.

Can I run more than one program,“deepstream-test5-app”,at a time.

11

Hi,

We can reproduce this issue in our environment.
The root cause is the second deepstream-test5-app will fail when the display is occupied.

An alternative is to change one of the config into FakeSink.
We can launch two deepstream-test5-app without after applying this update.

[sink0]
enable=1
#Type - 1=FakeSink 2=EglSink 3=File
type=1
sync=1
source-id=0
gpu-id=0
qos=0
nvbuf-memory-type=0
overlay-id=1

Thanks.