Installing hipMM

Installing hipMM#

2026-07-06

3 min read time

Applies to Linux

You can install hipMM via AMD PyPI, which is recommended for end users, or build and install it from source as described in Building and installing hipMM from source.

See System requirements for information regarding supported operating systems, ROCm versions, and AMD GPUs before installing hipMM.

Prerequisites#

The following ROCm components must be installed:

You should also have gcc 13.* and cmake 3.30.4 (or later) installed.

The steps in this guide require a Conda installation. A minimal free version of Conda is Miniforge.

Install hipMM via AMD PyPI#

Warning

Only install hipMM using AMD’s official package index. To ensure security, integrity, and supportability of your builds, consume packages exclusively from AMD’s official package index. Do not install, mirror, or resolve dependencies from any third‑party or unofficial indexes.

Packaged versions of hipMM and its dependencies are distributed via AMD PyPI. This section discusses how to install hipMM via this package index.

Create and activate a Conda environment with Python 3.12 and latest libstdcxx-ng using the minimal environment (conda/environments/install_rocm-latest_arch-x86_64.yaml) as shown below:

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.

channels:
  - conda-forge
dependencies:
- python=3.12 # Specify your Python version
- libstdcxx-ng
- pip
- pip:
   - --extra-index-url=https://pypi.amd.com/rocm-7.2.3/simple
   - rocm-llvm-python~=7.2.3.0
   - hip-python~=7.2.3.0
   - hip-python-as-cuda~=7.2.3.0
   - numba-hip~=0.1.6
   - amd-cupy~=13.5.1
   - amd-hipmm==4.0.0
name: hipmm

You can install both the environment and hipMM in it with a single command:

conda env create -f conda/environments/install_rocm-latest_arch-x86_64.yaml

After activating the hipmm environment, you can use import hipmm in your Python code.

Verify correct installation#

To verify the correctness of the installation, run:

conda activate hipmm

python3

Then enter the following code commands:

import hipmm

print(hipmm.__version__)

You should see output that is similar to:

Python 3.12.11 | packaged by conda-forge | (main, Jun  4 2025, 14:45:31) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import hipmm
>>> print(hipmm.__version__)
'4.0.00'