There are two instruction formats, referred to as "Type A" and "Type B". Most instructions were of type B. Type A instructions have, in sequence, a 3-bit
prefix (instruction code), a 15-bit
decrement field, a 3-bit
tag field, and a 15-bit
address field. There are conditional jump operations based on the values in the index registers specified in the
tag field. Some instructions also subtract the
decrement field from the contents of the index registers. The implementation requires that the second two bits of the instruction code be non-zero, giving a total of six possible type A instructions. One (STR, instruction code binary 101) was not implemented until the
IBM 709. Type B instructions have, in sequence, a 12-bit instruction code (with bits 2 and 3 set to 0 to distinguish them from type A instructions), a 2-bit
flag field, four unused bits, a 3-bit
tag field, and a 15-bit
address field. • Fixed-point numbers are stored in binary
sign/magnitude format. • Single-precision
floating-point numbers have a magnitude sign, an 8-bit excess-128 exponent and a 27-bit fraction (no hidden bit). • Alphanumeric characters were usually 6-bit
BCD, packed six to a word. The instruction set implicitly subdivides the data format into the same fields as type A instructions: prefix, decrement, tag and address. Instructions exist to modify each of these fields in a data word without changing the remainder of the word, though the
Store Tag instruction was not implemented on the IBM 704. The original implementation of
Lisp uses the
address and
decrement fields to store the head and tail of a
linked list respectively. The primitive functions
car ("contents of the address part of register") and
cdr ("contents of the decrement part of register") were named after these fields. ==Memory and peripherals==