Vitis AI安装步骤-包括Ubuntu、Docker安装

点击全文阅读

环境安装流程

Ubuntu 20.04 安装

下载镜像：ubuntu-20.04.6-desktop-amd64.iso

下载 rufus: 官网地址

制作USB启动盘

开机按 Delete 键进入 BIOS

启动项将首选项改为 USB 启动盘

重新启动进入安装程序

选择 English

有线网账号密码：psd

选择 Mininal installation

选择 Erase disk and install Ubuntu

设置用户名密码是：psd

安装配置 ssh

sudo apt-get updatesudo apt install openssh-serversudo apt-get install vimsudo vim /etc/ssh/sshd_config

端口号修改为23321后保存，继续执行

sudo systemctl restart sshdsudo ufw allow 23321

关闭自动锁屏：

点击右上角 Settings

点击 Privacy

点击 Screen Lock

关闭自动锁屏

---

CUDA 11.3 & cuDNN 安装

查看当前驱动

dpkg -l | grep nvidia

卸载原本的驱动并清理链接

sudo apt-get purge nvidia*sudo apt autoremove

查询可用驱动

ubuntu-drivers devices

自动安装推荐的驱动

sudo ubuntu-drivers autoinstall

重启，然后验证驱动是否安装成功

sudo rebootnvidia-smi

下载并运行 CUDA 11.3.1 安装程序

cd ~wget https://developer.download.nvidia.com/compute/cuda/11.3.1/local_installers/cuda_11.3.1_465.19.01_linux.runsudo apt install gccsudo sh cuda_11.3.1_465.19.01_linux.run

只勾选 CUDA Toolkit 11.3，然后安装

添加环境变量

sudo vim /etc/profile

export CUDA_HOME=/usr/local/cuda-11.3export PATH=$PATH:$CUDA_HOME/binexport LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda-11.3/lib64

下载并安装 cuDNN 8.9.2.26

cd ~wget https://developer.nvidia.com/downloads/compute/cudnn/secure/8.9.2/local_installers/11.x/cudnn-linux-x86_64-8.9.2.26_cuda11-archive.tar.xztar -xvf cudnn-linux-x86_64-8.9.2.26_cuda11-archive.tar.xzcd cudnn-linux-x86_64-8.9.2.26_cuda11-archivesudo cp -r ./bin/* /usr/local/cuda-11.3/binsudo cp -r ./lib/* /usr/local/cuda-11.3/lib64

验证是否安装成功

source /etc/profilenvcc -Vnvidia-smi

---

docker & nvidia docker 安装

安装docker

sudo apt-get install -y docker.iosudo systemctl start dockersudo systemctl enable dockerdocker version

安装 nvidia container toolkit

sudo apt-get install curlwget https://download.docker.com/linux/ubuntu/gpgsudo apt-key add gpgvim installNvidiaContainer.sh

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)sudo curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -sudo curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

保存后执行

./installNvidiaContainer.shsudo apt-get update && sudo apt-get install -y nvidia-container-toolkitsudo systemctl restart dockerrm installNvidiaContainer.sh

创建多个新用户并添加docker用户组权限

cd ~vim addUsers.sh

for name in aifpga maomao yangyang lichangyvdo    echo $name    useradd -d /home/$name -m -s /bin/bash $name    echo $name:$name | chpasswd    usermod -aG docker ${name}    # passwd --expire $name    echo "$user add successfuly"done

保存后执行

./addUsers.sh

---

Vitis & Vivado & Vitis HLS 安装

下载 Vitis 包并进入目录

配置 dash（键盘选择 No）

sudo dpkg-reconfigure dash

安装依赖包并执行安装程序

sudo apt-get install ocl-icd-libopencl1sudo apt-get install opencl-headerssudo apt-get install ocl-icd-opencl-devsudo apt install libstdc++6sudo apt install libncurses5sudo apt-get install libtinfo5sudo chmod +x xsetupsudo ./xsetup

选择安装内容（需在本机使用图形界面操作）

选择 Vitis

选择以下内容（共210.68GB）

Vitis Unified Software Platform

Vitis Model Composer

DocNav

Install devices for Alveo and edge acceleration platforms

Install Devices for Kria soMs and starter Kits

Devices for Custom Platforms

Engineering Sample Devices for Custom Platforms

其他配置默认，然后等待安装完成

配置环境

sudo vim /etc/profile

source /tools/Xilinx/Vivado/2023.1/settings64.shsource /tools/Xilinx/Vitis/2023.1/settings64.shsource /tools/Xilinx/Vitis_HLS/2023.1/settings64.sh

安装 USB 驱动

cd /tools/Xilinx/Vivado/2023.1/data/xicom/cable_drivers/lin64/install_script/install_driverssudo ./install_drivers

验证是否安装成功

source /etc/profilevitisvivadovitis_hls

无论执行哪一个都有图形界面弹出

---

Vitis AI 安装

克隆 Vitis AI 仓库

cd ~git clone https://github.com/Xilinx/Vitis-AI

构建基于 Pytorch-CUDA 的镜像

cd Vitis-AI/docker./docker_build.sh -t gpu -f pytorch

验证是否安装成功

cd .././docker_run.sh xilinx/vitis-ai-pytorch-gpu:3.5.0.001-a350fc104

---

点击全文阅读