Why is Python slow?

Many times people argue that python is slow compared to C++ or JAVA. They even argue that look at Javascript, it is also an interpreted language and yet python is slower compared to Javascript. I asked this question to python steering council member, Pablo Galindo. This council is a group of 5 people who decide the future direction of python programming language. In this video I will quickly summarize the why python is slow? If you want to watch entire conversation where we discussed other topics as well than watch this video,

#️⃣ Social Media #️⃣

❗❗ DISCLAIMER: All opinions expressed in this video are of my own and not that of my employers'.

[AlessandroBottoni]

Perfect explanation! Kudos! Thanks for having made clear these points to the people who were unaware of them. I'm using Python since 1998 and I used it for everything, from web development to data science passing from system administration. The only cases where I found it "unusable" was when there was not room enough for the Python interpreter (like in embedded systems). It has never been *slow*. It is just a matter to understand *how* to use it. Python is a "glue language" and is intended to be used to glue together C and C++ libraries, offering a better API to the developer. It is intended to be fast at *developing time*, not at *execution time*. The people who complain about its speed just do not understand this point and just do not know how to use it in the proper way.

[randomdude]

There is something called Numba that can be used to speed up Python.

[Carurd]

An excellent approach to the problem. !

[amritgurung]

Whenevef i see 20 min or more long videos, i am like Ehh.. 😕😕 but thats not the case with codebasic. I just like his super clear, concise and short yet meaningful explanation.

[krishanthadharmasiri]

just got me thinking, if numpy & pandas based on c/c++, why cann't we directly use c++ for ML which would be really really faster? thnaks for the video very informative

[softcoda]

Thanks for this...its very enlightening. I didn't know numpy/pandas are wrappers over C/C++, thus making heavy programs that use them faster.

[shreehari]

The speed of the programming language also depends on the processor, when i was doing data cleaning for a nlp project using .apply() function on my i3 processor laptop it took me about 30 minutes, and when did the same on my i7 processor it took me about 8 to 10 minutes. Processors also play an important role

[Sovereign_Lord]

Hi.. Can you please give a session on how to approach IPA? intelligent Process Automation. There are lot of example based on IPA but i would like to know where to start, tools to learn from Beginner perspective to become proficient.

[vuufke]

yeah.. NO, Lua is not even a fraction closer to being as popular as either Py or Js, it doesn't have any significant implementation anywhere, or any major funding behind it, yet and still beats both Js & Py.
that's not the problem here, the real problem is the Py interpretation layer not being good enough

[Travio]

If numpy and pandas use the C++ under the hood, why doesn't the plain python also use a wrapper approach?

[ThankYouESM]

88x is not near fast enough for me... it needs to be over 1000x faster like C++. However... I kept believing since around the year 2000 that somebody will soon easily find a way to make a free framework in which the heavy number crunching is to be remotely handled by a background C++ buffer application... therefore, we get the best of both worlds combined.

[sartajsingh]

Pls explain Data Science vs SRE as career choice in a video

[ArThoX]

At the end it's always about python fanboys trying to explain why they are slower than JS and extremely slower than C# and Java 😂

[codeforcoders]

Then why don't Tesla fund to make python Faster, because Tesla self driving car work on machine learning and AI

[laszlob]

3:20 how does chrome on a v8 engine like is it a hellcat tho

[Rudxain]

The Python council seems equivalent or analog to the TC39 commitee

[RodrigoDoTango]

So many miss information.... 1st, from your perspective, if we also use the best option available for each language, we also will have a performance boost that will take Python to be slow again during benchmarking, even with the best approach as possible. 2nd, C++ and special Java are also both high level languages and it's not needed to Google putting a lot of money to became a well performance language, you can make tests and even PHP8 is a lot faster than Python with best design pattern. The thing is that Python is not that slow as people complain, it's just slower than most of popular languages, but it's still faster than most of the languages, like C# (which is compiled) for instance, Lua between others. Python have several reasons to not be a faster language, one of them is what can make C# faster than Python, which is the use of concurrency in C#. For the design that Python have, it's not that bad performance, but it should be at least as fast as PHP8 to be considered a good performance language. The reason why Python is used so much in machine learning is not related to the speed, but because machine learning is difficult to code and Python is a lot easier than most of the languages that are faster.

[ShaarifDaGama]

Which Laptop I should buy for Preparing Data Scientists & Exel ML, My budget is under 30k

[Test-krfr]

Is their any plan to make django framework tutorials ?

[rajashekarm]

if numpy or pandas are recommended, I think it should be part of the base python pkg