Click above to get retro games delivered to your door ever month!

  There are various ways to produce an executable program whose memory map
  comprises separate code, data, and stack segments. The way a particular
  program addresses the different areas of its memory map is determined by
  the program's memory model.

  A memory model specifically describes how executable code and data are
  addressed within a program. For example, the 8086 imposes a limit of 64 KB
  in any given segment, so a program with more than 64 KB of executable code
  must be mapped into more than one executable code segment. Similarly, a
  program with more than 64 KB of data must store that data in at least two
  different data segments. Thus the simple memory model shown in Figure
  19-1 can be elaborated upon--into four different memory models. (See
  Figure 19-2.)

  The memory model you use affects how your program uses segment registers.
  In a small-model program, the CS and DS registers can be initialized at
  the start of a program and left undisturbed for the duration. Contrast
  this with a large-model program, where the CS register must be changed
  whenever the program branches from one code segment to another, and the DS
  or ES registers must often be updated whenever data from different
  segments must be accessed.

  Some high-level language compilers let you specify which memory model to
  use. (See your compiler documentation for more information.) If you know
  your program contains fewer than 64 KB of executable code and fewer than
  64 KB of data, you can explicitly request such a compiler to generate a
  small-model executable program. (This is the memory model we have used in
  all the assembly-language examples in previous chapters.) Other compilers
  can use a compact, medium, or large model, regardless of the program size.
  Whatever the case, you should know what memory model your compiler uses if
  you want to understand how the different parts of an executable program
  fit together.

  Model          Number of Code Segments      Number of Data Segments
  --------------------------------------------------------------------------
  Small          1                            1
  Compact        1                            More than 1
  Medium         More than 1                  1
  Large          More than 1                  More than 1
  --------------------------------------------------------------------------

  Figure 19-2.  Four common memory models.