Getting Started on Amazon EC2 VT1

Amazon EC2 VT1 Configurations

The following table summarizes the number of Alveo U30 cards and Xilinx devices for each of the different instance sizes of the Amazon EC2 VT1 family. For more details about the transcoding performance of these different configurations, refer to the performance tables in the Specs and Features page of this documentation.

Instance Type

Number of U30 Cards

Number of Transcode devices

vt1.3xl

1

2

vt1.6xl

2

4

vt1.24xl

8

16


Quick Start Guide for Amazon EC2 VT1

  1. Launch an Amazon EC2 VT1 Instance

    • Follow the instructions at Launch an Amazon EC2 Instance

    • When choosing the instance type in the EC2 console, make sure to select the Amazon EC2 VT1 instance family.

    • When choosing an Amazon Machine Image (AMI), make sure to select the Xilinx Video SDK AMI for VT1 Instances for Ubuntu 18.04 or for Amazon Linux 2

    • To get more information about Amazon EC2 VT1 instances sizes and pricing see the VT1 web page

  2. Connect to the Amazon EC2 VT1 Instance

  3. Clone the Xilinx Video SDK repository:

    git clone https://github.com/Xilinx/video-sdk -b v1.5 --depth 1
    
    • Cloning the repo downloads the source for the examples and tutorials.

    • It is not needed to install the binary packages included in the repository as they already installed in the public Xilinx Video SDK AMI.

  4. Set up the runtime environment:

    source /opt/xilinx/xcdr/setup.sh
    

    Note

    The script may issue a “Failed to start xrmd.service” message which can safely be ignored.

    • This step should be performed each time you open a new terminal on your instance.

    • You are now ready to use the video acceleration features of your Amazon EC2 VT1 instance.

  5. Run your first examples

    • See the tutorials and examples page to learn how to run video transcoding jobs on your Amazon EC2 VT1 instance.


Installing the Xilinx Video SDK on an existing AMI

In order to use a custom AMI on a Amazon EC2 VT1 instance, it is necessary to install the Xilinx Video SDK in the AMI. This section explains how to do this. These steps are not necessary when using the public Xilinx Video SDK AMI since it already includes a complete installation of the Xilinx Video SDK and can readily be used on a Amazon EC2 VT1 instance.

Additionally, AWS provides packer scripts to make it easy for customers to build their own AMI’s for use with VT1 instances. Those scripts can be found at the following repo: https://github.com/aws-samples/aws-vt-baseami-pipeline

Prerequisites

Install the following packages if they are not already present in your AMI.

  • For Ubuntu:

    sudo apt install linux-modules-extra-$(uname -r)
    sudo apt install python
    
  • For AL2:

    sudo amazon-linux-extras install epel -y
    sudo yum install kernel-devel-$(uname -r) kernel-headers-$(uname -r) boost-devel gcc-c++ -y
    

Installation Procedure

  1. Open a bash shell in your existing AMI

  2. Download the packages included in this repository:

    git clone https://github.com/Xilinx/video-sdk -b v1.5 --depth 1
    
  3. Navigate to the directory containing the packages corresponding to your Operating System:

    cd ./video-sdk/release/<os>
    
  4. Install the software components of the Xilinx Video SDK:

    ./install.sh -sw
    

    A successfull installation will end with the following messages:

    Successfully wrote (76557 bytes) to the output file: /opt/xilinx/xcdr/xclbins/transcode.xclbin
    Leaving xclbinutil.
    
  5. Always set up the runtime environment for the Xilinx Video SDK before running video transcoding jobs on your Amazon EC2 VT1 instance:

    source /opt/xilinx/xcdr/setup.sh
    

    Note

    The script may issue a “Failed to start xrmd.service” message which can safely be ignored.


Patching AMIs using v1.5 of the Video SDK

When using version 1.5 of the Xilinx Video SDK on an AWS VT1 instance running the latest version of the Xilinx firmware, sourcing the setup.sh script will give a “No U30 devices found” error.

In order to fix this, AMIs using version 1.5 of the Xilinx Video SDK need to be patched as follows:

wget https://raw.githubusercontent.com/Xilinx/video-sdk/v1.5/patches/u30_1.5_patch.sh
./u30_1.5_patch.sh

Working with Containers

For instructions on how to work with Docker containers, refer to the Container Setup page of this documentation.

In addition, the user guide for Amazon ECS can be found here: https://docs.aws.amazon.com/AmazonECS/latest/developerguide/getting-started.html


Working with Kubernetes

For instructions on how to deploy Docker containers with Kubernetes and EKS, refer to the Deploying with Kubernetes page of this documentation. A comprehense step-by-step guide on the AWS Compute Blog can be found here: https://aws.amazon.com/blogs/compute/deep-dive-on-amazon-ec2-vt1-instances/

In addition, the user guide for Amazon EKS can be found here: https://docs.aws.amazon.com/eks/latest/userguide/getting-started.html