Shulou(Shulou.com)06/03 Report--

This article introduces the relevant knowledge of "what are the plans to make Python faster". In the operation of actual cases, many people will encounter such a dilemma. Next, let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

It is true that Python is not as fast as C or Java, but there are some projects that are trying to make Python faster.

The Python code is simple and clean, but everyone knows that Python runs relatively slowly-- slower than C, Java, and Javascript on CPU-intensive tasks (but most services are not CPU-intensive)-- but some teams want Python to be perfect, so they are prepared to improve Python performance from the inside out.

If you want Python to run faster on specific hardware, you have at least two options, each with some drawbacks:

You can create a replacement for the Python runtime, but in the end you will find that you have rewritten CPython.

You can rewrite some of the existing code to take advantage of some performance optimization features, but the disadvantage is that programmers need to do more work. Here are five existing solutions to help you improve the performance of Python.

1. PyPy

PyPy is the most prominent alternative to CPython (for example, Quora uses it in a production environment). It also has the best chance of becoming the default interpreter, which is highly compatible with existing Python code.

PyPy uses just-in-time compilation to speed up Python, a technology also used by Google, which Google uses in the V8 engine to speed up Javascript. A number of performance enhancements have been added to the recent version of PyPy2.5, including a popular one that integrates with Numpy,Numpy and has also been used to speed up Python.

Code that uses Python3 needs to use PyPy3 accordingly. PyPy currently supports only Python3.2.5, and support for Python3.3 is in progress.

2. Pyston

Pyston, funded by Dropbox, uses the LLVM compiler architecture to speed up Python, as well as just-in-time compilation. Compared to PyPy,Pyston, which is still in its early stages, it supports only some of the features of Python. Pyston divides the work into two parts, one is the core features of the language, and the other is to improve performance to an acceptable level. There is still some way to go before Pyston can be used in a production environment.

3. Nuitka

In addition to replacing the Python runtime, some teams are trying to convert Python code into code in other languages that can run efficiently locally. One of the famous projects is that Nuitka-- converts Python code into C++ code-although the runtime still relies on the Python runtime. This limits its portability, but the performance improvement is considerable. In the long term, Nuitka also intends to enable the C language to call Nuitka-compiled Python code, so that the performance improvement will be more significant.

4. Cython

Cython (Python's C language extension) is a superset of Python that compiles Python code into C code and interacts with C and C++. It can be used as an extension of the Python project (to revisit the high-performance parts) or alone, without involving traditional Python code. The disadvantage is that you do not write Python, so you need to migrate manually, lack of portability.

It is said that Cython provides features to make code more efficient, such as variable typing, which is essentially required by C. Some scientific computing packages, such as scikit-learn, rely on some features of Cython to keep the operation simple and fast.

5. Numba

Numba combines the ideas of the above projects. I learned that Cython,Numba also uses a partial acceleration strategy, speeding up only CPU-intensive tasks; at the same time, it also learns from PyPy and Pyston, running Python through LLVM. You can use a decorator to specify the function you want to compile with Numba. Numba inherits Numpy to speed up the execution of the function. Numba does not compile at the right time, and its code is precompiled.

The father of Python said: most applications that think Python is slow are not using Python correctly. There are several ways to improve performance for CPU-intensive tasks-using Numpy for calculations, calling external C code, and avoiding GIL locks as much as possible. Since GIL locks are currently irreplaceable, many projects are starting to try some short-term viable alternatives, which may also be converted to long-term options.

This is the end of the content of "what are the plans to make Python faster?" Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.