[Numpy-discussion] NumPy sprints @ scipy 2013

Sebastian Berg sebastian@sipsolutions....
Fri Jun 28 08:10:41 CDT 2013


On Fri, 2013-06-28 at 07:44 -0500, David Cournapeau wrote:
> Hi there,
> 
> 
> It is very last minute, but I have set up a page to coordinate a bit
> scipy 2013's numpy sprints (Friday 28 and Saturday 29th, although I
> may not be there the second day).
> 
> 
> Depending on the audience, I will also look for low hanging fruits.
> Feel free to edit the wiki at will
> 
Just saw the "take vs. indexing" speed thing. To get it quite right, I
think you probably need to create a new mapiter (after all, take is not
ideal since it requires contiguity - though I guess it could be used as
a special case). Its only half related but this might be interesting if
someone wants to look into it:
https://github.com/seberg/numpy/compare/master...rewrite-index-parser

It does some smaller (but binary incompatible) tweaks to mapiter too,
though the speedup there is not nearly as big as should be possible (and
some fast special cases are not active). Mostly its a start of a
possible rewrite of the indexing machinery, but since the current
mapiter is rather organically interwoven with indexing...

Only a note, I would like to see it, but I doubt its a one day
project :).

- Sebastian

PS: Wondering about visiting EuroScipy (if my schedule works out), if
some of you are there, it would be something I could imagine working on
then (until then, I don't have any time).

> 
> cheers,
> David
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion




More information about the NumPy-Discussion mailing list