(see https://github.com/sylvain-prevost/deepstream_wsl for similar example tested with Deepstream 6.2/6.3 + python support)
nvdsgst_infer plugin adjustments to further augment Deepstream support (debugging/execution) under WSL2 when the host possess an nVidia dGPU. See nVidia driver 511.79 release note for list of compatible devices (https://developer.nvidia.com/cuda/wsl).
This is a work-around for lack of EGL symbols in the cuda for WSL2 library - it is extremelly basic as it consist in recompiling without the EGL calls (which can be safely excluded as they relate solely to Jetson devices).
Per SOFTWARE LICENSE AGREEMENT FOR NVIDIA SOFTWARE DEVELOPMENT KITS, 1.2 (iii):
This software contains source code provided by NVIDIA Corporation.
How to build the plugin (Within WSL2)
You need to follow nVidia DEEPSTREAM SDK v6.0 instructions for dGPU
CUDA_VER, GST_INSTALL_DIR & LIB_INSTALL_DIR are environment variables that are expected by the makefile.
Typical values would be:
CUDA_VER=11.4
GST_INSTALL_DIR=/opt/nvidia/deepstream/deepstream-$(NVDS_VERSION)/lib/gst-plugins/
LIB_INSTALL_DIR=/opt/nvidia/deepstream/deepstream-$(NVDS_VERSION)/lib/
Execute the following within WSL2
git clone https://github.com/sylvain-prevost/nvdsgst_infer-wsl2.git
cd nvdsgst_infer-wsl2
make clean
make ENABLE_WSL2=1
sudo make install
This will install the updated plugin at the GST_INSTALL_DIR location.
Due to problems with 'nvv4l2decoder' (not-yet-understood) and 'nveglglessink' (somewhat expected) under WSL2,
the deepstream-test1 has been modified with the following flow:
- filesource
- h264parse
- avdec_h264 [instead of nvv4l2decoder]
- nvvideoconvert (*addition to compensate absence of nvv4l2decoder)
- nvstreammux
- nvinfer
- nvvideoconvert
- nvdsosd
- nvvideoconvert (*addition to compensate absence of nveglglessink)
- autovideosink [instead of nveglglessink]
PS:
Do not forget to set DISPLAY environment variable in WSL2 to be able to visualize on Windows Host. :)
This has been tested on the following system:
-
Host:
- WIN10 21H2 (OS Build 19044.1526)
- VcXsrv v1.20.9.0 (XLaunch)
- [optional] Debug config:
- Visual Studio 2022 v17.2 Preview 1 (with CMakePresets)
- vcpkg
-
NVIDIA
- GeForce GTX 1080
- Driver v511.79
-
WSL2:
- Kernel version: 5.10.60 (5.10.60.1-microsoft-standard-WSL2)
- Ubuntu 20.04
- TensorRT 8.2.3.0-ga-20220113
- DeepStream SDK v6
- gstreamer 1.19.2 (compiled from sources)
- gcc Ubuntu 9.3.0-17ubuntu1~20.04