|
|
Fri Oct 2 12:42:34 EDT 2009
The Computer Merit Badge: how I'd complete the requirements NOW;
Part 3
--
This installment will not be as long or involved as the previous
2, but it does built upon what I discussed in them
The second part of requirement #1 says, "Describe the major parts
of a computer system." And this is what I will do, to some de-
gree.
Today, mainstream computers are based upon what is called, the
"von Neumann" architecture. This architecture was one proposed
by a man named Jon von Neumann, who in his 50 years of life con-
tributed significantly to many scientific fields. Computer Sci-
ence was one of them, and his contributions to the field arguably
put computer technology (and the US) in a position to advanced
significantly in the years to follow.
The von Neumann architecture was proposed as a solution to the
problem of how to construct a device that physically realized the
theoretical general purpose computing model proposed by Turing.
This model is briefly discussed in the first installment of this
series, but in short it provides the theoretical basis for con-
structing machines that are able to simulation other "machines".
The machines that a Turing Machine can simulate are realized to-
day as computer programs. The clear difference between the theo-
retical concept of a Turing Machine and what is physically real-
izable is, of course, the memory in a real machine can't be infi-
nite. So, technically, any computer today (of von Neuman archec-
ture or not), is really what is known as a "linearly bounded au-
tomata." But I digress.
So what does the von Neumann architecture consist of? Well:
1. a processing unit (i.e., a CPU)
2. memory, providing random access for the reading and writing of
programs and raw data
3. facilities for input and output devices
Is this literally similar to the Turing Machine? No. A Turing Ma-
chine actually consists of a tape (i.e., the memory + input) and
a "read/write" head to manipulate the tape. However, the
read/write head allows for random access to memory (albeit not
necessarily direct). There are technically no input or output de-
vices beyond the tape. The initial values of the tape and start-
ing position of the read/write head are all the TM provides for
input. And to make matters more complicated, the "input" of the
tape consists of data and program instructions encoded together.
TMs are very interesting devices, but I do not have the time nor
expertise to discuss them much here.
Now, back to the major parts of a computer system. Modern com-
puters are very complex, but the basic parts that make in them of
the von Neumann type are there.
1. a processor
2. RAM (i.e., random access memory)
3. Hard drive (i.e., a persistent instance of random access memo-
ry, a.k.a, "storage")
4. input (keyboard, mouse, cd-rom/dvd, etc)
5. output (monitor, printer, cd-rom/dvd, etc)
Are there other types of architectures? The answer is YES, and
they are called collectively as, "non-von Neumann" architectures.
And while some of them are modeled off of the Turing Machine con-
cept, not all are. For example, a man name Alonzo Church (i.e.,
Turing's PhD Advisor), created an equivalent but completely dif-
ferent may to investigate the questions of "computability" that
Turing was addressing with his models. Church's formulation has
come to be known as "Lamda Calculus," and is the basis for an ap-
proach to writing programs called, "functional programming." I
will discuss this more when I discuss the requirement about pro-
gramming.
Since the time of Turing, Church, and von Neumann, there have
been many conceptual models proposed that are as "powerful" as
the Turing Machine. Few, however, have provided the same basis
for building computational devices than von Neumann's. One inter-
esting factoid is that when one wishes to prove that their con-
ceptual model is as powerful as a TM (or "universal"), the easi-
est way to prove this is to build a functioning TM out of their
concept. This can be done, for example, using the set of rules
shown to be universal which are called, "Conway's Game of Life."
(check out a little tool called Golly)
So what sets the von Neumann architecture apart from the others,
including machines that mimick the operation of a Turing Machine
(well, linearly bounded automata)? The answer goes back effi-
ciently accessing that vital ingredient required for computation-
al power by a device - MEMORY! The von Neumann architecture is
conceptually meant to facilitate one thing - the efficient access
to memory. Whereas in a TM, there is literally a single tape and
read/write head containing the input data, the program's instruc-
tions, and the "working" memory required for the execution of the
program; the von Neumann architecture provides direct (the key)
access to all of the memory, which is intuitively how we think
about the memory on our computers today. Another way to say this
is, the von Neumann architecture gives one direct access to a
running program's "state," whereas other paradigms either do not
(in purely functional approaches) or in a very convoluted way
(e.g., a "real" TM or a TM built from cellular automata in the
GOL).
This way of thinking has also affected how most think about pro-
gramming and programming languages. Relatively few brave souls
engage in functional programming, as compared to those who engage
in programming that is conceptually based on the von Neumann ar-
chitecture. While today, many traditional programming languages
are incorporating "functional" features that give some conve-
niences to the programmer, writing a program that is meant to be
executed in a top-to-bottom way (i.e., procedurally) and that
stores values in "variables" relies on the view that the computer
executing the program has direct access to the place in memory
where this variable is storing its data. Since it is usually of
no concern where in memory the computer stores this information,
the programmer just assumes "on demand" access to random memory
locations. And this is why the most powerful access model of mem-
ory is called "random access." At some point when you take the
time to learn more about the Turing Machine and the hoops a "pro-
gram" must jump through to store and retrieve the data written to
the tape, an appreciation for the direct access to memory that
the von Neumann architecture will developed. Not only does the
von Neumann architecture actually facilitate programming lan-
guages and methods that assume direct access to memory anywhere
and at anytime, but through these languages and methods it facil-
itates the kind of programatic thinking that most Humans are in-
clined to learn the most easily. Some would argue that the von
Neumann architecure and the langauges and programming methods it
encourages are brutish and inelegant, but the simple fact is that
it has provide the world with a vast array of technologies that
might not have been practical otherwise.
Another characteristic of the von Neumann architecture that makes
it appealing, is that it is easy to manufacture computers that
implement this kind of architecture. Consider that the very first
commercially successful computer that was designed to be easily
manufactured and "upgradable" was the IBM 360. IBM pioneered the
standardization of parts and computer "families." And while some
might argue that their motivation was purely to sell as many com-
puters as possible, the motivation of the engineers was probably
just as strongly driven by the technical challenge of the
project. (I recommend reading Fred Brook's, "Mythical Man Month"
for insights into the effort to produce the 360 series).
There are many other reasons that the von Neumann architecture
persists, but it comes down to practicality - in design, manufac-
turing, and use (including programming). In the installment that
I write discussing the programming requirements of the Boy
Scout's Computer Merit Badge, I will discuss in more detail the
various approaches (or "paradigms") used by programming languages
and software developers - including more on "procedural" versus
"functional" programming, where "object oriented" programming
fits in, and will continue tying these issues back into the back-
ground I've provided here and in the other sections regarding
what a computer really is.
I will not be fulfilling the last part of requirement 1, which is
to list 4 uses of a computer. I believe one can do this on one's
own.
The next installment will be discussing requirement #2, which
says:
Do the following:
1. Tell what a program is and how it is developed.
2. Give three examples of programming languages and what types
of programming they are used for.
3. Describe a source program and an object program.
Until then, take care.
--
Powered by vee
Copyright © 2006-2010
|
|