You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Using Numpy matmul for some vector similarity functions, instead of @njit, can speed up processing from
CPU times: user 2.05 s, sys: 19.4 ms, total: 2.07 s
Wall time: 2.41 s
to:
CPU times: user 47.4 ms, sys: 8.95 ms, total: 56.4 ms
Wall time: 62.7 ms
On 1k x 1k spectra, each of 1k size. Purely on CPU.
Describe the solution you'd like
I already have some functions vectorized over at cudams. I'd like to move some of them to base matchms repo.
If that sounds fine with that proposal, I'll open a PR for it in a bit. Maybe there's some reason to avoid adding vectorized version (maybe somehow lower readability?)
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
Using Numpy matmul for some vector similarity functions, instead of
@njit
, can speed up processing fromto:
On 1k x 1k spectra, each of 1k size. Purely on CPU.
Here's a colab notebook to replicate the speedup.
Describe the solution you'd like
I already have some functions vectorized over at cudams. I'd like to move some of them to base matchms repo.
If that sounds fine with that proposal, I'll open a PR for it in a bit. Maybe there's some reason to avoid adding vectorized version (maybe somehow lower readability?)
The text was updated successfully, but these errors were encountered: