Python 3.13 Beta Out For Testing With Experimental JIT, Better Interactive Interpreter

[phoronix]

Phoronix: Python 3.13 Beta Out For Testing With Experimental JIT, Better Interactive Interpreter

The first beta of Python 3.13 is now available for testing ahead of its official release later this year...

https://www.phoronix.com/news/Python-3.13-Beta-1

[timofonic]

Will they manage to make Python considerably faster? I don't think so! Please someone prove me wrong!

[Setif]

Will they manage to make Python considerably faster? I don't think so! Please someone prove me wrong!

As many pointed previously in this forums, If you want speed don't use Python. Python is for scripting.

[edxposed]

Just build Python with frame pointer and O2, the slower Python is the more it meets the user's needs.

[timofonic]

As many pointed previously in this forums, If you want speed don't use Python. Python is for scripting.

Scripting languages are toys that underperform in demanding tech environments. Their rampant overuse signals a quiet calamity, undermining the software ecosystem and fostering a generation of coders accustomed to simplicity over security. An urgent shift is imperative for the sake of resilient, future proof and secure software.

[dEnigma]

Scripting languages are toys that underperform in demanding tech environments. Their rampant overuse signals a quiet calamity, undermining the software ecosystem and fostering a generation of coders accustomed to simplicity over security. An urgent shift is imperative for the sake of resilient, future proof and secure software.

You should turn down the bullshit dial in your ChatGPT settings.

[Old Grouch]

As many pointed previously in this forums, If you want speed don't use Python. Python is for scripting.

For some important use-cases python is faster than (bash) shell scripts, and also spreadsheet macros, so speed is relative. The flexibility and ease-of-use of python for inexperienced programmers means that it has a low barrier to entry in order to gain meaningful performance improvements in those, and other areas.

Comparing python to compiled code is not always the correct comparison: comparing it to other interpreted languages is.

Python can also be used for rapid prototyping of code. Once you have the functionality defined, you can optimise later by coding in a compiled language, Early optimisation is generally regarded as a bad thing.

[npwx]

Python can also be used for rapid prototyping of code. Once you have the functionality defined, you can optimise later by coding in a compiled language, Early optimisation is generally regarded as a bad thing.

I did just that and it was a waste of time, would have been way faster to write in Go from the beginning and saved me the horrible experience of dealing with Python mulithreading/tasking. Even if they drop the GIL, their threading abstractions are still garbage.

[Lycanthropist]

Python can be high performance, when using the right libraries, e.g. taichi-lang.

[sobrus]

I like both Python and Rust and use both. They both have their use cases and for most applications python is fast enough (you can use numba if you want JIT or run python code on a GPU).
But I agree that it's a waste of time to prototype in Python, as writing it in Rust or Go us usually not that much slower and you don't need to avoid features that your final language doesn't support (like operator overloading or something), and can't use what they offer at the same time (like channels). Basically it's writing the same thing twice.

Can they make it faster? Of course they can. Javascript used to be very slow too, but this is no longer the case.
Removing GIL and adding JIT is a good starting point.