PERF: tighter cython declarations, faster __iter__

[jbrockmendel]

• closes #xxxx
• tests added / passed
• Ensure all linting tests pass, see here for how to run them
• whatsnew entry

[mzeitlin11]

LGTM! Are the ndim=1 additions for readability or performance? Have always just assumed (but never checked) that cython infers it. If it helps perf we should do it everywhere - a quick regex shows lots of places we don't use ndim=1

[jreback]

lgtm

[jreback]

can you post benchmark changes in the top

[jbrockmendel]

Are the ndim=1 additions for readability or performance? Have always just assumed (but never checked) that cython infers it.

When I was working on this I convinced myself that it made a difference, but I can no longer remember why. Maybe @da-woods can weigh in?

can you post benchmark changes in the top

It's all going to be micro-perf that asvs don't pick up well

[da-woods]

Are the ndim=1 additions for readability or performance? Have always just assumed (but never checked) that cython infers it.

When I was working on this I convinced myself that it made a difference, but I can no longer remember why. Maybe @da-woods can weigh in?

https://cython.readthedocs.io/en/latest/src/tutorial/numpy.html#efficient-indexing

“ndim” keyword-only argument, if not provided then one-dimensional is assumed

I had a quick test of a simple example (looking at the annotated source) and the behaviour looks to match what the docs say. So I think there's no performance advantage to specifying ndim=1 but you may prefer to be explicit for readability reasons.

[jbrockmendel]

Thanks @da-woods.

If the ndim=1 is extraneous, I'm OK with either the more concise or more explicit versions.