A table with these columns might be helpful for performance comparisons: [
structure,
operation,
big_o,
keyset,  # signed ints, randstrs
timeit_output,
system_load,
architecture,  # x86_84, aarch64
os,
]

Does anyone have a recommendation for a tool that stores such benchmark data in JSON and merges the e.g. per-arch output?

Are there URIs for Big-O notation that could be added as e.g. structured attributes in docstrings or annotations?

I just added a ._repr_html_() method to the tabulate package so that, in addition to the many excellent ASCII formats tabulate prints, tabulate.tabulate(data, tablefmt='html') displays an html-escaped table in Jupyter notebooks? It's not yet released, so you'd need to
 `pip install -e git+https://github.com/astanin/python-tabulate@master#egg=tabulate`

On Tue, Dec 24, 2019, 3:53 AM Larry Hastings <larry@hastings.org> wrote:


On 12/23/19 8:09 PM, Kyle Stanley wrote:

Chris Angelico wrote:
> I think this is an artifact of Python not having an empty set literal.
> [snip]
> When both use the constructor call or both use a literal, the
> difference is far smaller. I'd call this one a wash.

Ah, good catch. I hadn't considered that it would make a substantial difference, but that makes sense. Here's an additional comparison between "{}"  and "dict()" to confirm it:

>>> timeit.timeit("{}", number=100_000_000)
2.1038335599987477
>>> timeit.timeit("dict()", number=100_000_000)
10.225583500003268


We could also go the other way with it, set / dict-map-to-dontcare with one element, because that way we can have a set literal:

>>> timeit.timeit("set()", number=100_000_000)
8.579900023061782
>>> timeit.timeit("dict()", number=100_000_000)
10.473437276901677
>>> timeit.timeit("{0}", number=100_000_000)
5.969995185965672
>>> timeit.timeit("{0:0}", number=100_000_000)
6.24465325800702

(ran all four of those just so you see a relative sense on my laptop, which I guess is only sliiiightly slower than Kyle's)


Happy holidays,


/arry

_______________________________________________
Python-Dev mailing list -- python-dev@python.org
To unsubscribe send an email to python-dev-leave@python.org
https://mail.python.org/mailman3/lists/python-dev.python.org/
Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/MQAKLPKWECT22NVPRITAD5XQBIFUS4UA/
Code of Conduct: http://python.org/psf/codeofconduct/