Skip to content

Latest commit

 

History

History
212 lines (176 loc) · 8.1 KB

prerequisite_software_installation_guide.md

File metadata and controls

212 lines (176 loc) · 8.1 KB
Raw

Pre-requisite Software Installation Guide

This article describes all the pre-requisite steps needed before installing Apollo.

  • Installing Ubuntu Linux
  • Installing NVIDIA GPU Driver
  • Installing Docker Engine
  • Installing NVIDIA Container Toolkit

tip_icon Working knowledge of Linux is assumed for successful software installation in this guide.

Installing Ubuntu Linux

Although other Linux distributions may be OK, we have only tested Apollo on Ubuntu systems. To be specific, Ubuntu 18.04.5 LTS (Bionic Beaver). So we would recommend using Ubuntu 18.04.5+ (including Ubuntu 20.04) as the host system.

The steps required to install Ubuntu 18.04+ are available at Tutorial from ubuntu.com on How to Install Ubuntu. Please follow the guide there for a successful Ubuntu installation.

Don't forget to perform software updates after the installation is done:

sudo apt-get update
sudo apt-get upgrade

Internet access is needed for successful software updates. Make sure either WiFi or Ethernet cable is connected to a network with Internet access. You might need to configure the network for your host if the connected network is not using the Dynamic Host Configuration Protocol (DHCP).

Installing NVIDIA GPU Driver for Nvidia GPU

The Apollo runtime in the vehicle requires NVIDIA GPU Driver.

According to How to Install NVIDIA Driver, the recommended way for Ubuntu is to use the apt-get commands, or use an official "runfile" from www.nvidia.com/en-us/drivers/unix/

For Ubuntu 18.04+, you can simply run:

sudo apt-get update
sudo apt-add-repository multiverse
sudo apt-get update
sudo apt-get install nvidia-driver-455

You can type nvidia-smi to check if NVIDIA GPU works fine on your host. (You may restart your host for the changes to take effect.) On success, the following message will be shown.

Prompt> nvidia-smi
Mon Jan 25 15:51:08 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.27.04    Driver Version: 460.27.04    CUDA Version: 11.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce RTX 3090    On   | 00000000:65:00.0  On |                  N/A |
| 32%   29C    P8    18W / 350W |    682MiB / 24234MiB |      7%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      1286      G   /usr/lib/xorg/Xorg                 40MiB |
|    0   N/A  N/A      1517      G   /usr/bin/gnome-shell              120MiB |
|    0   N/A  N/A      1899      G   /usr/lib/xorg/Xorg                342MiB |
|    0   N/A  N/A      2037      G   /usr/bin/gnome-shell               69MiB |
|    0   N/A  N/A      4148      G   ...gAAAAAAAAA --shared-files      105MiB |
+-----------------------------------------------------------------------------+

Installing ROCm for AMD GPU

  • Follow instructions at ROCm v5.1 or above upto running amdgpu-install.
  • When running amdgpu-install command use --usecase=hiplibsdk,rocm,dkms
  • Add user to the render group: sudo usermod -a -G render $USER.
  • Now you should be able to run rocminfo and rocm-smi
$ rocminfo
ROCk module is loaded
=====================
HSA System Attributes
=====================
Runtime Version:         1.1
System Timestamp Freq.:  1000.000000MHz
Sig. Max Wait Duration:  18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model:           LARGE
System Endianness:       LITTLE

==========
HSA Agents
==========
*******
Agent 1
*******
  Name:                    AMD EPYC Embedded 7292P 16-Core Processor
  Uuid:                    CPU-XX
  Marketing Name:          AMD EPYC Embedded 7292P 16-Core Processor
  Vendor Name:             CPU
  Feature:                 None specified
  Profile:                 FULL_PROFILE
  Float Round Mode:        NEAR
  Max Queue Number:        0(0x0)
  Queue Min Size:          0(0x0)
  Queue Max Size:          0(0x0)
  Queue Type:              MULTI
  Node:                    0
  Device Type:             CPU
  ...
*******
Agent 2
*******
  Name:                    gfx90a
  Uuid:                    GPU-4e8d8d5c57524677
  Marketing Name:
  Vendor Name:             AMD
  Feature:                 KERNEL_DISPATCH
  Profile:                 BASE_PROFILE
  Float Round Mode:        NEAR
  Max Queue Number:        128(0x80)
  Queue Min Size:          64(0x40)
  Queue Max Size:          131072(0x20000)
  Queue Type:              MULTI
  Node:                    1
  Device Type:             GPU
  ...
*** Done ***

$ rocm-smi
======================= ROCm System Management Interface =======================
================================= Concise Info =================================
GPU  Temp   AvgPwr  SCLK    MCLK     Fan  Perf  PwrCap  VRAM%  GPU%
0    31.0c  34.0W   800Mhz  1600Mhz  0%   auto  300.0W    0%   0%
================================================================================
============================= End of ROCm SMI Log ==============================

Installing Docker Engine

Apollo 6.0+ requires Docker 19.03+ to work properly. Just follow the Install Docker Engine on Ubuntu doc to install docker engine.

Docker-CE on Ubuntu can also be setup using Docker’s official convenience script:

curl https://get.docker.com | sh
sudo usermod -a -G docker $USER
sudo systemctl start docker && sudo systemctl enable docker

You can choose whichever method you would prefer. Just don't forget the Post-installation Actions for Linux, esp. the section on Manage Docker as Non-root User and Configure Docker to Start on Boot.

There is also a dedicated bash script Apollo provides to ease Docker installation, which works both for X86_64 and AArch64 platforms.

Installing NVIDIA Container Toolkit for Nvidia GPU

The NVIDIA Container Toolkit for Docker is required to run Apollo's CUDA based Docker images.

You can run the following to install NVIDIA Container Toolkit:

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-get -y update
sudo apt-get install -y nvidia-docker2

tip_icon Don't forget to restart the Docker daemon for the changes above to take effect.

sudo systemctl restart docker

Refer to NVIDIA Container Toolkit Installation Guide for more.

What's Next

With successful installation of the pre-requisites above, you can now move to "Git Clone Apollo Repo" section now.