Docker Setup

The following steps outline the process for pulling and running the Docker container required for the Vitis AI model quantization and compilation flows.

1. Pull the Docker image on a system with Docker installed. For installation instructions, see Install Docker Engine on Ubuntu and Linux Post-Installation Steps for Docker Engine. You can also contact your system administrator.

   docker pull amdih/vitis-ai:versal-2ve-release_v6.2_0612
   

   Note

   Docker image pulling can take a few minutes.

2. Verify the Docker image details (“REPOSITORY”, “IMAGE ID” and “TAG”):

   docker images
   

   The output appears in the following format:

   REPOSITORY	TAG	IMAGE ID	CREATED	SIZE
   amdih/vitis-ai	versal-2ve-release_v6.2_0612	8cd54102c274	xx hours ago	31.2GB

3. Get the AIEMLv2build build tools license (node locked/floating) as described in Obtaining License. Copy the node locked license file to your workspace at <yourLicenseDir>.

4. Start a Docker container and mount the required directories:

   docker run --ulimit stack=-1:-1 -it --rm --network host -v <yourLicenseDir>:/usr/licenses -v <yourDir>:/host_mount amdih/vitis-ai:versal-2ve-release_v6.2_0612 bash
   

   This command uses a node-locked license. If you have a floating license, see Docker Run Command with Floating License for example commands to run docker with Floating license.

   Note

   • Use the -v option to map the host directory (<yourDir>) to the container directory (/host_mount).

   • Large models can require a larger stack size. Use --ulimit stack=-1:-1 to remove the stack limit.