Talk Preview: A new dragon in the den: fast conversion from floating-point numbers
summary
Covers a new algorithm that converts IEEE floats and doubles to textual form by deriving the conversion from first principles, implementing it in C++ and comparing it to existing Dragon, Grisu, Ryu, Errol and Dragonbox approaches. Claims a faster and simpler implementation that still guarantees correct roundtrip/shortest representations, explains the actual arithmetic and table choices that make it fast, and shows microbenchmark results and a working library.
AI-written summary · about
related talks
11mPrinting floating point numbers is surprisingly hard!!
01h 41mFast Conversion From Cpp Floating Point Numbers
14mPrinting Floating Numbers is Surprisingly Hard!
58mPortable floating-point calculations
01h 00mFloating-Point ＜charconv＞: Making Your Code 10x Faster With C++17's Final Boss
32mEnter the Dragon: Taming LLVM
26mSlaying the Dragon
06mLightning Talk: A Compiler Optimization and the 2-Adic Numbers