Double Float Mathematical Routines

This document serves as a library write-up for the "DBLFLT" series of mathematical subroutines designed for LINC-mode computers, including the LINC, LINC-8, and PDP-12. These routines perform double-precision floating-point operations using a 24-bit mantissa and a 12-bit exponent.

Key components of the document include:

• DBLFLT: The original mathematical routines written for the classic LINC computer.
• DBLFLT1: A revised and more manageable version of DBLFLT, featuring simplified entry points and mnemonic symbols for easier programming on the PDP-12.
• DBLFLT3: An extended package of subroutines built upon DBLFLT1. It provides a broader set of mathematical operations, including sine, cosine, square root, logarithms, and power series, along with teletype input/output routines.

Core Technical Information:

• Functionality: The routines handle basic arithmetic (add, subtract, multiply, divide, absolute value, complement), conversion between fixed and floating-point formats, and data movement.
• Operation: Users interact with these routines via JMP instructions to specific entry points. Detailed calling sequences, argument handling (including the use of index registers), and memory field requirements are provided for each version.
• Format: The document explains the "DBLFLT" normalized binary floating-point format, which consists of three consecutive memory words (one for the signed exponent and two for the signed mantissa).
• Tables and Appendices: The document includes comprehensive tables detailing subroutine entry points, argument requirements, register usage, and typical execution times.
• Modification: It provides instructions for adapting older programs written for the original DBLFLT to use DBLFLT1 or DBLFLT3, and outlines how users can customize the packages for specific memory configurations or speed requirements.