High-performance applications are often written in a lower-level language, such as C++ of Rust, rather than Python. This is common in the world of AI, where organisations including Tesla develop models in Python, and then rewrite them in C++ to improve performance. The primary allure of lower-level languages lies in their ability to provide finer control over system resources, thereby unlocking a significant boost in performance.

With a foundational understanding of C++, courtesy of my past engagement with microcontroller programming in C, the venture of elevating a basic Python package to C++ piqued my interest. As an Earth Science graduate, working in software engineering, it seemed fitting to work on something geospatial. In previous projects I’ve used fmalina/blocl-bnglatlon, which converts British National Grid (OSBG 36) to lat-lon (WGS84). Whilst efficient, I suspected I could improve the performance of this package by rewriting it in C++.

Aside from utilizing constants, not much has been altered in the code; it largely represents a straightforward translation to C++. The code’s available on GitHub.

Results

The following results were gathered over 100 runs.

Item CPU time (x 10^-4 seconds) Memory (MiB)
Python 9.27458 20.828125
C++ 0.81958 20.812500
∆ 8.455 ~0
  • Duration: The duration for processing in C++ is significantly lower, showcasing a reduction of −91.15%. This reduction in time is a massive gain in performance, which is essential in high-performance applications.
  • Memory: The memory consumption is nearly identical between Python and C++, with a slight reduction in memory usage in the C++ implementation. This was expected.

Conclusion

This exercise served as a valuable lesson in understanding the performance gains achievable when transitioning from a higher-level language like Python to a lower-level language such as C++. The drastic reduction in processing time reaffirms the importance of language choice in high-performance computing scenarios.

The code and the benchmarks clearly demonstrate the benefits of using a lower-level language for performance-critical tasks. For those looking to improve the performance of their applications, especially in the realm of AI or high-performance computing, transitioning to a lower-level language like C++ is definitely worth considering.