Open rht opened 7 years ago
It's definitely very possible to make this faster with numba. It's currently set up to work with pypy, and that's what I've been using when speed becomes a bottleneck. It's possible that numba would provide a better speed/simplicity tradeoff but I haven't felt hindered by speed as it is.
I was trying to reproduce the result of https://github.com/Futrell/cliqs/blob/1c72b064f9048f9344ca88c90d6dface616d7ee4/run_mindep.py#L49.
I found that pooling did result in a 2x speed up of the run.
Without parallel:
With parallel (pmap):
This was ran on "Intel(R) Core(TM) i5-4200U CPU @ 1.60GHz", quadcore. I think the run could be ~an order of magnitude faster by inserting several numba
@jit
s to deptransform/depgraph. So far I had tested with@jit
-inggen_row
but didn't observe any speed up.