I have not looked closely at this benchmark but I do know that the Ray authors worked hard to improve on Dask performance and that may matter for us. Unfortunately the Ray team does not seem to care much about workloads like ours, though experimental/array does have a little code.
I have not looked closely at this benchmark but I do know that the Ray authors worked hard to improve on Dask performance and that may matter for us. Unfortunately the Ray team does not seem to care much about workloads like ours, though experimental/array does have a little code.