mpi4jax
enables zero-copy, multi-host communication of JAX arrays, even from jitted code and from GPU memory.
The JAX framework has great performance for scientific computing workloads, but its multi-host capabilities are still limited.
With mpi4jax
, you can scale your JAX-based simulations to entire CPU and GPU clusters (without ever leaving jax.jit
).
In the spirit of differentiable programming, mpi4jax
also supports differentiating through some MPI operations.
mpi4jax
is available through pip
and conda
:
$ pip install mpi4jax # Pip
$ conda install -c conda-forge mpi4jax # conda
If you use pip and don't have JAX installed already, you will also need to do:
$ pip install jaxlib
(or an equivalent GPU-enabled version, see the JAX installation instructions)
In case your MPI installation is not detected correctly, it can help to install mpi4py separately. When using a pre-installed mpi4py
, you must use --no-build-isolation
when installing mpi4jax
:
# if mpi4py is already installed
$ pip install cython
$ pip install mpi4jax --no-build-isolation
Our documentation includes some more advanced installation examples.
from mpi4py import MPI
import jax
import jax.numpy as jnp
import mpi4jax
comm = MPI.COMM_WORLD
rank = comm.Get_rank()
@jax.jit
def foo(arr):
arr = arr + rank
arr_sum, _ = mpi4jax.allreduce(arr, op=MPI.SUM, comm=comm)
return arr_sum
a = jnp.zeros((3, 3))
result = foo(a)
if rank == 0:
print(result)
Running this script on 4 processes gives:
$ mpirun -n 4 python example.py
[[6. 6. 6.]
[6. 6. 6.]
[6. 6. 6.]]
allreduce
is just one example of the MPI primitives you can use. See all supported operations here.
If you have a question or feature request, or want to report a bug, feel free to open an issue.
We welcome contributions of any kind through pull requests. For information on running our tests, debugging, and contribution guidelines please refer to the corresponding documentation page.
If you use mpi4jax
in your work, please consider citing the following article:
@article{mpi4jax, doi = {10.21105/joss.03419}, url = {https://doi.org/10.21105/joss.03419}, year = {2021}, publisher = {The Open Journal}, volume = {6}, number = {65}, pages = {3419}, author = {Dion Häfner and Filippo Vicentini}, title = {mpi4jax: Zero-copy MPI communication of JAX arrays}, journal = {Journal of Open Source Software} }