Developers: | University of Massachusetts Amherst (UMass Amherst) |
Date of the premiere of the system: | August 2023 |
Branches: | Information Technology |
Technology: | Application Development Tools |
2023: Product Announcement
In late August 2023, researchers at the University of Massachusetts Amherst announced the creation of the Scalene profiler, which allows you to repeatedly speed up the work of programs written in Python.
It is noted that Python-based software is often very inefficient in terms of performance: it can be up to 60,000 times slower than code written in other programming languages. Despite this, because of its usability, Python became very popular in the era of data sciences and machine learning.
To combat Python inefficiency, developers can use tools called profilers. Such tools determine which parts of the program work slowly. However, existing profilers are not very useful: at best, they point to an ineffective section of code without making any recommendations to fix the problem. The new Scalene tool is designed to bypass existing restrictions.
According to the creators, Scalene not only accurately identifies ineffective blocks of Python code, but also uses artificial intelligence to propose measures to improve the program. In the process, the profiler determines what exactly the software spends the most time on: in this case, the use of resources of the central processor, graphics accelerator and memory is analyzed. Once trouble spots are identified, Scalene uses AI algorithms to identify ways to optimize individual lines or entire blocks of code. Thus, theoretically, the tool is able to speed up the work of programs written in Python by a thousand times. It is said that Scalene is actively used by the development community: by the end of August 2023, this profiler was downloaded more than 750,000 times.[1]