search

CovertLab / arrow

Stochastic simulations in python
MIT License
3 stars 1 forks source link

Removed support for alternate forms, rearranging tests to allow for profiling #28

Closed prismofeverything closed 5 years ago

jmason42 commented 5 years ago

This will be helpful. Incidentally I played with Numba a bit and got mixed results. I think we can start JIT'ing choose and propensity but step and evolve pose a number of issues

prismofeverything commented 5 years ago

I saw something similar, yet overall runtime is still lower without jiting anything.... which is weird. Might try cython next, or I might just write this in C.

jmason42 commented 5 years ago

I saw something similar, yet overall runtime is still lower without jiting anything.... which is weird. Might try cython next, or I might just write this in C.

That's odd. I was getting 3x speed-ups when JIT'ing those two functions.

prismofeverything commented 5 years ago

Alright I'll try it with the full sim and see what happens.