Installing Deepstream on V100 or NVIDIA DRIVE AGX

Please provide complete information as applicable to your setup.

Machine 1
NVIDIA DRIVE AGX
DeepStream SDK 5.0
• Tensor RT 5.1.4.2

Machine 2
• GV100GL [Tesla V100 PCIe 16GB]
• DeepStream SDK 5.0
•JetPack Version (valid for Jetson only)
• 5.1.5.0-1+cuda10.0 (Inside Docker)
• Driver Version 435.21

I have two devices as listed above where I would like to install Deepstream for inference on a model trained by Transfer Learning Toolkit. As per the documentation both of my devices are not listed. Is there any way I can use Deepstream on either of the machines mentioned above?

Hi
We maintain and test deepstreamSDK on Tesla series cards and Jetson devices, DRIVE AGX should be with QNX os as I know, and should be not supported, and for V100, you could download the package and give a try.

I have installed the Deepstream app on the V100 machine and I am getting the following error.

cuGraphicsGLRegisterBuffer failed with error(219) gst_eglglessink_cuda_init texture = 1
0:00:02.861790110 114804 0x56551d89d1e0 WARN                 nvinfer gstnvinfer.cpp:1946:gst_nvinfer_output_loop:<primary_gie> error: Internal data stream error.
0:00:02.861812109 114804 0x56551d89d1e0 WARN                 nvinfer gstnvinfer.cpp:1946:gst_nvinfer_output_loop:<primary_gie> error: streaming stopped, reason not-negotiated (-4)
ERROR from primary_gie: Internal data stream error.
Debug info: gstnvinfer.cpp(1946): gst_nvinfer_output_loop (): /GstPipeline:pipeline/GstBin:primary_gie_bin/GstNvInfer:primary_gie:
streaming stopped, reason not-negotiated (-4)
Quitting
ERROR from sink_bin_queue: Internal data stream error.
Debug info: gstqueue.c(988): gst_queue_handle_sink_event (): /GstPipeline:pipeline/GstBin:processing_bin_0/GstBin:sink_bin/GstQueue:sink_bin_queue:
streaming stopped, reason not-negotiated (-4)
ERROR from qtdemux2: Internal data stream error.
Debug info: qtdemux.c(6073): gst_qtdemux_loop (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstBin:src_sub_bin5/GstURIDecodeBin:src_elem/GstDecodeBin:decodebin5/GstQTDemux:qtdemux2:
streaming stopped, reason not-negotiated (-4)
ERROR from qtdemux22: Internal data stream error.
Debug info: qtdemux.c(6073): gst_qtdemux_loop (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstBin:src_sub_bin22/GstURIDecodeBin:src_elem/GstDecodeBin:decodebin22/GstQTDemux:qtdemux22:
streaming stopped, reason not-negotiated (-4)
ERROR from qtdemux28: Internal data stream error.
Debug info: qtdemux.c(6073): gst_qtdemux_loop (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstBin:src_sub_bin27/GstURIDecodeBin:src_elem/GstDecodeBin:decodebin27/GstQTDemux:qtdemux28:
streaming stopped, reason not-negotiated (-4)
ERROR from qtdemux12: Internal data stream error.
Debug info: qtdemux.c(6073): gst_qtdemux_loop (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstBin:src_sub_bin12/GstURIDecodeBin:src_elem/GstDecodeBin:decodebin12/GstQTDemux:qtdemux12:
streaming stopped, reason not-negotiated (-4)
App run failed

I am getting this error after trying to run the example caffe model by the following command:-

deepstream-app -c source30_1080p_dec_infer-resnet_tiled_display_int8.txt

You may change sink type to 1 Fakesink or 3 File or 4 RTSPStreaming in config.
another option, V100 is compute card, you may follow this Deepstream/FAQ - eLinux.org 5A to setup virtual display, the FAQ is for T4 or P4, but you may give a try.