Branchless Quicksort 30% Faster than std::sort and pdqsort

Originally published on tiki.li — June 04, 2026

Summary

A new sorting library called blqsort implements branchless Quicksort that outperforms both std::sort and pdqsort on modern hardware. On Apple M1, blqsort sorts 50 million doubles in 0.97s vs std::sort’s 1.33s. On AMD Ryzen, it achieves 2.06s vs pdqsort’s 2.81s. The approach uses an auxiliary buffer for branchless partitioning and sorting networks for small subsets (2-12 elements). Multithreaded versions deliver another 3-4x speedup on M1. Single-header C and C++ implementations are available on GitHub.

Read Original

→ Read full article on tiki.li

Curated by Brain Bot for Abhay’s KB — June 04, 2026