Getting Started - Evaluating PVA ISP
This section provides a guide to obtaining, installing, and testing the RidgeRun PVA ISP evaluation version.
Step 1: Request the Evaluation Binaries
RidgeRun provides a fully-featured evaluation version of the plugin with a limited processing time, as shown in table 1:
| Feature | Professional | Evaluation |
|---|---|---|
| PVA ISP Plugin | Y | Y |
| Unlimited Processing Time | Y | N (*) |
| Source Code | Y | N |
(*) With the evaluation version, you will have access to all the features for a limited period of time.
In order to request an evaluation binary for a specific architecture, please Contact Us by providing the following information:
- Platform (i.e.: i.Jetson AGX Orin, etc...)
- L4T Version (i.e.: 36.4.4)
- GStreamer version and kernel version, you can provide the output of the following commands:
gst-launch-1.0 --gst-version uname -a
Step 2: Install the dependencies
The following packages are required on the Jetson target for both evaluation and deployment:
| Package | Purpose |
|---|---|
nvidia-l4t-pva
|
PVA support package matching the installed JetPack/L4T version |
pva-sdk-2.9-l4t
|
PVA SDK 2.9 runtime |
pva-allow-2 version 2.0.5
|
Enables PVA ISP execution on the PVA |
yaml-cpp
|
Required for sample examples |
cuda
|
The CUDA version of the system. |
rrv4l2src
|
The plugin for MIPI cameras with a patch for Bayer. |
Install CUDA
CUDA is required when using the GStreamer integration.
Install CUDA from the JetPack repositories:
sudo apt update sudo apt install nvidia-cuda sudo apt install nvidia-cuda-dev
Alternatively, CUDA can be installed through NVIDIA SDK Manager by selecting the CUDA components for the corresponding JetPack version.
Verify the version as follows.
nvcc --version
Install PVA Dependencies
The PVA ISP requires the NVIDIA PVA driver, cuPVA runtime, and authorization utilities.
Install the PVA driver and authorization package:
sudo apt update sudo apt install nvidia-l4t-pva sudo apt install pva-allow-2=2.0.5 sudo apt install libyaml-cpp-dev
For JetPack 6, download the PVA SDK 2.9 runtime package manually from:
https://repo.download.nvidia.com/jetson/
wget https://repo.download.nvidia.com/jetson/common/pool/main/p/pva-sdk-2.9-l4t/pva-sdk-2.9-l4t_2.9.4_arm64.deb
sudo apt install ./pva-sdk-2.9-l4t_<version>_arm64.deb
Verify the installed packages:
dpkg-query -W -f='${Package}\t${Version}\n' \
nvidia-l4t-core \
nvidia-l4t-pva \
pva-sdk-2.9-l4t \
pva-allow-2 \
libyaml-cpp0.7
The nvidia-l4t-pva version must match the installed JetPack/L4T release.
Compiler tools, development packages, ASIP Programmer, and NVVPU are not required for evaluation.
Allow PVA Execution
For evaluation purposes, disable PVA VPU application authentication:
sudo nvidia-pva-allow disable
This is intended for evaluation and debugging only. Production deployments should use an application-specific PVA allowlist instead.
Build rrv4l2src
We have a port of the Video For Linux 2 (v4l2) GStreamer plug-ins, supporting GStreamer 1.14 through 1.24 and 10-bit or 8-bit Bayer for MIPI cameras. You can evaluate a non-production version of the video source using a patched version available in this repo.
Please consider the following instructions:
1. Clone the repo:
git clone https://gitlab.ridgerun.com/open/gstreamer/gst-v4l2src.git -b bpp-10 cd gst-v4l2src
2. Configure the build
# Create the folder
mkdir -p ${HOME}/custom-gst-v4l2src
# Configure the path in build, compile and install
meson builddir --prefix=${HOME}/custom-gst-v4l2src --libdir=${HOME}/custom-gst-v4l2src/lib
ninja -C builddir install
3. Export the environment variables
export LD_LIBRARY_PATH=${HOME}/custom-gst-v4l2src/lib:${LD_LIBRARY_PATH}
export GST_PLUGIN_PATH=${HOME}/custom-gst-v4l2src/lib/gstreamer-1.0:${GST_PLUGIN_PATH}
These exports can be used in the .bashrc to automate them.
4. Test the installation
gst-inspect-1.0 rrv4l2src
It shall show a valid inspect.
