Python 3.11 to be 2X faster

The upcoming version of Python is set to be twice as fast as the previous versions.

While Python is one of the most widely used languages commonly utilized for data analytics and Machine Learning than for actual market-making software. This is due to Python’s higher level of abstraction, which makes it slower as an interpreted language than alternatives like C++ or Java.

The modifications may help users of Python-based products, according to Guido Van Rossum, the author of Python. Banks like JPMorgan and Bank of America, which are heavy users of Python in their risk pricing systems — though JPMorgan has been slow to move away from Python 2 — and finance firms that utilize the language for data analysis may theoretically be included.

His plan includes an adaptive bytecode interpreter, frame stack optimization, and ‘zero overhead’ exception handling. If these changes double Python’s speed, Van Rossum believes that future changes may include a solid ABI (application binary interface) or machine code generation to boost Python’s performance even more.

However, if we think beyond big banks, a 2 to 3X speed improvement might mean a lot for data scientists and others who generate scripts that take a while to complete, assuming that the limitation is not input/output. However, in those cases, speed increases are possible when working with input files of different types CSV files are much faster to read than Excel files, and parquet or pickle files are even faster. So in short, who wants to wait, I will take any speed improvements that I can get. The presentation provides more details of his work.

FasterCPythonDark.pdf

Download PDF • 201KB