rocm/tensorflow repository overview

This repo host the docker images for Tensorflow releases with ROCm backend support. To configure docker environment for ROCm, please refer to the following installation guide:

https://github.com/RadeonOpenCompute/ROCm-docker/blob/master/quick-start.md⁠

The -dev docker containers have all dependencies to rebuild tensorflow-rocm from the source.

The docker images been hosted in this registry will run on gfx900(Vega10-type GPU - MI25, Vega56, Vega64), gfx906 (Vega20-type GPU - MI50, MI60) and gfx908 (MI100), gfx90a (MI200) and gfx1030 (Navi21).

Here's the recommended command to launch the TF2.5 docker image:

alias drun='sudo docker run -it --network=host --device=/dev/kfd --device=/dev/dri --ipc=host --shm-size 16G --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v $HOME/dockerx:/dockerx'

drun rocm/tensorflow:latest

Images built after ROCm3.1 are based on Ubuntu 18.04, earlier images are based on Ubuntu 16.04 unless otherwise specified in the tag.