You are on page 1of 10

COMPUTER

CIRCUITRY 339

to provide a framework foractivitiesin computer built with electric relays and then with vacuum tubes
chess and to encourage advances in the field. There as the first electronic components. The evolution in
are currently about 500 members. It published the electronic component technology has been motivated
ICCA Journal, the leading publication in the field. primarily by the desire to have faster and smaller
Information on the ICCA can be obtained by writing computers that consume very little energy. Electronic
to Don F. Beal, Secretary, ICCA, Department of computers built since the late 1960s have used metal-
Computer Science, Queen Mary and Westfield oxide semiconductor (MOS) transistors. Prior to that,
College, Mile EndRoad,London El 4NS, England bipolar junction transistors (BJT) gates were used;
(http://www.dcs.qmw.ac.uk/-icca; email: icca we discuss these transistor types later. Theearliest
@dcs.qmw.ac.uk). semiconductor computer circuits consisted of gates
mounted on printed circuit boards connected by cop-
Bibliography per wires. This technology has also undergone major
changes and today gates and interconnect wires are
1950. Shannon, C. “Programming a Computer for Playing
Chess,” Philosophy Magazine, 41, 256-275.
integrated into a singlesilicon wafer in VLSI (Very
1953. Turing, A. M. “Digital Computers Applied to Games,” in LargeScale Integration) technology (see MICROCOM-
Faster than Thought (ed. B. V. Bowden), 286-295. London: PUTER CHIP).
Pitman.
1975. Newborn, M. Computer Chess. New York: Academic This article is restricted to discussing the structure of
Press.
1977. Frey,P. (ed.) Chess Skill in Man and Machine.New York: MOS and BJT gates. It describes a systematic method
Springer-Verlag. to design two classes of electronic circuits: combina-
1990. Levy, D. N. L., and Newborn, M. How Computers Play tional circuits, which implement arithmetic functions,
Chess. New York: W. H. Freeman. for example; and sequential circuits, which implement
1991. Hsu, F.-H., Anantharaman, T., and Nowatzyk, A.
“A Grandmaster Chess Machine,” Scientific American, 263, 4 memory. Finally,it describes the use of “program-
(October), 44-50. mable logic” devices, which facilitate the often tedious
1997. Newborn, M. Kasparov Versus Deep Blue: Computer process of designing computer circuits.
Chess Comes of Age. New York: Springer-Verlag.
Monty Newborn
Boolean Algebra
Boolean algebra forms the theoretical cornerstone on
COMPUTERCIRCUITRY which modern digital computers are built. Boolean
algebra deals with functions and variables that take on
only two values, commonly denoted by either T and F
or 1 and 0. Using the axioms of Boolean algebra, it can
be shown that any Boolean function (a function over
Boolean variables), no matter how complex, can be
composed from at most three primitive operations:
Although the development of digital computers can be and, or, and not. This set of logic operations is therefore
traced back to Charles Babbage, who conceived a said to be functionally complete. In fact, there are two
mechanical machine with toothed wheels to perform primitive operations, widelyusedin the design of
arithmetic processes, electrical principles first found computer circuitry, that are functionally complete in
application in digitalcomputers in the form of electro- themselves. These are nand (equivalent to a not fol-
mechanical relays. The most prominent examples of lowing an a n d ) and nor (equivalent to a not following
this type of computer are theBell Labs relay machines an O Y ) . The implicationis that one needs only to design
( 4 . v . )and the Harvard Mark I (4.v.) and Mark 11. Even circuits for a functionally complete set of Boolean
while these machines were under construction in the operations in order to have the basic building blocks
early and middle 1940s, it was recognized that an elec- for a digital computer.
tronic computer would offer great advantages in terms
of computational speed. Electronic computers use Theassociation that isusually made between an
electronic circuits that interconnect electronic compo- abstract Boolean operation and circuitry that imple-
nents called gates. Gates implement basic operations ments that operation is through voltage levels. Thatis,
called Boolean or logic functions. This article starts digital circuitry is designed to respond to two voltage
with a brief overview of Boolean algebra, the theory levels, designated high and low (e.g. +5 volts and 0
that underlies such circuits. volts). The conventional method uses the high voltage
(or V d d ) to represent a 1 and the low voltage (or Gnd
The physical realization of electronic gates has under- (Ground)) to represent a 0. Other associations of
gone major changes since the 194Os, when they were Boolean values and circuit quantities are possible.
340 COMPUTER CIRCUITRY

nand and not

D- or nor
Figure 2. Symbols for FETs.

Figure 1. Symbols for logic gates.

connection, and the pattern ofFig. 3b is a parallel


MOS Logic Gates connection. These elementary connection patterns can
be used to build larger networks, as shown in Fig. 3c.
Circuits that implement the most primitive Boolean
functions are called gates. The symbols that are used We say that anetwork N of switches is activated if and
to implement the commonly used gates are shown in only if the two terminals X, Y of N are connected
Fig. 1. A small circle used in conjunction with any gate through a set of switches. For example, in Fig. 3a, if
denotes negation of that gate’s function. A = B = 1, both switches are closed and X is con-
Variations of metal-oxide semiconductor (MOS) tech- nected to Y. Therefore the network of Fig.3a is
nologies are currently used to design computers. activated by the assignment A = B = 1. One can verify
In such technologies logic gates are buildusing that, if A = B = 0, the network of Fig. 3c is activated.
controlled switches. As shown in Fig. 2, such switches
MOS logic gates are constructed by using networks of
consist of three terminals and can be either “n-type”
switches. As shown in Fig. 4, there are two distinct
or “p-type” (‘n’for negative and ‘p’ for positive). In an
models of logic gates. The type-a gate uses two net-
n-type switch, if G equals 1, terminal S is connected to
works of switches-pullup and pulldown networks.
terminal D. If G equals 0, then terminal S is dis-
The pullup network consists only ofpFETs and the
connected from terminal D. In a p-type switch, if G
pulldown network consists only of nFETs. These gates
equals 1, terminal S is disconnected from terminal D.
are designed so that, for any assignment of values
If G equals 0, then terminal S is connected to termi-
to the input variables, only one of the two networks
nal D. Physically, the “n-type” switch is an n-type field
is activated. If the pullup (pulldown) network is
effect transistor (nFET) and the “p-type” switch is a
activated, the output of the gate is set to the logic
p-type field effect transistor (pFET). For the purpose of
valu‘e 1 (0).Fig.5ais an example of a type-a gate.
this discussion, we will not be concerned with the If A = B = 1, the pulldown network is activated and
physics of these devices (see INTEGRATED CIRCUITRY). the output is set to 0. If A = B = 0, then the pullup
As shown in Fig. 3, such switches may be intercon- network is activated and the outputis set to 1. One can
nected to form a network having two distinct terminals, verify that the gate of Fig. 5a implements the nand
X and Y.The connection pattern of Fig. 3a is a series function.

A C
0 0

B
0

- D E
-
Yl
Y
(a) Series
connection (b) Parallel connection (c) Series-Parallel graph

Figure 3. Series-parallelconnections.
COMPUTER CIRCUITRY 341

vdd vdd
4
t
Input Pullup
network variables
-1 output

I Pulldown
Output Input
variables
Pulldown
network
network

Figure 4. Structure of MOS gates.

For any assignment of values to the input of type-b only one type of switch. Therefore, it was easier to
gates, the output is set to 0 if and only if the assign- perfect the nMOS technology. However, technologi-
ments activate the pulldown network. One can verify cal innovation has made CMOS an equallyfeasible
that Fig.5bis a type-b nand gate, and Fig.5cis a technology, and today almost all circuits are CMOS
type b nor gate. circuits.
Type-a gates in which the p d u p network consists only
One of the major reasons fordeveloping the more
of pFETs and the pulldown network consists only of
complicated CMOS technology can be understood by
nFETs are known as CMOS (complementary metal-
re-examining the structure of the nMOS and CMOS
oxidesemiconductor) gates. Circuitsconsisting of
logic gates. Consider the nand gates of Fig.5a, b.
CMOS gates are known as CMOS circuits. Type-b
If A = B = 1, the pulldown network of Fig. 5a is acti-
gates in which the pulldown network consists only of
vated but thepullup network of Fig. 5a isnot activated.
nFETs are known as nMOS gates. Circuits consisting
Therefore, in Fig. 5a, there is no conducting path from
of nMOS gates are known as nMOS circuits.
V d d (both 'd's refer to the transistor drain terminal) to
Historically, nMOS circuits preceded CMOS circuits Gnd. In fact, in the gate of Fig.5a, there is no
because nMOS circuits, unlike CMOS circuits, require assignment of values to the input of the gate that will

I vdd

B
C 0
A

output
A
- ? output

-l

A - L
B I-
B---l
I - -
A B
L I
Gnd Gnd I
Gnd
(a) CMOS nand gate (b) nMOS nand gate (c) nMOS nor gate

Figure 5. Examples of MOS gates.


342 COMPUTER CIRCUITRY

result in a conducting path from V d d to Gnd. On the We use the symbol of Fig. 6a to represent a BJT where
other hand, inFig.Sb, when A = B = 1, there is a G is the base, E is the emitter, and Cis the collector. If
conducting path from vd/dd to Gnd. Therefore, current the voltage at C is “sufficiently higher” than the
flows through the pullup resistor R, thereby dissipating voltage at E, terminal C is electrically connected to
power, something that does not occur with CMOS cir- terminal E (the switch is ON); otherwise, terminal Cis
cuits. This, in turn, hinders the integration of a large disconnected from terminal E (the switch is OFF).
number of nMOS gates on a single chip. This is one of
Logic gates using BJTs are similar (if not identical) to
the many reasons for using CMOS circuits for denser
type-b gates. In resistor-transistor logic (RTL),only
chips.
nor gates are available for designing circuits. It is for
this reason that RTL is said to be “nor logic.” A two-
input RTL nor gate is shown in Fig. 6b. It is similar to
Other Logic Families the nMOS nor gate of Fig. 5c.
Although MOS, more specificallyCMOS,isnow the
In diode-transistor Iogic (DTL), unlikeRTL, only nand
dominant technology, a number of technologies have
gates are availablefordesigning the circuits. These
been used for manufacturing computers. These tech-
nand gates use both diodes and transistors as active
nologies differ inthe type of switching devices used, as
devices. The symbol for a diode is shown in Fig. 6c.
well as how the devices are used in the design of logic
A diode is a unidirectional two-terminal device. If the
gates. The earliest computer circuits used vacuum
voltage at X is higher than the voltage at Y, terminal X
tubes, but since they are no longer in use, we will
is electrically connected to terminal otherwise, X is
not discuss the implementation of logic gates using
disconnected from Y. A DTL nand gate is shown in
vacuum tubes.
Fig. 6d. To understand the operation of this gate, note
The logic familiesthat we will discuss here use bipolar that if A(B) has the logic level 0 (Le. voltage 0), diode
junctiontransistors (BJTs)as the switching device. d l (d2) conducts. This implies that the voltage at P
LikeFETs,BJTs are also three-dimensional devices. is 0. V - is a negative voltage. Therefore, diodes d3,

l u x
(c) Diode
dl c-

d2
I
V-
(b) RTL nor gate (d) DTL nand gate

v+. I v+ I

VS

5
S

R3 V -

( e ) TTL nand gate (0ECL nand gate


Figure 6 . Implementation of some universal gates using BJT-basedlogic families.
COMPUTER CIRCUITRY 343

d4 conduct and Q is at a negative voltage (by adjust- Classes of Computer Circuits


ing the resistance value R3). This cuts T, the driver
transistor, OFF and the output F is at logic level 1. Depending onthe functions they implement, com-
If A = B = 1, diodes d l , d2 do not conduct. Diodes d3, puter circuits are divided into two classes: combina-
d4 conduct and Q is at a positive voltage (adjusted by tional circuits and sequential circuits. Combinational
resistances R1, R3). The driver Tis ON and the output circuits are the simpler of the two classes and, along
F is pulled down to logic level 0. with other circuit elements, are used in the design of
sequential circuits.
In transistor-transistorlogic (TTL), the only gates
available are nand gates. A TTL nand gate is shown in To understand the difference between combinational
Fig. 6e. It is very similar to a DTL nand gate, but is and sequential circuits, refer to Fig. 7a. The time
faster than either RTL or DTL.If either A or B is at interval is divided into subintervals 11, 12, . . . of equal
logic level 0, there is enough base-to-emitter voltage size. The intervals are marked by a periodiclogic
difference at either T1 or T2 to turn one or both of signal known as clock. The peviod of clock is equal to
them ON. This brings down the voltage of P to a the length of the interval. Every interval is divided into
sufficiently low value to cut T OFF. The output F is two parts. During the first part, clock is 1, and during
therefore at logic level 1. When both A, B are 1, then the second part, clock is 0. We will refer to them as the
both T1 and T2 are cut OFF and P returns to a high one and zero periods of the interval.
voltage. This turns TON and the output F is pulled to Every such circuit can be represented as shown in
logic level 0. Fig. 7b. Computation proceeds as follows. During the
In DTL and TTL, there is a driver BJT and circuitry is one period, the values of the inputs change. During
added to the base of this BJT to implement the nand zero period, the logic valuesat the outputschange and
function. In emitter-coupled logic (ECL), the base of settle down to their steady state value. This is known
the driver BJT is grounded and circuitry is added to as single-phase clocking.
its emitter to implement the nor function. ECL circuits Combinationalcircuits are circuits whose output,
are potentially faster than RTL, DTL, or TTL because, during any time interval, depends only on the values
unlike those logicfamilies, the driver BJTis never of the inputs during the current time interval and is
driven into saturation and can therefore switch much independent of the values of the inputs during the
faster from one state to another. If either A or B in preceding time intervals. The output of sequential
Fig.6fis at logiclevel1, either T1 or T2 conducts. circuits, during any time interval, on the other hand,
Therefore there is current through R1. This brings Q depends on both the values of the inputs during the
to logic level 0. The emitter of T4 is now at either 0 or current interval, as well as the values of the inputs
a small positive voltage(depending on the values of R1 during the preceding time intervals. Having noted this
and R3) and this cuts T4 OFF. S is therefore at logic difference, we will next look at some combinational
level 1. When both A, B are 0 , neither T1 nor T2 con- and sequential circuits and try to understand some
ducts. Q is at logic level 1. The emitter ofT4is at a simple design procedures for these classes of circuits.
negative voltage (close to 0). T4 conducts and S is set
to logic level 0. Note that Q and S are complements.
Combinational Circuits
One advantage of using logic gates implemented using
BJTsis that they have faster switching speeds than Boolean functions are functions overBooleanvari-
gatesusing MOS devices. However, the low power ables whose result can be only one of the two Boolean
dissipation of MOS gates, among other reasons, made values 0, 1. Combinational circuits implement Boolean
it easier to integrate MOS devices. Nevertheless, many functions. An example of such a function is the odd
BJT devices are still in use for special-purpose circuits parity function P, of n variables. P, equals 1 if and
or where higher driving capabilities are required.

r-l 1 n n -
only if an odd number of the n input variables equal 1.

'
Clock

Intervals 1 I3 I4
I I
-b
I
Time
I
1 Clock

(4 (b)
Figure 7. Global clock.
344 COMPUTER
CIRCUITRY

Table 1 get the following sum of products expression P3: P3 =


R1’~~T;x3+xl’X2’x3+f~’~~7i~3$X~’
X1 x2 x3 From such an expression, weget the gatelevel de-
..................................................... .........,....................
p3

scription shown in Fig. 8. Note that, for each term,we


0 0 0 0
0 0 1 1 have an and gate and thereis one or gate that is driven
0 1 0 1 by all the and gates.
0 1 1 0
1 0 0 1
1 0 1 0 Programmable Circuits
1 1 0 0 The gate leveldescription is a description of the circuit
1 1 1 1
that isusedfor fabricating it. Prior to fabrication,
the physical location on a silicon wafer of the devices
(like FETs, resistors, etc.) has to be determined (place-
Boolean functions are defined using truth tables. Such ment). Thisisfollowedby determining how the
a table definesthe value of the function for each combi- devices are tobe interconnected (routing). For circuits
nation of values of the input. The truth table of the using a small number of devices, placement and rout-
parity function of three variables is shown in Table 1. ing canbedone manually.For larger circuits, the
Circuitswithmultiple outputs can alsobespecified entire process has to be automated. Many of the steps
using truth tables. of placement and routing can easily be automated if
Fig. 8 is a description of a circuit for the parity the circuit topology is regular. Moreover, the integra-
function. Such a pictorial representation is a gate level tion process becomes simpler and more cost effective
description of the circuit. A simple, but not necessarily if a variety of circuits can be physically implemented
efficient, way to derive a gate level description from a by minor variations (or programming) of a “master
truth table is as follows. An input variable Xi or its piece.” These factors have led to the evolution of a
complement (Le. negation) X i is known as an input number of design styles like programmable logic arrays
literal. A conjunction of literals is known as a term. (PLAs), Weinburger arrays, gate matrix arrays, etc.
Corresponding to each row of a truth table, we have a Such design styles are being used extensively. To gain
term. For example, for row 1 of Table 1, we have the insight into these styles, we will have a brief look at
term X I . X 2 . X3 (“.” denotes the “and” operation). nMOS PLAs.
The term correspondingto a row of a truth table for a Let X I , X Z , X3 be input variables.The nor expres-
Boolean function f is a one-term for f if and only if the
value of f for that input combination is 1. A zero-term
+ +
sion (X1 x2 X,) can be implemented logically as
shown in Fig. 9a. To implement this expression
issimilarlydefined.FromTable 1, . x2 X3 is a
physically, if there are n input variables, we have a
zero-term and X1 . x2 . X3 is a one-term for P3.
physical row of 2 n nFETs. For everyinput variable Xi,
Let a l , . . . , at be the one-terms of a function f. Then, there exist two nFETs, one driven by Xi and the other
+ + +
al a2 . . at (“+” denotes the or operation) is a driven by xi. An example is shown in Fig. 9b. For this
sum-of-products expressionfor f. FromTable 1 we row of nFETs, some of the “links” are “broken.” This

x2
-

x3

T
Figure 8. Example of sum-of-product circuits.
COMPUTER
CIRCUITRY 345

Links to be broken

I-l-l

Figure 9. Programmable NOR gate.

effectively removes some of the FETs from the circuit, The inputs of the FETs in a row of the OR-PLANE are
resulting in the desired gate shown in Fig. 9c. This is driven by the output of the same nor gate of the AND-
the basic idea used in a PLA. PLANE. Given such an array of FETs, a set of sum of
products expressions is implemented by “selectively
A PLA is shown in Fig. 10. It consists of two parts, an disconnecting” some of the FETs from the circuit. For
AND-PLANE and an OR-PLANE. Each of these two example, to implement the followingBoolean func-
parts is a two-dimensional array of FETs. Every rowof tions, the FETs to be disconnected are shown by x ” I‘

the AND-PLANEis arranged to form a potential nor in Fig. 10.


gate. AU FETs in a column of the AND-PLANE are
driven by the same input literal. Every column of the fl = XIX2 +m
OR-PLANE is arranged to form a potential nor gate. + XlX2

+
f2 = XlX2

OR
AND-PLANE

PLANE %

- - -

><

1._1-1?+-
: I

- -
x1 xi x2 x2 fi f2

Figure 10. Example of a programmable logic array.


346 COMPUTER
CIRCUITRY

Figure 11. Example of a state diagram.

Sequential Circuits Another way to express this behavior of MOD-3 is with


the help of the state diagram of Fig. l l b . Since we
We will use a simple example to illustrate how sequen- need to know if rj is 0, 1, or 2 , we say that the circuit,
tial circuits are designed. Consider a circuit with one at the end of I;, is in state So if rj is 0 , SIif rj is 1, and
input X and one output Y, defined as follows.Let ni be Sz if rj is 2. The arrow marked a / b , which starts at Si
the number of intervals, including and up to I;, during and ends at S t , is to be interpreted as follows: if,at the
which Xhad the value 1, At the end of Zj, Yis set to 1if end of the interval Zj-1, rj-1 is i and during interval 4
and only if n; is divisible by 3. We call such a circuit the value of X is a, then, at the end of interval Z;, the
MOD-3. value of Y should be b and Y , should be t.
In order to compute thevalue of Y during the interval While designinga circuit from a state diagram, we have
Ij, MOD-3 must remember some characteristics of the to make sure that thecircuit "remembers" the state of
pattern of Os and 1s during the intervals 1 to Ii-1. It is the system. In order to do that, the states are encoded
enough for the circuit to remember the value of using two statevariables ao, a1 asfollows: SO =
r;-1 = n;-1 mod 3, which is nothing but the remainder (a0 = 0 , a1 = 0); SI= (a0 = 0 , a1 = 1); and& = (a0 = 1,
left after dividing nj-1 by 3. Also note that if X,during a1 = 0). The circuit now needs to remember thevalues of
+
I;, equals 1, then r; = (rj-1 1) mod 3; and if r; equals the state variables. In other words, the circuit should
0, then Y is to be set to 1. Therefore, at the end of any be capable of remembering two bits of information.
interval, we need to remember if the remainder was 0, This leads to the question: how exactly does a circuit
1, or 2. remember two or more bits of information?

ICLOCK 1 WR Q

\ n

-.--

Figure 12. Structure of a flip-flop


COMPUTER CIRCUITRY 347

Table 2 JK flip-flops,etc. (see SEQUENTIAL MACHINES).Fig. 12c


depicts a slightvariation of a D flip-flop that uses
si st
nFETs. In this case, in order to modify the values of Q
input
.......................................................... output
X a0 a1 Do Dl Y and Q, both CLOCK and WR have to be 1.
........................................................................................................
0 0 0
To continue our discussion of the design process, we
0 0 1
1 0 0 0 1 0 derive the state table from the state diagram. The state
0 0 1 0 1 0 table for our example is shown in Table 2. For each
1 0 1 1 0 0 edge in the state diagram, we have a row in the state
0 1 0 1 0 0 table. The pair (ao, a l ) defines the state Si of Fig. l l c ;
1 1 0 0 0 1
0 1
(DO,01)defines the state .ST; X defines the value a; and
dc 1 dc dc
1 1 1 dcdc dc Ydefines the value b. Note that DO,Dl will become the
new value of ao, a1 at the start of the next interval. In
dc-don’t care. order for that to happen, the circuit should be of the
form shown inFig. 13a, where C must satisfy the
conditions of the state table (Table 2). Since there is no
As theydidforlogic gates, circuit designers have
state S3, the case a0 = a1 = 1 cannot arise; hence
devised circuit elements called flip-flops. A flip-flop
there are “don’t care”values in these rows. The com-
can store (i.e. remember) one bit of information. The
plete circuit is shown in Fig. 13b. Note that Table 2 is
structure of a flip-flop isshown in Fig. 12aand we will
like a truth table for the combinational block C of
use the symbol of Fig. 12bto represent such a flip-flop.
Fig. 13a. Thisis a simple, but notnecessarily an
Here Q is the output of the flip-flop and Q is another
efficient way to implement sequential circuits.
output, such that the value of Q is always the com-
plement of the value of Q. D is the input to the flip-
flop. When CLOCK is 1 and D is 1, Q is set to 1. When
Conclusion
D is 0 and CLOCKis 1, then Q is set to 0. When This article has provided an overview of the building
CLOCK is 0 , the value of Q or Q cannot change. What blocks of computer circuits. Since the 1960s, these
we have just described is a D flip-flop. There are a circuits havebeenincreasingly compact and dense
number of other classes of flip-flops like SR flip-flop, semiconductor devices,initiallybuilt out of discrete

I ‘
(b)
Figure 13. Implementation of the state diagram of Fig. 11.
348 COMPUTER
CONFERENCINC

transistors, and then as integrated circuits that packed tems (CS), Computer ConferencingSystems(CCS),
many transistors, and hence, many gates, on a silicon Computer-MediatedCommunication Systems (CMCS),
chip. They have evolved from small-scale integration Computer-Supported Cooperative Work (CSCW), Co-
(SSI) to medium and large scale integration (MSI and ordination Systems (CS), Electronic Meeting Systems
LSI) to the current very large scale integration (VLSI) (EMS), Group Decision Support Systems (GDSS),
that can put several million transistors on a silicon Groupware, Networking, Teamware, and Teleconfer-
wafer about 1 cm in diameter. Such microcomputer encing. All these types of systems have one thing in
chips can hold a powerful processor and fast cache common: they are designed with the objective of using
memory (4.v.). the computer to aid inthe process of facilitating group
communications. The most appropriate inclusive term
Sometimes computers are characterized by their gen-
is Computer-MediatedCommunications (CMC),a term
eration: the first generation used vacuum tubes; the
that includes all the others and claimeddifferences
second, discrete transistors, and the third, integrated
among them.
circuits. These circuits, introduced at the end of the
1960s, have evolved greatly, but more than 30 years With the emergence of theInternet (4.v.) and the
later, they still providethe fundamental material out of World Wide Web ( q . v . ) , there are currently hundreds
which computers are built. of group systemscommerciallyavailable: however,
very fewof them reflect the ability of the user or group
Bibliography to tailor the process. The demonstration of that con-
1980. Mead, C.,and Conway, L. Introduction to VLSI Systems.
cept is still largely limited to R&D systems.
Reading, MA: Addison-Wesley.
Thediscussion or conference part of a CMC sys-
1993. Weste, N., and Eshraghian, K. Principles of CMOS VLSI
Design, 2nd Ed. Reading, MA: Addison-Wesley. tem may be represented as a specific hypertext (4.v.)
1998. Unger,S. H. The Essence of Logic Circuits, 2nd Ed. Upper structure. Most current systems can be described as
Saddle River, NJ: Prentice Hall. in Table1, which shows the relationships (links)among
1998. Wolf, W. Modern VLSI Design-Systems on Silicon,
2nd Ed. Upper Saddle River, NJ: Prentice Hall.
the objects (nodes) that characterize the system.
2000. Wakerly, J. F. Digital Design: Principles and Practice, Objects in CMC systems include both the elements of
3rd Ed. Upper Saddle River, NJ: Prentice Hall. discourse and the individuals or members. The individ-
Sreejit Chakravarty uals are objects which are linked to content objects by
privileges associated with their human roles. There may
be significant additional functionality with respect to
the different roles and associated software powers a
person can have in a given conference. However, the
basic discourse structure is usually some combination
of temporal occurrence, comment/reply hierarchies,
and keyword association of comments.
An example that goesbeyond these basics is the
“Virtual Classroom” (Hiltz, 1994), where a number of
special communication structures exist inaddition to a
basic computer conferencing structure. For exam-
ple, an instructor can ask a discussion question but no
student can see the other discussion answers until he
or she has answered. What this example indicates is

Table 1. Common CMC object and relationship structures.

Comment Reply Person

Comment Laterlearlier In response to Author of/


than temporal editor of/
reader of
Additional/ Author of/
alternative editor of/
reader of
Person Member of
same
conference

You might also like