Recent searches
Search options
Administered by:
Is Python Really That Slow? https://blog.miguelgrinberg.com/post/is-python-really-that-slow
@miguelgrinberg urllib3 and Trio support PyPy and we reported many PyPy bugs over the years, and maybe one big CPython bug? (To PyPy’s credit, those bugs are often fixed in 24 hours.) Also most C extensions don’t work? PyPy isn’t that easy to use outside of toy examples I’m afraid.
@quentinpradet I remember this being the case a few years ago. I will revisit and see what can be done today. Maybe it is enough material for another blog post.
Very interesting read. I used Python extensively for data analysis and found the biggest advantage to be the extensive ecosystem of libraries. The speed of Python itself is not terribly relevant if all the heavy lifting is done in native libraries such as scipy or pandas.
I am not aware of such complete libs in other languages. So I dare say with these libs, the speed differential to other ĺanguages is negligible (especially if you need to implement complex algos yourself)
@robertmx Yes, fully agree. As I mention in the article, I think for most people there is a mix of native and Python dependencies. For my benchmarks I chose examples that are pure Python, but this is unusual in typical applications. In cases like yours, where most of the work is done by native language libraries you are already running at near native speeds, so none of the optimizations in the Python space will lead to noticeable improvements.
@miguelgrinberg ah yis, always appreciate your opinion! Same story. Just wait until No GIL, JIT. Then watch the community switch to mojo when it's ready. For certain it is worth it to be a Python dev. AI is C++ & Python. The ships have sailed. First & last. Python!!!
Here is an mp3 my TTS service created. It's this blog post.
