5.3 Universal registers for 16-bit and 32-bit modes
----------------------------------------------------
When creating function libraries it is frequently necessary to write
variants of a function for use in 16-bt and 32-bit modes which differ only
in their use of 16-bit or 32-bit registers. The new register syntax makes
it possible to write only a single function. If the compiler encounters
the syntax:
(E)AX=0;
it will use register AX when compiling 16-bit code and EAX when compiling
32-bit code.
Automatic registers are used to simplify library files and make them
more comprehensible.