23 Jan
2019
23 Jan
'19
11:30 p.m.
Hi all, Inevitably, codes have to execute in parallel to make best use of modern processors. One such PR caught my eye today: https://github.com/scikit-image/scikit-image/pull/3670/files Do we have any precedent for how we handle parallel processing code? We have used dask in the past, but it is an optional dependency. We have Cython code that uses prange. It would be good to have a set of guidelines around which parallel paradigms we embrace. Best regards, Stéfan