Faster than Memcpy

Explore optimization techniques for high-performance image processing in this code::dive 2019 conference talk. Delve into optimization patterns and pitfalls encountered in industrial machine vision software, where algorithms must operate faster than memory copy. Learn about efficient multi-threaded code and optimal SIMD instruction usage for the x86-64 architecture. Discover how to achieve real-time performance in image processing algorithms, going beyond computational complexity analysis. Gain insights from speaker Mateusz Brzeszcz, an experienced software developer specializing in real-time image processing algorithms.

Intro
Measuring
Vectorization
Filtering
Questions