We use mpi4py for parallelism:


We are experimenting with using dark for some things, but yt predates dask and has its own capabilities for out of core and parallel computation. Deeper integration with dask is amethyst g we might explore in the future but we have no current concrete plans.

Also please direct questions like this to the mailing list (cc’d) rather than to the mailing list admins.

On Tue, Jul 24, 2018 at 7:07 PM Aaron Chu <xweichu@ucsc.edu> wrote:
Dear teams,

We're working on a project which needs yt lib to access boxlib data such as Amex. And we consider to distribute the computation across multiple nodes. Is there a way that yt can be used with distributed computation framework like dask(https://dask.pydata.org/en/latest/)?  Or any features are under development to support Dask?

Appreciate your time. Thanks.

Best Regards,
Aaron Chu