The author implemented a program to visualize a floating point timeseries compression method called Gorilla compression, where subsequent values are XORed for compression. The unique algorithm helps compress values, and a simple heuristic aids in choosing between two compression methods. Real data application shows distinct compression patterns. A tweak to the algorithm is proposed for a more optimal compression strategy by resolving a potential failure mode. The implementation showcases efficient compression compared to single precision floating point values, making the compression factor significant. The article gives detailed insights into the compression process and provides code references for implementation in Rust.
https://clemenswinter.com/2024/04/07/the-simple-beauty-of-xor-floating-point-compression/