deepstream-test4-app metadata in T4

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi Sir/Madam,

I try to run “deepstream-test4-app” inside deepstream 4.0 package in my T4 GPU. However, I found some error as below

My Input:

deepstream-test4-app -i /home/ubuntu/Server/deepstream-4.0/samples/streams/sample_720p.h264 -p /opt/nvidia/deepstream/deepstream-4.0/lib/libnvds_kafka_proto.so --conn-str=“172.0.0.1;9092;ubuntuDeepstreamAppMsg” -s 1 --topic ubuntuDeepstreamAppMsg

I got error as below:

Now playing: /home/ubuntu/Server/deepstream-4.0/samples/streams/sample_720p.h264
Creating LL OSD context new
0:00:00.544274958 48008 0x560c9a1e4a40 INFO                 nvinfer gstnvinfer.cpp:519:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:initialize(): Trying to create engine from model files
0:00:02.214752378 48008 0x560c9a1e4a40 WARN                 nvinfer gstnvinfer.cpp:515:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:log(): TensorRT was compiled against cuBLAS 10.2.0 but is linked against cuBLAS 10.1.0
0:00:12.188906720 48008 0x560c9a1e4a40 WARN                 nvinfer gstnvinfer.cpp:515:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:log(): TensorRT was compiled against cuBLAS 10.2.0 but is linked against cuBLAS 10.1.0
0:00:12.228602896 48008 0x560c9a1e4a40 INFO                 nvinfer gstnvinfer.cpp:519:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:generateTRTModel(): Storing the serialized cuda engine to file at /home/ubuntu/Server/deepstream-4.0/samples/models/Primary_Detector/resnet10.caffemodel_b1_int8.engine
0:00:12.237860474 48008 0x560c9a1e4a40 WARN                 nvinfer gstnvinfer.cpp:515:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:log(): TensorRT was compiled against cuBLAS 10.2.0 but is linked against cuBLAS 10.1.0
0:00:12.240571545 48008 0x560c9a1e4a40 WARN                 nvinfer gstnvinfer.cpp:515:gst_nvinfer_logger:<primary-nvinference-engine> NvDsInferContext[UID 1]:log(): TensorRT was compiled against cuBLAS 10.2.0 but is linked against cuBLAS 10.1.0
Running...
Creating LL OSD context new
cuGraphicsGLRegisterBuffer failed with error(999) gst_eglglessink_cuda_init texture = 1
Frame Number = 0 Vehicle Count = 3 Person Count = 2
Frame Number = 1 Vehicle Count = 3 Person Count = 2
0:00:12.586206627 48008 0x560c9a163a80 WARN                 nvinfer gstnvinfer.cpp:1830:gst_nvinfer_output_loop:<primary-nvinference-engine> error: Internal data stream error.
0:00:12.586219031 48008 0x560c9a163a80 WARN                 nvinfer gstnvinfer.cpp:1830:gst_nvinfer_output_loop:<primary-nvinference-engine> error: streaming stopped, reason not-negotiated (-4)
ERROR from element primary-nvinference-engine: Internal data stream error.
Error details: gstnvinfer.cpp(1830): gst_nvinfer_output_loop (): /GstPipeline:dstest4-pipeline/GstNvInfer:primary-nvinference-engine:
streaming stopped, reason not-negotiated (-4)
Returned, stopping playback
Frame Number = 2 Vehicle Count = 4 Person Count = 2
0:00:12.591713700 48008 0x560c9a163a80 WARN                 nvinfer gstnvinfer.cpp:1830:gst_nvinfer_output_loop:<primary-nvinference-engine> error: Internal data stream error.
0:00:12.591725729 48008 0x560c9a163a80 WARN                 nvinfer gstnvinfer.cpp:1830:gst_nvinfer_output_loop:<primary-nvinference-engine> error: streaming stopped, reason not-linked (-1)
Frame Number = 3 Vehicle Count = 4 Person Count = 2
Frame Number = 4 Vehicle Count = 3 Person Count = 2
Frame Number = 5 Vehicle Count = 3 Person Count = 2
Frame Number = 6 Vehicle Count = 3 Person Count = 2

So, without “–no-display”, it cannot work well. May I know how to fix the display problem? Is that i need to change to “RTSP” in default to fix it? if so, may i know the “variable naming” of rtsp in sink group inside C programming?

My Environment:
Ubuntu 18.0
GPU T4 with driver 418.56
Cuda: 10.1
Gstreamer: 5.1.5

THanks

2

I have similar questions, anyone can answer

3

If change nveglglessink to be fakesink, there is no this issue ?

4

I got the similar issue when I tried to run “deepstream-app -c samples/configs/deepstream-app/source4_1080p_dec_infer-resnet_tracker_sgie_tiled_display_int8.txt” with Deepstream 4.0 docker images on T4. If I change the sink0 from EglSink to FakeSink, this issue won’t happen.

My Environment:
Ubuntu 18.04.2
GPU T4 with driver 418.70
Cuda: 10.1
Gstreamer: 5.1.5

Thank you.

5

Yes, if I use fakesink, no such problem. Is that T4 GPU not support “EglSink”?

6

T4 does not have display port, but you can follow this Deepstream/FAQ - eLinux.org
to setup virtual display, then you can use nveglglessink.