Numsectorlisp is a library for SectorLISP, for scalar, vector, and matrix multiplication of fixed-point numbers. The number of bits for the fractional part and the decimal part can be easily configured. The default setting implements 16-bit fixed-point numbers with 12 bits for the fractional part, 3 bits for the integer part, and 1 bit for the sign. Instructions for configuring the bit sizes are available inside the source code.
An in-depth explanation of the implementation details are available in my blog post, Building a Neural Network in Pure Lisp Without Built-In Numbers Using Only Atoms and Lists.
Examples uses of numsectorlisp are available in these projects: