[eslaught]

> GPU support

If you haven't tried cuNumeric [1], you really ought to. It's a drop-in NumPy wrapper for distributed GPU acceleration. Would be interesting to see if it works for this.

[1]: https://github.com/nv-legate/cunumeric

[VHRanger]

The problem with drop-in replacements between CPU and GPU code is that performance GPU code requires rethinking the dataflow often -- so even if the code itself is a drop-in, the "make it good" part still requires some rewriting.

I'd be curious how that library compares to other numeric python GPU libraries