#hpc #simd

branchless

Algorithms and data structures designed to maximize performance on superscalar processors

2 releases

0.1.1 Jan 2, 2024
0.1.0 Jul 12, 2023

#427 in Science

Unlicense

76KB
255 lines

Branchless

Algorithms and data structures designed to maximize performance on superscalar processors. It uses platform-specific SIMD instructions to avoid using branches, where possible.

These crates are experimental, and their APIs are subject to change at any time.

Ipv4 address parsing

Benchmark Standard library (ns) branchless (ns) CPU
Parse single ipv4 address 27 9 i7-7500U
Parse 10,000 ipv4 addresses 526,811 103,089 i7-7500U

Dependencies