MIPS
A
processor obeys the instructions which make up a program. How fast the program
is carried out depends on how many instructions the processor can obey per
second. Following this idea to its logical conclusion you might think that it
would be possible to’ give each processor a speed rating in terms of the ‘number
of instructions per second’ it can obey, analogous to a ‘miles per hour’ rating
for a car. Unfortunately this isn’t possible because not all instructions that
the processor is called on to obey take the same amount of time. In other
words, how many instructions per second a processor can execute depends on the
particular instruction in which you are interested! For example, to make the
comparison between two different processors seem favorable in a particular
direction all you have to do is count how many simple instructions per second
one can execute compared to how many complex instructions the other manages.
You might think that this sleight of hand would be easy to detect, but in
practice it is not so easy to determine what is a simple and what is a complex
instruction when dealing with two completely different processors.
The idea of a measure of the number of
instructions per second can be rescued by introducing a standard mix of
instructions. that can be used to find out how many instructions on average the
processor will obey in a second. This corresponds to the often quoted measure
of processor speed - MIPS or Millions of Instructions Per Second. So a machine
that is capable of say 2 MIPS will obey 2 million instructions per second on
average.
Average MIPS is a useful measure of how
fast a processor runs but it can still be misleading. For example, suppose a
processor was very good at a particular operation - addition say - and your
application used this operation very heavily. The fact that it was
exceptionally good at addition would only increase the MIPS rating a little,
but it would have a powerful impact on your application! The point is that MIPS
is an
EmoticonEmoticon