Explore an in-depth case study on optimizing string performance for computer graphics applications in this CppCon 2022 conference talk. Learn about SIMDString, an open-source replacement for std::string designed for graphics applications that perform frequent copy construction and concatenation on small strings. Discover how SIMDString leverages SIMD instructions, employs a freelist optimized allocator, and utilizes various micro-optimizations to achieve over 5x faster performance than std::string. Examine the tradeoffs between different string implementations and understand why certain optimizations are particularly effective for graphics applications. Gain insights into improving GPU shader performance through optimized string handling techniques presented by research scientists from NVIDIA and Roblox.
Overview
Syllabus
Optimizing A String Class for Computer Graphics in Cpp - Zander Majercik, Morgan McGuire CppCon 22
Taught by
CppCon