search

castcollab / tesserae2

Tesserae2: Fast recombination-aware global and local alignment.
Other
3 stars 0 forks source link

Speed improvements through effective use of numpy? #1

Closed winni2k closed 4 years ago

winni2k commented 4 years ago

Karl Johan and I dug into the tesserae code today. The __recurrence method does not really use any numpy vectorization, when it really should.

We think that perhaps the for loop at https://github.com/castcollab/tesserae2/blob/master/src/cortexpy/tesserae.py#L355, which iterates over targets could be parallelized using numpy? @kvg what are your thoughts on this?

kvg commented 4 years ago

Oh wow, yeah.  It does indeed look like the calculation for each target is independent of the other ones.  Let’s do it!

  Thanks,     -Kiran On Mar 24, 2020, 6:13 AM -0400, Winni Kretzschmar notifications@github.com, wrote:

Karl Johan and I dug into the tesserae code today. The __recurrence method does not really use any numpy vectorization, when it really should. We think that perhaps the for loop at https://github.com/castcollab/tesserae2/blob/master/src/cortexpy/tesserae.py#L355, which iterates over targets could be parallelized using numpy? @kvg what are your thoughts on this? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

winni2k commented 4 years ago

@karljohanw, I think you were looking to implement this, right?

winni2k commented 4 years ago

5 addresses this issue, but I'm not sure it has been fixed. See #14