Get the latest tech news

Lessons learned from profiling an algorithm in Rust

cently, I've been working on a new approximate nearest neighbor search algorithm called RaBitQ. The author has already provided a C++ implementation that runs quite fast.

Then I found that the CPUs are not completely idle as I have VSCode + Rust Analyzer, it seems they don't consume much CPU but they do affect the benchmark results heavily. Also, a mature linear algebra library provides many useful functions for manipulating the matrix and vectors, making it easier for me to implement the algorithm. Imagine that implementing an algorithm involving matrix multiplication, projection and decomposition in Python without numpy, it's a nightmare.

Get the Android app