Building CUDA ISP for NVIDIA Jetson
| CUDA ISP for NVIDIA Jetson | |
|---|---|
 | |
| CUDA ISP for NVIDIA Jetson Basics | |
|
|
|
| Getting Started | |
|
|
|
| User Manual | |
|
|
|
| GStreamer | |
|
|
|
| Examples | |
|
|
|
| Performance | |
|
|
|
| Contact Us | |
|
|
Contents
Dependencies
 | lleon: I think we should provide more hints. Ok. Here is a list but, where can we find them? (please remove this box when addressed) |
CUDA ISP has the following dependencies for building from source:
- CUDA
- NPPI
- libspdlog
- GstCUDA
Core
Please, install them using (in Debian-based systems):
sudo apt install -y \
libspdlog-dev \
python3-dev
CUDA L4T pacakge (NPPI library)
| lleon: Be careful. Jetson is not compatible with CUDA 11.6. Most of X86 provides support but Jetson is a bit delicate. I would create a subsection dedicated to each platform (please remove this box when addressed) |
The Jetpack default installation will flash and install all the necessary CUDA packages, including NPPI library. This section describes how to perform a manual installation, which is typically __NOT__ needed
Installation guide for CUDA:
1. You will need to know which distribution version you are using:
lsb_release -a
2. Go to the CUDA 11.6.1 download archive and follow the selection menu.
- For Jetson platform you can follow this link: CUDA 11.6.1 Jetson Configuration
- For Ubuntu x86 platform you can follow this link: CUDA 11.6.1 Ubuntu x86 Configuration
Select the version depending on the lsb_release command executed above. We recommend using the deb (local) option for the installation.
3. Once you have selected your system setup, NVIDIA will show you the installation instructions for those requirements.
Installation guide for NPPI library:
In case installing CUDA does not include the NPPI library, you can follow this step to install NPPI library:
1. Download the package for NPPI library. Access this link download NPPI library and find the version of NPPI library you want. For this project we used: libnppicom10_10.1.243-3_amd64.deb
2. Then run:
sudo apt-get update
sudo apt-get install libnppicom10
GstCUDA
| lleon: Please, highlight that it is required for CUDA ISP in case they want to use it on GStreamer (please remove this box when addressed) |
GstCUDA is another RigdeRun product. You can see more about this product in this link: Get GstCUDA
Documentation (optional)
For the documentation, consider the following dependency:
- doxygen
Please, install it using (in Debian-based systems):
sudo apt install -y \
doxygen \
Meson building system
At the building system level:
- ninja-build
- python3-pip
- pkg-config
- meson
Please, install them using (in Debian-based systems):
sudo apt install -y \
python3 \
python3-pip \
ninja-build \
pkg-config
The process should be similar in other systems, not based on Debian.
Then, install meson and other elements.
| lleon: We do not need numpy nor pre-commit (it is only in developer mode) (please remove this box when addressed) |
sudo pip3 install meson \
pre-commit
pip3 install numpy
Building the project
Once fulfilled the dependencies, you can run a default compilation with:
meson build
ninja -C build
sudo ninja -C build install
For additional customization, you may refer to the following table of options:
| lleon: Fix this table accordingly. You can find the options by using meson config (please remove this box when addressed) |
| Configure Option | Description |
|---|---|
| -Denable-python=disabled | Skip the Python API compilation. Enabled by default. |
| -Dwith-python-version=3.8 | Sets the Python version. It takes '3' by default. |
| -Denable-docs=disabled | Skip documentation generation. Enabled by default. |
| --prefix /usr | Set the installation path of the library |
Table 1. CUDA ISP configuration options (Meson)
