前往小程序,Get更优阅读体验!
立即前往
首页
学习
活动
专区
工具
TVP
发布
社区首页 >专栏 >NVIDIA Clara-AGX-Developer-Kit installation

NVIDIA Clara-AGX-Developer-Kit installation

原创
作者头像
vanguard
修改2022-07-02 05:09:45
1.3K1
修改2022-07-02 05:09:45
举报
文章被收录于专栏:vanguard

The healthcare and life science industries are being transformed by advancements in accelerated computing and AI. These advancements provide a foundation that makes it easier to develop, use, and maintain AI-powered medical instruments, plus glean insights from data faster and more accurately. At the center of this medical instrument revolution is NVIDIA Clara AGX™.

用户手册

装机过程 (Jetpack5+):

1. 准备工作

1.1 Host PC with Ubuntu 20.04 with SDKmanage 装好软件的Linux主机

1.2 Access to Clara AGX SDK 高质量的USB_TYPEC数据线连接设备和主机

1.3 AC Power Cord 交流电源线

1.4 Display, mouse and keyboard 显示器、鼠标和键盘

1.5 Account 账户权限申请

https://developer.nvidia.com/clara-holoscan-sdk-program

2. 刷机过程

2.1. Connect display to HDMI out port, mouse and keyboard to Clara AGX.

2.1.将显示器连接到 HDMI 输出端口,将鼠标和键盘连接到 Clara AGX。

2.2. Attach Clara AGX Developer Kit to host PC through USBC port.

2.2.通过 USBC 端口将 Clara AGX 开发人员套件连接到主机 PC。

2.3. Run SDKmanager on host PC and install the latest Clara AGX SDK.

2.3.在主机 PC 上运行 SDKmanager 并安装最新的 Clara AGX SDK。

2.4. After OS flash, SDKmanager will pause.

2.4.操作系统刷新后,SDKmanager 将暂停。

2.5. Complete setup on Clara AGX Developer Kit until desktop screen.

2.5.在 Clara AGX Developer Kit 上完成设置,直到桌面屏幕。

2.6. Complete SDKmanager flash on the host.

2.6.在主机上完成 SDKmanager刷机。

如果不顺利可以用Recovery+Reset模式(开盖板),操作见用户手册

代码语言:shell
复制
sudo apt update
# sudo apt upgrade # 升级内核非必要
# 依赖库问题考虑强制覆盖可修复
# sudo apt --fix-broken install -o Dpkg::Options::="--force-overwrite"
# 安装固件工具
sudo apt install mstflint
# 检查固件版本
lspci | grep Mellanox
sudo mstflint -d 0000:09:00.0 q full
# 按用户手册要求是否需要升级
# scp或者samba传文件
sudo reboot

3. 独显驱动

代码语言:shell
复制
sudo cp /opt/nvidia/l4t-gputools/bin/nvgpuswitch.py /usr/local/bin/
# 查询
# nvgpuswitch.py query iGPU # (nvidia-l4t-cuda, 34.1.2-20220524101639)
# 换卡
sudo nvgpuswitch.py install dGPU
# 再查
# nvgpuswitch.py query dGPU # (cuda-drivers, 510.73.08-1) Quadro RTX 6000, 24576 MiB
# 测试DP输出,
# 测试nvidia-smi
# 设置环境变量(如需)
# export PATH=/usr/local/cuda-11.6/bin:$PATH 
# export LD_LIBRARY_PATH=/usr/local/cuda-11.6/lib64:$LD_LIBRARY_PATH
# 换卡*2 (如需)
# sudo nvgpuswitch.py install iGPU

硬件配置:

1 NVIDIA® JETSON AGX XAVIER™

2 NVIDIA RTX™ 6000  

3 NVIDIA CONNECTX®-6 100GBE NETWORK INTERFACE CARE (NIC) WITH TWO PORTS -QSFP28 FOR 100GBE AND RJ45 FOR 10GBE  

4 HDMI 2.0 INPUT  

5 2X PCIE GEN4 WITH EIGHT SLOTS  

6 250GB M.2 SATA STORAGE

安装 Clara Holoscan SDK Clara Holoscan NGC

Building a Global Defense System Against Coronavirus (SARS-COV-2)

NVIDIA SDK Manager for Your Development Environment Setup

 NVIDIA Clara AGX Developer Kit - Early Interest Program

Developing End-to-End Real-time Applications with the NVIDIA Clara AGX Developer Kit

AI 助力的医疗健康解决方案 & 下一代产品

代码语言:python
代码运行次数:0
复制
def install_dgpu(commands, l4t_repo):
    """ Add commands to install the dGPU driver packages. """
    packages_to_install = [
        "nvidia-l4t-*",
        "nvidia-driver-510",
        "nvidia-dkms-510",
        "nvidia-utils-510",
        "cuda",
        "nvidia-container-runtime",
        "libnvinfer-bin",
        "mstflint"
    ]
    install_string = " ".join(packages_to_install)
    add_command(commands, "Remove blacklist for dGPU driver",
        "rm -f /etc/modprobe.d/blacklist-nvidia.conf")
    add_command(commands, "Add blacklist for iGPU driver",
        "echo 'blacklist nvgpu' > /etc/modprobe.d/blacklist-nvgpu.conf")
    add_command(commands, "Add modprobe options for dGPU driver",
        "echo 'options nvidia NVreg_EnableGpuFirmware=0 NVreg_DmaRemapPeerMmio=0' > /etc/modprobe.d/nvidia-holoscan.conf")
    add_command(commands, "Add public key for dGPU L4T apt repo",
        f"apt-key adv --fetch-keys http://{l4t_repo}/jetson-ota-public.asc")
    add_command(commands, "Add dGPU L4T apt repo",
        f"echo 'deb http://{l4t_repo}/dgpu-rm r34.1.2 main'" +
        " >> /etc/apt/sources.list.d/l4t_rm.list")
    add_command(commands, "Add public key for nvidia-container-runtime apt repo",
        "apt-key adv --fetch-keys https://nvidia.github.io/nvidia-container-runtime/gpgkey")
    add_command(commands, "Add nvidia-container-runtime apt repo",
        f"echo 'deb https://nvidia.github.io/libnvidia-container/stable/ubuntu18.04/$(ARCH) /'" +
         " >> /etc/apt/sources.list.d/l4t_rm.list && " +
        f"echo 'deb https://nvidia.github.io/nvidia-container-runtime/stable/ubuntu18.04/$(ARCH) /'" +
         " >> /etc/apt/sources.list.d/l4t_rm.list")
    add_command(commands, "Install dGPU drivers",
        f"apt update && apt install -y {install_string}")
    add_command(commands, "Repair dGPU library config",
        "echo '/usr/lib/aarch64-linux-gnu/tegra' >> /etc/ld.so.conf.d/nvidia-tegra.conf && ldconfig")
    add_command(commands, "Configure nvidia-container-runtime",
        "mkdir /etc/systemd/system/docker.service.d && " +
        "echo '[Service]' > /etc/systemd/system/docker.service.d/override.conf && " +
        "echo 'ExecStart=' >> /etc/systemd/system/docker.service.d/override.conf && " +
        "echo 'ExecStart=/usr/bin/dockerd --host=fd:// --add-runtime=nvidia=/usr/bin/nvidia-container-runtime'" +
        " >> /etc/systemd/system/docker.service.d/override.conf")
    add_command(commands, "Install dGPU nvpmodel configuration",
        "ln -sf /etc/nvpmodel/nvpmodel_t194_e3900_dGPU.conf /etc/nvpmodel.conf")
    soc = os.popen("cat /etc/nv_boot_control.conf | grep COMPATIBLE_SPEC | cut -d '-' -f 7").read().strip()
    if soc == "clara":
        driver = os.popen("lspci | grep 0000:09:00.0 | cut -d ' ' -f 4").read().strip()
        if driver == "Mellanox":
            add_command(commands, "Upgrade Mellanox CX-6 firmware",
                "mstflint -d 0000:09:00.0 -i /opt/nvidia/l4t-cx6-firmware/fw-ConnectX6.bin burn")

原创声明:本文系作者授权腾讯云开发者社区发表,未经许可,不得转载。

如有侵权,请联系 cloudcommunity@tencent.com 删除。

原创声明:本文系作者授权腾讯云开发者社区发表,未经许可,不得转载。

如有侵权,请联系 cloudcommunity@tencent.com 删除。

评论
登录后参与评论
0 条评论
热度
最新
推荐阅读
目录
  • 装机过程 (Jetpack5+):
    • 1. 准备工作
      • 2. 刷机过程
        • 3. 独显驱动
        • 硬件配置:
        相关产品与服务
        容器服务
        腾讯云容器服务(Tencent Kubernetes Engine, TKE)基于原生 kubernetes 提供以容器为核心的、高度可扩展的高性能容器管理服务,覆盖 Serverless、边缘计算、分布式云等多种业务部署场景,业内首创单个集群兼容多种计算节点的容器资源管理模式。同时产品作为云原生 Finops 领先布道者,主导开源项目Crane,全面助力客户实现资源优化、成本控制。
        领券
        问题归档专栏文章快讯文章归档关键词归档开发者手册归档开发者手册 Section 归档