Aloha ACCESS-NRI folks. Here’s an interesting and possibly helpful call from the new Pangeo Distributed Arrays Working Group.
They want your “problematic workloads”.
This WG is aiming to collect a range of such examples in this repository , to eventually create benchmarks out of them.
NB: my personal impression is that PANGEO folks in USA and elsewhere are less-HPC focused and more into commercial Cloud applications but there should still be some commonalities to workflow hurdles?