ABSTRACT
This chapter presents the formats and standards used in storing numerical data in the Intel math units. Supported data types include integer and real binary numbers and binary-coded decimals. It examines the development of conversion routines for input of user data into the math unit, and for converting math unit numeric data into strings of user-readable ASCII digits. In the Intel math unit binary integer numbers can be stored in three formats. The word integer format occupies two bytes, the short integer format takes up a doubleword, and the long integer format a quadword. The designers of the math unit provided some compatibility between the binary integer data types and the data types used by the 80x86 CPU family. The memory image of a number encoded as a decimal integer consists of 10 bytes, with the high-order byte located at the lowest numbered memory address.
