Skip to content

h5py.rst

Latest commit

 

History

History
37 lines (25 loc) · 1.37 KB

h5py.rst

File metadata and controls

37 lines (25 loc) · 1.37 KB

H5PY-backend

The H5PY-backend relies on an MPI compatible build of HDF5, which should be linked to the same MPI implementation as DOLFINx relies on.

The DOLFINx docker images (ghcr.io/fenics/dolfinx/dolfinx:stable) comes with an already configures MPI compatible HDF5 installation, and h5py can in turn be installed with

HDF5_MPI="ON" HDF5_DIR="/usr/local" python3 -m pip install --no-binary=h5py h5py --no-build-isolation

If you are using apt on Ubuntu, this can for instance be achieved with the following commands (here using Docker). Note that this code block does not install DOLFINx, it just illustrates how to get the correct h5py.

FROM ubuntu:24.04 AS base
RUN apt-get update && apt-get install -y python3-dev python3-pip python3-venv libhdf5-mpi-dev libopenmpi-dev


ENV VIRTUAL_ENV=/test-env
ENV PATH=/test-env/bin:$PATH
RUN python3 -m venv ${VIRTUAL_ENV}

ENV HDF5_MPI="ON"
ENV HDF5_DIR=/usr/lib/x86_64-linux-gnu/hdf5/openmpi/
ENV C_PATH=/usr/lib/x86_64-linux-gnu/openmpi/include/:${C_PATH}
RUN python3 -m pip install setuptools cython numpy pkgconfig mpi4py
RUN CC=mpicc python3 -m pip install --no-binary=h5py h5py --no-build-isolation
.. automodule:: io4dolfinx.backends.h5py.backend
    :members:
    :exclude-members: read_point_data