Pyston 2.1 Is Blowing Past Python 3.8/3.9 Performance

[phoronix]

Phoronix: Pyston 2.1 Is Blowing Past Python 3.8/3.9 Performance

With this past week's release of Pyston 2.1 as an alternative Python interpreter I was curious to see how the performance compared to that of upstream Python... So here are some weekend benchmarks with a Ryzen 9 5900X system...

http://www.phoronix.com/scan.php?page=news_item&px=Pyston-2.1-vs-Python-3.8-3.9

[Setif]

How about Intel Python distribution?
Installing Intel® Performance Libraries and Intel® Distribution for...

[david-nk]

Looks good, but if you consider how hilariously slow Python is, even twice as fast is still very slow.
Regardless, since for some reason Python has become the standard language for doing ML, any improvements are welcome.

[bachchain]

Python has become the standard language for doing ML, any improvements are welcome.

I imagine the hot paths for most ML libraries are already written in assembly, so this probably won't help too much

[Setif]

Looks good, but if you consider how hilariously slow Python is, even twice as fast is still very slow.
Regardless, since for some reason Python has become the standard language for doing ML, any improvements are welcome.

I think for the most ML programs, python is just an interface while the core is written in C/C++.

[rmfx]

It’s known that Cpython is the slowest and meant for conformity.

Pypy would be totally the best. It’s implemented not in c but python and faster.

I really wonder why they didn’t swallow their pride and go with pypy.

[Anarchy]

It’s known that Cpython is the slowest and meant for conformity.

Pypy would be totally the best. It’s implemented not in c but python and faster.

I really wonder why they didn’t swallow their pride and go with pypy.

because a lot of python extensions don't work with pypy. it would break a lot of code.

[Viki Ai]

No source, no interest.

Life is too short to be bogged down by code you can't fix yourself.

[bug77]

No source, no interest.

Life is too short to be bogged down by code you can't fix yourself.

Our plan is to open-source the code in the future, but since compiler projects are expensive and we no longer have benevolent corporate sponsorship, it is currently closed-source while we iron out our business model.

I'll admit, it's a bit strange to lose your sponsors right when the language is seeing a lot of adoption, but it is what it is.

[lyamc]

I wonder how tcc as a JIT compiler would perform vs python, ruby, javascript, etc.