Thank you for posting a what's new for python vid, it is good to see more changes to the GIL and the adding of a JIT. It will be interesting to see where this goes, also liking the repl changes as I would have more of a reason to use it on small tests than before with the multi-line code etc.
They just made the JIT available with no optimizations implemented. Those optimizations will come on 3.14. Loops where already optimized whit the adaptative interpreter and that's how they made 3.11 25% faster than 3.10.
Two of the core developers run a podcast called core.py (which is amazing), and it was just a funny little joke based on something they said in their latest episode haha.
Aah right, idk what the specific meaning for that was. My best guess was maybe the view count of one of the episodes, or perhaps a video one of them had done, but I'm not sure.
Ok so it’s still Python, it still sucks at doing loops and still needs 10x+ performance speed ups to actually become usable for anything that isn’t automation or scripting. Gotcha
When you jit compile a function using @jit from numba module and numpy (just adding decorator for a function), Code runs 40-70x faster. Python can run anything, except an OS. Jit compiled Python is in avrg just 4x sloweer than unoptimized C++ code. Sometimes it's just as fast. For 99% of appications, you'll see no differences in a Python jit compiled code vs JS, Go, Lua, java... or even C because it's simply fast enough and HW became really fast in the las 10 years... You can write Autocad in Python and it will run fast. Just remeber Python code in REAL apps are executing C code most of time. Every lib you import and use is a C lib.
@@smanzoli I wonder how this would apply to those android builds as well. Multiplatform compiled python binaries could be interesting.. if it wasn't for the nightmare that is python package dependency management.
Python 3.14 will be awesome.
3.1416 would be more precise ;-)
Pithon
@@quettle xD
@@quettle πthon
It’s still Python
Thank you for posting a what's new for python vid, it is good to see more changes to the GIL and the adding of a JIT. It will be interesting to see where this goes, also liking the repl changes as I would have more of a reason to use it on small tests than before with the multi-line code etc.
They are heavily inspired by PyPy ;-)
Thanks for keeping us updated! Always good to know how the language evolves :)
Yet, they still don't have a docstring for each standard library module
Does the JIT have any perf improvements for simple loops? If it's only a few % it doesn't seem like it.
Loops were used as an example in the article I took a lot of info from, though I don't know exactly how the JIT has been implemented.
They just made the JIT available with no optimizations implemented. Those optimizations will come on 3.14. Loops where already optimized whit the adaptative interpreter and that's how they made 3.11 25% faster than 3.10.
can somwone explain outro?
Two of the core developers run a podcast called core.py (which is amazing), and it was just a funny little joke based on something they said in their latest episode haha.
i got that lol, for some reason it thought it had some other meaning i didint understand(about the 47 views)
Aah right, idk what the specific meaning for that was. My best guess was maybe the view count of one of the episodes, or perhaps a video one of them had done, but I'm not sure.
You're using Dell computers? Dell is Hell on earth
I use Dell _monitors_. I use a Mac mainly, but have a custom PC as well.
Dell computers are hot garbage. I have 3.
Ok so it’s still Python, it still sucks at doing loops and still needs 10x+ performance speed ups to actually become usable for anything that isn’t automation or scripting. Gotcha
When you jit compile a function using @jit from numba module and numpy (just adding decorator for a function), Code runs 40-70x faster.
Python can run anything, except an OS. Jit compiled Python is in avrg just 4x sloweer than unoptimized C++ code. Sometimes it's just as fast.
For 99% of appications, you'll see no differences in a Python jit compiled code vs JS, Go, Lua, java... or even C because it's simply fast enough and HW became really fast in the las 10 years... You can write Autocad in Python and it will run fast. Just remeber Python code in REAL apps are executing C code most of time. Every lib you import and use is a C lib.
what a butthurt comment
Lol it's funny whenever JS developers cry about python
@@architmishra015yeah js is so slow arm literally added cpu instructions to handle javascript functions 😂 Not even python is that bad
@@smanzoli I wonder how this would apply to those android builds as well. Multiplatform compiled python binaries could be interesting.. if it wasn't for the nightmare that is python package dependency management.