M.I.T. Floating Point Arithmetic Package

This document provides a technical overview and the assembly code implementation for a suite of floating-point arithmetic subroutines developed at the Massachusetts Institute of Technology.

Summary: The M.I.T. Floating Point Arithmetic Package represents numbers in the format $f \times 2^e$, where $f$ is an 18-bit one's complement fraction and $e$ is an 18-bit one's complement integer exponent. The package includes routines for:

• Basic Arithmetic: Floating addition (including unnormalized versions), multiplication, and division.
• Mathematical Functions: Square root, base-2 logarithm, exponentiation ($2^x$), and a selection of trigonometric and inverse trigonometric functions (sine, cosine, secant, cosecant, tangent, cotangent, and arctangent).
• I/O Operations: Routines for handling floating-point input (including character validation) and formatted floating-point output.

The document contains the full assembly language source code for these routines, which are designed for use with the Digital Equipment Computer Users Society (DECUS) library. Each routine is documented with instructions on how to call it, the required registers for arguments and results, and descriptions of specific behaviors such as error handling (e.g., division by zero or negative square root arguments).