Installing hip-ucxx#
2026-11-17
2 min read time
You can install hip-ucxx via AMD PyPI as described below. This is recommended for users of the code. For developers interested in modifying or contributing to the Open Source hip-ucxx component, see the Building hip-ucxx from source.
Requirements#
See System requirements for information related to supported operating systems, ROCm versions, and AMD GPUs before installing hip-ucxx.
hip-ucxx also requires a working installation of UCX
(version 1.18.0 or later) built with ROCm support. Installing hip-ucxx from AMD
PyPI as described below pulls in the amd-libucx package automatically, which
compiles a ROCm-enabled UCX at install time, so no separate UCX step is normally
needed. See Building UCX with ROCm support if you prefer to supply UCX yourself, either via
amd-libucx directly or by building from source.
Install hip-ucxx via AMD PyPI#
Packaged versions of hip-ucxx and its dependencies are distributed via AMD PyPI. This section discusses how to install hip-ucxx via this package index.
Set up a Python environment#
Option A: Conda
Create and activate a Conda environment with a compatible Python version, such as 3.11 or 3.12 as shown below. For more information on compatible Python versions, see System requirements.
conda create --name hip-ucxx python=3.12 # Specify your Python version
conda activate hip-ucxx
Option B: Python virtual environment
Create and activate a virtual environment:
python3 -m venv hip-ucxx-env
source hip-ucxx-env/bin/activate
Install packages#
Note
The amd-libucx dependency is a source distribution and compiles a
ROCm-enabled UCX during pip install. Install its build prerequisites
first (on Ubuntu):
sudo apt install make gcc g++ patchelf wget \
rdma-core libibverbs-dev librdmacm-dev libnuma-dev
With your environment activated, install hip-ucxx using pip and the AMD PyPI URL:
pip install amd-hipucxx==0.2.0 --extra-index-url=https://pypi.amd.com/rocm-10.0.0/simple/
This will also install the amd-libhipucxx dependency, which provides the underlying C++ shared
library.
To also install the Dask Distributed backend:
pip install amd-distributed-hipucxx==0.2.0 --extra-index-url=https://pypi.amd.com/rocm-10.0.0/simple/