C++ implementation - Show & Tell (sametuca)

C++ Implementation - 1BRC (Show & Tell)

Hi! I've implemented a high-performance C++ solution for the 1 Billion Row Challenge. I know the official challenge is Java-only, but I'd like to share it here as mentioned in the FAQ (Show and tell area for non-Java implementations).

Repository: https://github.com/HappyCerberus/1brc (PR #4 - sametuca)

Optimizations applied:

• Memory-mapped I/O with OS-level sequential prefetch hints
• Custom open-addressing hash map with inline key storage
• wyhash, SSE2 SIMD newline scanner, branchless temperature parse
• Thread affinity, prefetch, cache-line alignment, direct output

Compile: g++ -O3 -march=native -std=c++17 -o sametuca src/sametuca.cpp -lpthread

Happy to discuss or share benchmarks if of interest.