Jetson Nano 2G Armbian 22.04(Jammy) Cuda Installing について調べてみました。
Installing CUDA on Nvidia Jetson Nano の記事を見かけたので、出来るのか考えてみました。
Jetson Nano 2G Armbian 22.04(Jammy) のインストールは、
jetson Nano 2G が、Armbian で動くみたい。 を参照してください。
To install CUDA toolkit on Jetson Nano (or any other Jetson board), there are two main methods:
1.Installing through JetPack SDK
2.Installing from Debian (Ubuntu) repositories
3.Installing from CUDA repositories
の 2.Installing from Debian (Ubuntu) repositories が出来るのか?
Installing from Debian repositories
Before installing CUDA on your Jetson Nano, make sure that you have completed the pre-install
requisites found in this guide to ensure a smooth and hassle-free installation.
の this guide ページには、
2.1. Verify You Have a CUDA-Capable GPU
To verify that your GPU is CUDA-capable, go to your distribution's equivalent of System Properties,
or, from the command line, enter:
lspci | grep -i nvidia
とあるので、Jetson Nano 2G armbian で、さっそくチェック。
$ lspci | grep -i nvidia
PcI bridge : NVIDIA Corporation Device 0faf(rev a1)
と出ました。
後は、
$ sudo apt list 2>/dev/null | grep cuda
で、いくつかパッケージが出てきます。
$ sudo apt install nvidia-cuda-toolkit
$ sudo apt install nvidia-cudnn
error
How to recovery.
$ sudo apt remove --purge nvidia-cudnn
$ cd /var/lib/dpkg/info
$ ls |grep cudnn
nvidia-cudnn.config
nvidia-cudnn.list
nvidia-cudnn.md5sums
nvidia-cudnn.postinst
nvidia-cudnn.postrm
nvidia-cudnn.prerm
nvidia-cudnn.templates
$ sudo rm nvidia-cudnn.postinst nvidia-cudnn.postrm nvidia-cudnn.prerm
$ sudo apt remove --purge nvidia-cudnn
Retry
$ sudo apt install --reinstall nvidia-cudnn
same error occured.
Never install.
reffer from
apt-getのpost-installationスクリプトエラー
check the follwing repos.
developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/sbsa/
How about these ?
cuda_11.8.0-1_arm64.deb
libcudnn8-dev_8.7.0.84-1+cuda11.8_arm64.deb
Check version_11.8.0.json
Other wise, How abou Jetson nano 2G JetPack 4.5 cuda repo?
Let's check JetPack 4.5 file
/usr/apt/sources.list.d/cuda-l4t-10-2-local.list
deb file:///var/cuda-repo-l4t-10-2-local /
$ ls /var/cuda-repo-l4t-10-2-local
cuda version 10.2