We hear specifications of microcomputers which classify them as
n-bit computers, for example
16-bit computers.
n used to be equal to the length of the
accumulator1, since the accumulator was the recipient of incoming values (from the
memory) and the results of
arithmetic and
logical operations, which in turn were mainly transferred back to memory. Due to this fact the
accumulator was usually designed to have equal size to the size of the
data bus. Thus an n-bit microcomputer implied a
data bus and an
accumulator of size n.
However, life is rarely so simple, and putting the blame on large multi-billion
corporations is fun.
Intel, after announcing the popular microprocessor 80386DX which was a true 32-bit processor with a 32-bit accumulator and a 32-bit
data bus, decided this was too expensive a solution and released the 80386SX which had a cheap 32-bit accumulator, and cut the data bus down to 16-bits, in turn, forcing two
CPU cycles to load
data into the accumulator.
This scenario had once before been encountered, the 8088 and 8086 microprocessors were identical internally but had a different
data bus width (8 / 16 respectively).
1 The coolest register of them all, back with the 8080 they only had only one register to perform mathematical functions on, called The Accumulator, da daa!!