|
CPU Revisited
Whether it was for scientific use or business
use, early computers had specific processors developed for each
particular purpose. The biggest problem with this, was that
if you required a computer for your particular operation, then
it had to be designed with that purpose in mind. A processor
has a limited number of instructions that it understands called
its instruction set. Each instruction involves a series of logical
operations (possibly thousands) that are performed to complete
each task.
CISC Technology
IBM realized that, instead of having to develop a new processor
for every new computer, what was needed was a processor that
could be used for multiple purposes. The saving in research,
development and manufacturing costs alone would be staggering.
CISC technology (Complex Instruction Set Computing) combined
the different instructions into one single CPU, and each instruction
had the ability to perform several different tasks based on
mini-programs or microcode integrated into the processor.
Of course, this technology would have its drawbacks.
An increased number of instructions (200 to 300) meant a much
more complex processor, requiring millions of transistors. Instructions
were different lengths, using 8, 16, or 32 bits for storage.
This resulted in a great deal of the processor's time being
spent calculating where each instruction began and ended.
One of the biggest drawbacks however, wasn't
realized until later. With thousands of Application Software
programs being written for the processor, any new chips Intel
developed had to be backwards compatible, which limited the
ability for them to take advantage of a lot of new discoveries
and advancements being made in processor technology.
RISC Technology
Not having to stay backward compatible, allowed other developers
the opportunity to research and improve on processor technology.
It was discovered that some of the more complex instructions,
and the mini-programs (microcode) integrated into CISC processors,
were not entirely necessary. Using several of the less complex
instructions together could complete the same tasks in less
time. RISC technology (Reduced Instruction Set Computing)
takes advantage of this. Using fewer instructions (128+) requires
fewer transistors, which results in reduced manufacturing costs
and a more stable, cooler operating CPU. Also, each instruction
is a fixed size (32 bits). This means that the processor doesn't
have to use up any of its valuable time figuring out where each
instruction begins and ends.
This is a basic introduction to RISC and CISC
technology. There is of course, a lot more to it. If you're
interested, by all means do some research into it. An excellent
resource is the Winn L Rosch Hardware Bible. Really though,
whether your computer uses CISC or RISC technology is really
a moot point. There's a lot more software out there for CISC
computers but, although the two technologies are not compatible,
information can be translated between the two.
Intel's processors now use a hybrid of both
technologies.
Pipelining
No matter which instruction set a processor is using, it often
has to perform several functions to execute each instruction.
Different circuits, or different parts of the processor, can
be involved in performing each function. Early processor circuits
could only handle one function at a time, and required an entire
clock cycle to perform each function. Thus, a single instruction
could take up several clock cycles. Pipelining is the process
whereby each function is recognized, and pipelined to that particular
part of the processor that deals with it. A CPU can receive
a single instruction, begin the first function, and receive
another instruction before it's done with the first. This allows
for more instructions to be performed in fewer beats of the
clock, and averages out to about one instruction per clock cycle.
continued...
|
