                              RISC

  By: TIM FARLEY

  RISC stands for "Reduced Instruction Set Computer", which is a
  theory that someone came up with that if you severely reduce the
  complexity of operations that a computer can perform, limiting it
  to a small set of very simple operations, you can build a much
  faster computer.

  For instance:  the CPU's in your PC-compatible have a very
  complex, involved instruction set.  The instructions can be from
  one byte to umpteen bytes long (I forget the exact maximum).  The
  execution times vary from one clock cycle to hundreds of clock
  cycles per instruction. All of this makes the CPU very complex
  and causes it to be very hard to predict the execution times of
  software.  Optimization of software is also very hard.

  On a RISC computer, all instructions are generally designed to be
  (nearly, or exactly) the same length in memory, and to execute in
  (again, nearly or exactly) the same amount of time.  This makes
  the internal design of the processor simpler, and it means that
  you can process instructions more evenly as you go.  This in
  theory leads to faster processors and better optimized programs.

  What you lose in a RISC machine is neat instructions like, for
  instance, the built-in integer MULtiplication and DIVision
  instructions in the 386.  Instead, these get replaced with
  software subroutines that run nearly as fast.

  The jury is out on RISC, as it remains unclear whether it has
  real benefits.  Meanwhile, vendors of non-RISC processors like
  Intel have included "RISC like" features in their CPU designs to
  increase speed.

  For instance, the 486 will "pipeline" instructions so it can
  partially execute the next instruction before it is done with
  this one.  This is a feature that originally came about in RISC
  machines.
