Live data from Hacker News

NumPy 1.24

github.com

31–34 of 34 posts

Re: NumPy 1.24

#31
post #3

How do people usually pronounce numpy? num-pee or num-pie?

I first read numpy all in lower case as imports and pronounced it num-pee, rhyming with “this road is bumpy.”

I am frequently stupidly oblivious, so the first time I noticed the docs spelling it “NumPy” I thought that an odd spelling for a package called num-pee.

Only when I heard it in a presentation did it all click and I laughed at my misunderstanding.

I now attempt to say “num-pie” but frequently mispronounce in error.

Re: NumPy 1.24

#32
post #24
post #3

How do people usually pronounce numpy? num-pee or num-pie?

YouGlish is a handy resource for this sort of question. 209 video clips with searchable transcripts use "NumPy": https://youglish.com/pronounce/NumPy/english ? Out of the first 16 clips I checked (which contain a few duplicate speakers), all say "num-pie". Next, "sqlite": https://youglish.com/pronounce/sqlite/english ? 15 for "sequel-light", 1 for "S-Q-L-light". 430 left to check. I recently learned YouGlish also sup…

Checked MySQL and i see almost everyone pronounces it as My Sequel. Well, with the exception of Marten Mickos, who was the CEO of MySQL before same was acquired by Sun Microsystems. I guess, this would be another tomato, tomahto.

Re: NumPy 1.24

#33
post #13
post #10

Has anyone noticed numpy running slower on Windows vs. Linux given the same hardware? I would have expected similar performance, but it seems like on my setup the same python runs about 20-30% faster on linux, all other things being equal.

That seems about on par with my experience. Even my own numeric code in C++ frequently runs significantly faster on Linux on the same hardware.

It could be something clever like maybe gcc uses a better allocator or maybe you have better, I dunno, OpenMP libraries on Linux (or whatever the C++ equivalent is)

On the other hand, when I boot into Windows for games, it always surprises me how much CPU Windows uses just… sitting there, doing nothing.

Re: NumPy 1.24

#34
"Passing dtype instances other than the canonical (mainly native byte-order) ones to dtype= or signature= in ufuncs will now raise a TypeError."

This update breaks Seaborn library. I need to keep numpy on v1.23

Post reply on HN