Emulation of 3Sum, 4Sum, the FMA and the FD2 instructions in rounded-to-nearest floating-point arithmetic
Stef Graillat, Jean‐Michel Muller · HAL (Le Centre pour la Communication Scientifique Directe) · 2024
We give high level algorithms that make it possible to compute the rounded-to-nearest sum of 3 or 4 floating-point numbers in binary floating-point arithmetic. They can easily be adapted to emulate the FMA (ab + c correctly rounded) and FD2 (ab + cd correctly rounded) instructions.