Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
Nov 2000
Computer Hardware
Table of Contents
MICROCOMPUTER HARDWARE ............................................................................................................... 2
Computer Basics ....................................................................................................................................... 2
Input → Process → Output .................................................................................................................... 2
Bits; Bytes; Binary and Hexadecimal ........................................................................................................ 2
Power of 2 .............................................................................................................................................. 3
THE MICROCOMPUTER ............................................................................................................................. 4
THE SYSTEM BOARD .............................................................................................................................. 4
The System Board Buses ...................................................................................................................... 4
The Local Bus ........................................................................................................................................ 5
The Data Bus ......................................................................................................................................... 5
The Address Bus ................................................................................................................................... 5
The Expansion Bus ................................................................................................................................ 5
Direct Memory Access ........................................................................................................................... 6
The Central Processing Unit (CPU) .......................................................................................................... 6
Clock Speed ........................................................................................................................................... 6
Word Size .............................................................................................................................................. 6
Instruction Set ........................................................................................................................................ 6
Intel Microprocessors ............................................................................................................................. 7
Other Microprocessors........................................................................................................................... 9
Interrupts and I/O Ports ........................................................................................................................... 10
Common Interrupts .............................................................................................................................. 10
Input/Output Ports ................................................................................................................................ 11
Expansion Slots ....................................................................................................................................... 12
ISA Bus ................................................................................................................................................ 13
Micro Channel Bus ............................................................................................................................... 13
EISA Bus .............................................................................................................................................. 13
VESA Bus ............................................................................................................................................ 13
PCI Bus ................................................................................................................................................ 14
NuBus .................................................................................................................................................. 14
Memory .................................................................................................................................................... 14
Memory Types ..................................................................................................................................... 15
Single In-Line Memory Modules .......................................................................................................... 16
Memory Usage ..................................................................................................................................... 16
Storage Systems ..................................................................................................................................... 17
MICROCOMPUTER HARDWARE
A network administrator must have a solid background in the fundamentals of microcomputer hardware
and software components.
1.0 Computer Basics
A microcomputer operates on the same basic internal principles as its larger siblings, mini-computers and
mainframe computers. All computers, regardless of their size or capacity, need to perform input,
processing, storage and output operations.
Input
Process
Output
The input units include such devices as keyboards, scanners, and the mouse, which convert data into
electronic on/off signals that computer circuits can transmit and store. The processing hardware houses
the computer circuits that are responsible for processing the data.
As bits of data arrive at the system unit, they are stored in electronic circuits called memory buffers. A
memory buffer is like a receiving room where the data bits wait until the computer is ready to process
them.
The central processing unit (CPU), also called a microprocessor on a microcomputer, contains the
electronic circuits that interpret the program instructions stored in memory and then perform the specified
operations on the data, including input, arithmetic functions, decision making, and output.
A major component of the processing hardware is the primary memory, also called random-access
memory or RAM. The primary memory stores program instructions and data in a form that is directly
accessible to the CPU. Data storage in RAM is temporary because the RAM contents are lost when the
computer is turned off or rebooted. In order to store software and data for later use, it is necessary to
record the information from RAM into a storage system. The storage of the computer consists of magnetic
disks that allow the computer system to store and access software and data files.
Some data such as that needed to initially start the computer is stored in read only-memory or ROM.
This data can be read into the computer’s memory for processing, but the computer cannot write
information back.
Output units consist primarily of video monitors, printers, and storage devices. The video card of your
computer plays a very important role in how fast your computer performs and what software you can run
by providing compatibility with several video standards in use today.
The hardware components found in all microcomputers can be divided into six major categories:

Microprocessors

Memory

System boards

Storage systems

Video monitors

Power systems
2.0 Bits; Bytes; Binary and Hexadecimal
The network administrator needs to be familiar with the basics of the binary and hexadecimal number
systems in order to better understand and configure computer hardware. Binary refers to a number
system with only two values: 0 and 1 (mathematicians refer to this as Base 2. It’s like a light switch –
either the switch is off (0 – the light is off) or the switch is on (1-the light is lit). The 0 and 1 are called
binary digits and all data and instructions in a computer are stored as binary digits. A single binary digit is
Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
Nov 2000
called a bit (short for binary digit). Since one bit by itself cannot convey much information, eight bits are
grouped together for form a byte.
27
128
0
Power of 2
Value
Bit
26
64
1
25
32
1
24
16
0
23
8
0
22
4
1
21
2
0
20
1
1
The maximum value that can be stored in a byte (11111111) is equal to 255 (125+64+32+16+8+4+2+1).
The number of different values that can be stored in a byte is 256: the numbers from 1 (00000001) to 255
(11111111) plus 0 (00000000).
To represent characters, a coding system is used so that a byte represents one character of data. The
most common coding system used on microcomputers is the American Standard Code for Information
Interchange (ASCII) developed by the American National Standards Institute (ANSI). Listed below is
a partial chart.
Character
Dec
Hex
Character
Dec
Hex
Character
Dec
Hex
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
41
42
43
44
45
46
47
48
49
4A
4B
4C
4D
4E
4F
50
51
R
S
T
U
V
W
X
Y
Z
a
b
c
d
e
f
g
h
82
83
84
85
86
87
88
89
90
97
98
99
100
101
102
103
104
52
53
54
55
56
57
58
59
5A
61
62
63
64
65
66
67
68
I
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
69
6A
6B
6C
6D
6E
6F
70
71
72
73
74
75
76
77
78
79
7A
Another solution is to use the hexadecimal number system to make binary numbers and ASCII codes
more manageable. The hexadecimal number system is based on powers of 16 (mathematicians refer to
this as Base 16) and uses 16 different values. You are already familiar with the symbols for the first 10 of
these values (0-9); the other six values are represented by the first six letters of the alphabet (a-F), where
the letter A represents ten, B represents eleven, C twelve, D thirteen, E fourteen, and F fifteen. The
advantage of using the hexadecimal system is that when you divide a byte into two 4-bit sections, called
nybbles, each nybble has a range from zero (0000) through 15 (1111). This means that each nibble can
easily be represented by one hexadecimal digit (0-F).
Nybble
Hex
Nybble
Hex
0000
0001
0010
0011
0
1
2
3
1010
1011
1100
1101
A
B
C
D
Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
0100
0101
0110
0111
1000
1001
4
5
6
7
8
9
1110
1111
Nov 2000
E
F
Hexadecimal numbers are often written with the letter “h” after the number to indicate that
characters are hexadecimal numbers. For example, the hexadecimal number 4E would be written
as 4Eh.
Character
Decimal
Hexadecimal
Binary
N
78
4E
01001110
O
79
4F
01001111
V
86
56
01010110
E
69
45
01000101
L
76
4C
01001100
L
76
4C
01001100
3.0 The Microcomputer
The microcomputer is commonly called a personal computer or PC. The people who use them to get the
job done are referred to as end users or just users. PCs used by users are often referred to as PC
workstations or just workstations.
3.1 System Board
The system board (also called the motherboard) is the most important component of a microcomputer
because it links all the individual system components. The design of the system board directly affects the
performance of a computer system. In this section we’ll discuss how the system board components are
linked by using buses and the following sections will discuss the components themselves.
3.2 The System Board Buses
The system board circuits that connect components are called buses. Buses are the pathways for
electronic communication between parts of the computer. They can be visualized as a set of wires
running together (in “parallel”) from component to component. In actuality, circuits etched into the system
board are instead of wires, but the idea is the same. Buses vary in size in size (number of “wires”) and
speed, and are usually referred to by names that reflect their purpose. In this section we’ll discuss the
data bus, the address bus, the expansion bus, and the local bus. A typical bus structure is shown in
Figure 2-7.
Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
Nov 2000
CPU
Data bus and
Address bus
Local bus
L2 cache
controller
Expansion
bus controller
Expansion bus
L2 cache
memory
Expansion
slots
Memory
3.2.1 The Local Bus
The system board circuits that connect the CPU to memory and other system board components are
referred to as the local bus of the computer. The local bus is closely associated with the functions of the
microprocessor chip and is designed to support the data and address bus of a specific microprocessor.
3.2.2 The Data Bus
As shown in Figure 2-8, the data bus is the highway that transfer data bits to and from the
microprocessor registers where the microprocessor stores the data it uses. Just as the number of lanes
on a highway determines the amount of traffic that can flow, the size of the data bus determines the
number of bits that can be transferred into the microprocessor at one time. Transferring information
between the memory unit and the processor is called fetching. For example, an 80386SX
microprocessor, which has a 16-bit data bus, requires two fetches from memory in order to load one
register. An 80386DX microprocessor, which has a 32-bit data bus, can load a register with just one fetch.
3.2.3 The Address Bus
Just as each box in a post office is given a unique number to identify it, each byte in the computer’s
memory is identified by a binary number called an address. The microprocessor uses an address to
identify the memory byte to or from which it is transferring data. The address bus carries the address of
the memory byte from the microprocessor to the memory unit. When the memory unit receives the
address along with a signal to read, it responds by placing the contents of that memory byte on the data
bus. The number of bits in the address bus determines the maximum amount of memory the
microprocessor can access directly. If an address bus consists of only two wires, for example, it maximum
binary number is 11. A computer with this address bus is limited to a maximum of four byte addresses:
00,01,10 and 11. The 24-bit address bus on the 80286 and 80386SX can access up to 16MB; and the 32bit address bus on the 80386DX, 80486, Pentium, and Pentium Pro can access up to 4 GB.
3.2.4 The Expansion Bus
Some microcomputer components, such as video display controllers and network interface controllers,
are not usually built into the system board. Instead these components are built on separate cards, called
expansion cards, that can be attached to system board by inserting them into an expansion slot on the
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
system board. The expansion bus is the bus that connects the expansion card slots to the components
of the system board. The expansion bus connection is made through a specialized chip set on the system
board that controls the operation of the expansion bus and the expansion cards attached to it. There are
several types of expansion buses, which are discussed in the section on expansion slots later in this
chapter.
3.3 Direct Memory Access
A direct memory access (DMA) channel is part of the local bus that is used to automate the transfer of
data between the computer’s memory and external devices such as disk drives and NIC’s. DMA channels
are assigned to specific devices. This means that when you are configuring a device such as a NIC, you
need to be sure to assign an unused DMA channel number.
Because local buses are designed to support a specific microprocessor, it is impossible to
upgrade the microprocessor chip on the system board in microcomputers. For example, if you
want to upgrade a system from an Intel 80386SX to an Intel 80486 processor, you need to
replace the system board. Newer system boards are designed for microprocessor upgrading.
However, these system boards still have the original local bus built into the, which will limit the
performance of the upgraded system. For example, Intel now sells a Pentium microprocessor to
replace 80486 processors on properly designed system boards. But these boards were designed
with a 32-bit data bus, whereas the Pentium is designed to work with a 64-bit data bus.
3.4 The Central Processing Unit (CPU)
The CPU, also referred to as a microprocessor, is the brain of the microcomputer system. Built into the
silicon of modern microprocessors are more than 1 million transistors that make up circuits to interpret
and control the execution of program instructions and perform arithmetic and logical operations.
In order to compare microprocessors, you first need to understand the parameters that determine the
performance and functionality of microprocessors: clock speed, word size, instruction set, data bus set,
and address bus size.
3.4.1 Clock Speed
If the microprocessor is the brain of the computer, then the clock is the heartbeat of the system unit, and
its beat synchronize all the operations of the internal components. The microprocessors clock is used to
provide precisely timed signal pulses called cycles. Each clock cycle consists of an electronic pulse that
is transmitted to each component of the system unit to trigger and synchronize processing within the
computer system. Each clock pulse received by the microcomputer causes its circuits to perform part or
all of an instruction.
Clock speed is measured in millions of cycles per second, called megahertz (MHz). Wait states are
clock cycles during which the processor does not perform any operations; they are necessary to slow
down high-speed processor chips and allow them to work with slower devices. In general, higher clock
rates mean faster processing speeds. The processing speed, when combined with the speed of a
computer’s disk storage and video card, determines its throughput performance.
3.4.2 Word Size
A microprocessor chip holds instructions and data temporarily in storage areas called registers. Each
processor chip has several registers for various purposes. A microprocessor’s word size is the number of
bits each register can hold. A larger word size enables a microprocessor to work on more data per cycle
clock. Older processor chips, such as Intel’s 8088 and 80286, have 16-bit registers. Newer processors,
such as the Pentium and Pentium Pro, have 32-bit registers.
3.4.3 Instruction Set
The instruction set, also called the machine language, is the group of commands that the
microprocessor chip has been designed to process. All software must be converted to the
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
microprocessor’s machine language before it can run. This is often accomplished with the aid of a special
program called a compiler, which converts English-like commands to the binary language of the
processor chips. DOS machine language programs use the filename extension .COM or .EXE. On a
NetWare server, machine language programs have the extension .NLM.
A machine language program can be run only on the processor for which it was designed. Intel and
Motorola processors, for example, have very different instruction sets, making it impossible for the
Motorola chip to run a machine language program written for an Intel chip. The NetWare 3.12 and 4.1
operating systems were written for the instruction set of an Intel 80386 microprocessor and therefore
cannot be run on earlier Intel processors or on Apple computers, which all use Motorola processor chips.
Computers with Intel and most Motorola chips are classified as complex instruction set computers
(CISC) because their instructions have a wide range of formats and because one instruction can require
many clock cycles. The resultant sped of the microprocessor is often expressed in millions of
instructions per second (MIPS).
Companies such as Cyrix and Advanced Micro Devices produce Intel-compatible processor
chips that are used in some IBM-compatible systems.
To maximize speed, many engineering workstations running CAD applications (such as SUN
workstations) are based on processors called reduced instruction set computers (RISC). RISC
processors are very fast and efficient because their instructions are all the same length, and each
instruction performs a very specific process. The disadvantage of RISC processors is that the software
development is more complex and requires sophisticated compilers to convert programs to the machine
language format.
The biggest advantage of RISC- over CISC-based computers is the increased speed of floatingpoint math calculations. This speed advantage is the reason RISC-based processors are often
used in workstations running engineering, CAD, or scientific applications.
A math coprocessor is an extension of a chip’s basic instruction set that allows the
microprocessor to perform more complex arithmetic operations such as square root and trigonometric
functions. Math coprocessors are built into 80486DX, Pentium, and Pentium Pro processors. Math
coprocessors can greatly increase the speed of spreadsheet programs and applications used for
engineering and CAD, which typically perform many square root and trigonometric calculations.
3.4.4 Intel Microprocessors
The Intel family of microprocessor chips is probably the best known because of the wide acceptance of
IBM-compatible computers based on this processor design. You need to know the basic features of Intel
processor chips and how they relate to the capabilities of the computer systems you are likely to
encounter.
A. Intel history.
The Intel 8088 processor chip, included in the IBM PC, which was introduced in 1981, started the IBM
PC-compatible industry. The 8088 has a 4-MHz clock speed, a 20-bit address bus that can access up to
1 MB of RAM, and a 16-bit register system. It allowed designers to create everything a PC user would
need in the then-foreseeable future. Running the instruction set that comes with the original Intel 8088
microprocessor is referred to as operating in real mode. Real mode instructions use 16-bit data registers
and can directly access only 1 million bytes of memory. The DOS operating system—and the thousands
of DOS software applications still in use—were designed specifically for the original 8088 microprocessor.
This means that even of you have the latest and fastest Intel processor chip, your workstation computer is
limited to 640 KB of RAM and 16-bit instructions when it runs DOS-based software in real mode.
The need for more powerful processor chips led to the development of the 80286 processor, which
provided up to seven times the performance of the 8088 processor while providing compatibility for real
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
mode programs. Included in the IBM AT, which was introduced in 1984, the 80286 microprocessor added
three new capabilities:

The address bus was increased to 24 bits to allow for up to 16 MB of system RAM.

The clock speed was increased to between 8 and 20 MHz.

It can switch between real mode and protected mode.
Real mode operation allows the microprocessor to act like a very fast 8088; protected mode allows it to
run multiple programs more reliably by preventing one program from affecting the operation of another.
Although the 80386 is capable of 32-bit processing (because of its 32-bit internal registers and data
paths), most PC add-on boards and software were designed for older 8- or 16-bit processors and are
therefore unable to make optimum use of the 80386’s 32-bit capability. Because of this, Intel introduced a
less expensive version of the 80386 chip, the 80386Sx, and named the original chip 80386DX. The main
difference between the SX and DX versions is that the 80386SX has the same 16-bit external data bus
and 24-bit address bus as the 80286 processor. It is therefore limited to a maximum of 16 MB of system
RAM and generally runs at a slower speed. In addition to providing 32-bit processing, the 80386DX
incorporated the following new features:

It can directly access up to 4 GB of system RAM.

It has the ability to switch between real and protected modes without the need to reset the processor.

The use of virtual memory enables the 80386 processor to use hard disk space to simulate a large
amount of internal RAM. Although the use of virtual memory slows down the computer’s throughput, it
also allows you to run large programs that would not otherwise fit in the existing RAM.

The addition of virtual real mode enables multiple real mode programs to run simultaneously.

It can run at a variety of clock speeds ranging from 16 to 40 MHz.
B. Current Intel microprocessors
Although you will still encounter workstations and older NetWare servers based on earlier versions of the
Intel processors, these early members of the Intel microprocessor family are no longer being produced,
and all new Intel-based workstations and NetWare servers you install will be based on either 80486,
Pentium, or Pentium Pro microprocessors. The 80486, Pentium, and Pentium Pro microprocessors all
provide compatibility with software written for earlier processors and the computing power needed for
high-speed graphics-based software.
80486: The 80486 is basically a supercharged version of the 80386 chip. It incorporates more than 1
million transistor components.

Higher clock speeds ranging from 33 to 100 MHz.

8-KB high-speed memory cache (L-1 cache) allows the processor to access commonly used memory
locations without going through the slower external data bus

Built in math co-processor.
There are two main versions of the 80486 chip. The 80486DX contain all the features just described. The
80486SX is a less expensive version of the 80486DX chip and does not include the math coprocessor.
(Interestingly, the 80486SX actually has the math co-processor , but it has been disabled to allow Intel to
market the chip at a lower price). A modified 80486DX chip, called the 80486DX2, uses a clock-doubling
technique that doubles the processing throughput of the chip. An 80486DX at 33 MHz becomes an
80486DX2 at 66 MHz. Yet another version of the 80486 chip, called the 80486DX4, can triple (not
quadruple) clock speeds. Thus an 80486DX at 33 MHz becomes an 80486DX4 at 100 MHz.
Pentium: Intel’s Pentium chip represents a major leap ahead of earlier Intel chips by incorporating two
80486-type microprocessors on a single chip that can process two instructions simultaneously. The
Pentium chip can operate at least 200 MHz with 64 bit registers and more than 3 million transistors. The
Pentium math co-processor has been re-designed to achieve a 300% improvement in geometric
computations over 80486 chips, allowing graphics-intensive applications to work at much faster speeds
Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
Nov 2000
Pentium MMX: Intel has usually tried to move users to each new microprocessor as quickly as possible.
With the Pentium, however, Intel has continued developing the chip to extend its useful life span. The
result is the Pentium MMX microprocessor, a version of the Pentium released in late 1996 that has been
enhanced to speed up multimedia functions. For example, one enhancement allows the Pentium MMX to
perform an ADD instruction on eight bytes of data in the time formerly required for just one byte of data.
This will result in significantly faster multimedia software processing but will not have as much of an effect
on standard spreadsheet or word processing software. The need for multimedia processing is increasing
as more software products are delivered on CD-ROM and as Internet content includes more audio, video,
and 3-F components. Speeds may include a 200 MHz. version, and these enhancements will keep the
Pentium chip a viable choice for PC workstations
Pentium Pro: The Intel Pentium Pro processor was designed for optimal performance with 32 bit software
while maintaining compatibility with previous Intel processors. It still uses 32-bit word, but its design
includes 5.5 million transistors in the chip, and it operates at speeds up to 200MHz. The Pentium Pro chip
also includes a 256-KB L2 cache to accelerate data input. One of the Pentium Pro’s main features is
called Dynamic Execution. Dynamic Execution combines three processing techniques: multiple branch
prediction, dataflow analysis, speculative execution. In multiple branch prediction, the Pentium Pro looks
several steps ahead in the program and predicts which steeps will be processed next. Dataflow analysis
is then used to set up an optimized schedule for performing the program steps, which leads to speculative
execution which performs the steps as scheduled by the dataflow analysis.
Word
Size
Data
Bus
Address
Bus
Max. Clock
Speed (MHz)
8088
16
8
20
10
No
33
80286
16
16
24
20
No
3
80386SX
32
16
24
33
No
5
80386DX
32
32
32
33
No
11
80486SX
32
32
32
33
No
41
80486DX
32
323
32
100 (DX4)
Yes
80
Pentium
32
64
32
200
Yes
>100
Pentium MMX
32
64
32
200
Yes
>100
Pentium Pro
32
64
+(8
ECC)
36
200
Yes
>100
Microprocessor
Math CoProcessor?
Instructions
(MIPS)
The Pentium Pro runs best with a true 32 bit operating system such as Windows NT
Workstation or O/S 2 Performance suffers if Windows 95 is used as the operating system. A
Pentium system running Windows 95 can actually outperform the Pentium Pro system when
using typical user applications such as word processing.
3.4.5 Other Microprocessors
Apple Macintosh computers are based on the Motorola 68000 line of microprocessors; Apple PowerMac
computers are based on the PowerPC microprocessor.
A. Motorola Processors.
The Motorola MC68000 chip was used on earlier Macintosh computers and is similar to the Intel 8088
microprocessor. The MC68000 chip was used in various Macintosh II machines and is comparable to the
Intel 80386Dx or 80486SX processors The MC68030 represents the first attempt to do multiprocessing
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
within a microprocessor chip and offers an internal memory cache, a math coprocessor, and clock speeds
of up to 40 MHz. The MC68030 is used in the Macintosh SE and various Macintosh II computers. The
MC68040 is comparable to the Pentium in that it offers 64-bit communication and higher clock speeds.
PowerPC. Intel’s microprocessors are Complex Instruction Set Computers (CISC) processors. The
PowerPC processor, developed cooperatively by Motorola, IBM, and Apple, is a member of the
Reduced Instruction Set Computer (RISC) family of microprocessors. Because it is based on the RISC
architecture, the PowerPC achieves much higher processing speeds than the existing CISC-type
Motorola processors in the 68X family and competes well with Intel’s microprocessors. Both Apple and
IBM are using special versions of PowerPC processors to produce high speed computers that can
support both existing software and newer PowerPC specific software. They do this with special emulator
software that resides in ROM and automatically translates 68000-based instructions into PowerPC
instructions.
Microsoft has a PowerPC version of the powerful Windows NT operating system that is intended to make
the PowerPC chips more widely used on networked workstations. Newer versions of the PowerPC chip,
such as the 603, 604, and 620, will provide lower power usage, higher clock speeds, and a larger word
size. The PowerPC 620 is scheduled to have 64-bit registers along with a 64-bit external data bus to
provide extremely high-speed processing required by such applications as speech recognition.
3.5 Interrupts and I/O Ports
In order to provide for input from and output to the computer system, you need to be able to attach such
devices as keyboards, printers, monitors, network cards, and the mouse. These devices are commonly
known as peripherals because they are added on to the system board. Each peripheral device attached
to the system board—from the hard disk to the keyboard—must be controlled and monitored by the
microprocessor. This monitoring is accomplished by interrupts and input /output I/O ports. You need to
know how to configure network adapter cards correctly.
3.5.1 Common Interrupts
An interrupt request (IRQ) is a signal that a device or controller card sends to the processor to inform it
that the device or controller card needs attention.
Each device in a computer system needs to have its own unique interrupt so that the processor will not
misinterpret the source of the interrupt signal, two devices using the same interrupt number in a computer
system cannot interact correctly with the processor, and your system performance will be sporadic at
best.
Because of the limited number of system interrupt numbers, it is impossible to assign unique numbers to
every category of computer peripheral. There are some general usage guidelines for system interrupts,
however. Figure 2-10 shows interrupt numbers used on several of the most common system devices.
Each manufacturer allows you to adjust the interrupt settings of its peripheral device, so you can choose
an interrupt setting that does not conflict with other system devices.
Local area network (LAN)
Revision 0
Topic 2 – Computer Hardware
IRQ
CPU
Nov 2000
0 — System Clock
8 — Clock (time of day)
1 — Keyboard
9 — Unavailable
2 — Available
10-12 — Available
3 — COM2:/COM4:
13 — Math coprocessor
4 — COM1:/COM3:
14 — AT hard disk
5 — LPT2 (PC Hard Disk)
6 — Floppy disk drive
7 — LPT1:
3.6 Input/Output Ports
An input/output (I/O) port is a memory location that the processor uses to send control commands to a
peripheral device and read back status information. To communicate with each device separately, each
peripheral attached to the computer system needs a unique I/O port address range. Figure 2-11 lists I/O
port addresses for several common peripherals. To avoid conflicts with other devices in a computer, each
peripheral controller card manufacturer provides a number of different I/O port addresses options. The
network administrator’s job includes assigning unique I/O port settings for network cards.
Device
Interrupt
I/O Address
COM1
COM2
LPT1
LPT2
LPT3
IDE disk controller
XT disk controller
IBM token ring (primary)
IBM token ring (secondary)
4
3
7 (if used)
5 (if used)
none
14
5
2
3
3F8-3FF
2F8-2FF
3BC-3BE
378-37A
278-27A
IFO-IF8 170-177
320-32F
A20-A23
A24-A27
3.6.1 Parallel Ports.
The parallel port connects the computer to a parallel cable, which transfers data from the computer to a
peripheral device eight bits at a time on parallel wires. The parallel port is commonly referred to as the
printer port because almost all printers use a standard parallel port interface. This makes it easy to plug
almost any printer into the parallel port of a computer. The use of the parallel port for printers was
standardized by the Centronics printer company, and the parallel cable attaches to a 25-pin connector on
the back of the computer and a larger 36-pin card edge connector on the printer. As a result of the early
popularity of the Centronics standard, all IBM-compatible computer and printer manufacturers today
include the Centronics parallel port on their systems.
3.6.2 Serial Ports.
The serial port on IBM-compatible computers sends only one bit of data at a time. One advantage of the
serial port is its ability to send information between devices over long distances by using only few wires in
a twisted-pair cable. Timing is very important in serial communications in order for the receiving device to
correctly interpret the signals coming from the transmitter.
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
The speed of the serial signal is the baud rate. The baud rate is one, indicating one signal change per
second. When a modem sends analog frequencies over the telephone, the bit pattern is represented by a
change in frequency, which allows several bits to be transmitted for each baud. The rate of bits per
second is often much faster than the baud rate.
Serial communication can be either synchronous or asynchronous. Synchronous communication,
commonly used with LAN cards to send packets consisting of 1,500 or more bytes between computers,
takes place at very high speeds ranging from 4 to more than 100 Mbps (megabits per second).
Asynchronous communication is much simpler, sending only one character at a time. It is often used by
modems to transmit information between microcomputers, or between microcomputer and on-line
information service provider such as America Online, CompuServe or Prodigy.
In asynchronous communication, each character is transmitted separately and is encapsulate with a start
and stop bit and optional parity bit for error checking. See figure 2-12. Parity works well for single-bit
errors, but its reliability falls off when more than one bit is changed.
Most system board manufacturers build one or more asynchronous ports into their IBM-compatible
systems. These ports are generally referred to as COM1 through COM4, have either 25-pin or 9-pin
connectors, and are located on the back of the computer.
Two types of serial connectors, known as RS232 connectors, were standardized by the Electronic
Industry Association (EIA) in the early days of computing. A special type of RS232 cable called a null
modem cable is used to connect two DTE computers without the use of a modem.
Figure 2-13 has certain wires crossed in order to allow the signal from the sending computer to go to the
correct connectors on the receiving computer.
3.7 Expansion Slots
The expansion slots connect add-in cards, such as video cards and NICs, to the system board. They are
connected to other components by the expansion bus. As the capabilities of microprocessors have
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
improved, the demands placed on the expansion bus have grown. Part of the change in the expansion
bus has been changes in the connectors used in the expansion slot.
3.7.1 ISA Bus
The industry standard architecture (ISA) bus was introduced in 1984 with the IBM AT computer. Its
expansion slots support 16-bit data and 24-bit address buses running at 8 MHz. It also contains a local
bus that can support up to 32-bit data and address paths at high clock speeds (such as 33 MHz) between
the microprocessor and memory. The main disadvantage of the ISA bus can be seen in graphics
applications that require high-speed video processing. ISA slots on the system board have 16-bit card
slots composed of two sockets placed together, one containing 31 pins and the other containing 18 pins.
3.7.2 Micro Channel Bus
The Micro Channel Bus architecture is owned by IBM and can support 32-bit expansion slots running at
high clock speeds (such as 33 MHz). It was developed for IBM’s PS/2 line, but lack of success with that
line and with licensing the architecture forced IBM to discontinue using Micro Channel in its PCs. IBM
now uses Intel’s PCI bus in its products. A major advantage of the Micro Channel architecture is that it
keeps card configuration information in CMOS (complimentary metal oxide semiconductor) memory
on the system board, allowing software setting of card options and configurations. Although Micro
Channel was used by IBM for its PS/2 line of computers, it never was adopted by the industry in general,
and IBM returned to using the ISA bus for many of its less-expensive models before finally switching to
the Intel PCI bus.
3.7.3 EISA Bus
When IBM introduced its proprietary Micro Channel bus, other PC manufacturers who wanted to sell
systems with the increased performance of IBM’s 32-bit bus slots were required to pay IBM royalties and
redesign their systems. Because the Micro Channel bus does not accept the older ISA cards, users must
purchase the more expensive Micro Channel cards even for slower devices such as modems and
printers. In reaction to this, a number of IBM-compatible computer manufacturers cooperated on the
design of an enhanced version of the ISA bus that would support 32-bit expansion cards and higher clock
speeds. The result is the extended industry standard architecture (EISA) bus, which supports 32-bit
data and address expansion slots that can support adapter cards at 8-MHz clock speeds. Because the
EISA bus is an extension of the ISA bus, it includes 16-bit expansion slots that accept other ISA cards. Of
course, ISA cards placed in these slots still use the limited ISA address and data bus sizes. For example,
a 32-bit disk interface card placed in an EISA bus slot provides better disk access times than a 16-bit card
in an ISA-compatible slot. A typical EISA system board includes two 32-bit slots and six 16-bit ISA slots.
3.7.4 Bus mastering
It is a technique built into EISA and Micro Channel bus systems, enables adapter cards to off-load such
tasks as moving information into memory in order to improve overall system performance. Bus mastering
is an important option to consider when selecting a NetWare server computer. Much of a NetWare
server’s processing involves moving information to and from memory. The use of bus mastering can
greatly improve the performance of a NetWare server by making the system’s CPU available more
frequently.
Note. Although servers that use the Pentium or Pentium Pro microprocessor use the PCI bus, these
same servers are often also equipped with EISA buses. Before the PCI bus became common, systems
based on the EISA bus were good choices for medium- to large-capacity NetWare servers that supported
more than 50 users, used more than 50 MB of disk storage, had multiple network cards, and contained
additional peripheral devices.
3.7.5 VESA Bus
Not long after the 80486 chip was introduced, IBM-compatible system board manufacturers struggled to
provide systems that would enable the video and hard-drive peripherals to match the increased speed of
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
the latest microprocessors. The Video Electronics Standards Association (VESA) cooperated with Intel to
design a new system bus architecture that would allow peripheral cards such as the video adapter to
have direct access to the local bus of the system board at the same clock speed as the system board.
With the advent of the Pentium chip, Intel’s PCI bus became the commonly used expansion bus. You
might encounter the VESA bus in PC workstations and NetWare servers that have an 80486
microprocessor. The VESA bus consists of an extension to the 16-bit ISA slot, enabling the slot to be
used for either a VESA-compatible device or a 16-bit adapter. This extension allows a card placed in a
VESA slot to become part of the local bus of the system board and achieve much better data
transmission speeds. According to some industry experts, VESA cards can yield performances that are
up to 10 times better than standard ISA cards. VESA slots can be included on 80486 system boards that
have both ISA or EISA expansion slots. On PC workstations using graphics-intensive applications, a
system’s VESA slot is best used for the video card, as this will greatly increase the performance of
graphics-based applications, which need to send millions of bits per second to the screen. On NetWare
servers, which must be able to move many large blocks of data to and from the disk and the network
cards, the VESA slot is often used for high-speed disk controllers and NICs.
3.7.6 PCI Bus
The latest local bus designed by Intel is called the Peripheral Component Interconnect (PCI) bus. The
PCI bus improves on the older VESA bus design by avoiding the standard input/output bus and using the
system bus to take full advantage of the Pentium chip’s 64-bit data path. In addition, the PCI bus runs at
the 60- or 66-MHz speed of the processor (compared with the 33-MHz maximum speed of the VESA bus)
and, as a result, is being used on many new system boards designed around the Pentium processor.
Another advantage offered by the PCI bus is hardware compatibility between Intel-based computers and
Apple PowerPC-based machines, due to the use of PCI slots in Apple Macintosh computers. The biggest
limitation of the PCI bus is that it will support only three to four slots on a system board. However, new
systems are being designed that will provide multiple PCI buses on the same system board in order to
provide eight or more PCI slots.
Trend. Intel will continue to develop the PCI bus, and it will be used as a standard expansion bus in new
workstations and servers. The PCI bus is the dominant bus on Pentium and Pentium Pro workstations
and can also be found on later 80486 workstations. The VESA bus is found on older 486 workstations.
Combination buses will be the normal configuration, as the PCI bus is usually combined with an ISA bus
(on PC workstations) or the EISA bus (on servers).
3.7.7 NuBus
The Apple NuBus was originally developed by Texas Instruments and was then adopted by Apple for use
on its Macintosh line of computers. The NuBus offers a 32-bit address and data bus running at 10 MHz
and is similar to Micro Channel in that it supports self-configuring boards. With self-configuring boards, a
computer’s system will automatically identify and configure each peripheral added to the system,
eliminating the need to configure them manually by using DIP switches or jumpers.
4.0 Memory
The purpose of the computer’s primary memory unit it to store software and data in a manner that allows
the microprocessor unit to access each storage cell directly. Memory is composed of millions of tiny
switches built into silicon memory modules that can be turned on or off to represent a binary one or zero.
The memory switches are arranged in groups of eight to form memory cells called bytes. Every byte is
assigned a unique number or address that distinguishes it from other memory bytes. Each memory
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
byte can then be used to store one character of data or part of an instruction. The microprocessor can
access memory by sending the address number of the desired byte on the address bus and then
receiving the contents of the memory cell(s) on the data bus. On a 32-bit data bus, four sequential
memory bytes can be sent to or from the microprocessor with one memory access.
4.1 Memory Types
There are four primary types of memory used in microcomputer systems: RAM, ROM, CMOS, and highspeed cache. Each of these memory types has a specific function in the processing of information in a
computer system.

RAM. Random-access memory (RAM) is considered a volatile form of memory because it
depends on constant power; when power is turned off, the contents of the RAM are erased. A
computer’s RAM is its primary workspace where programs and data are stored during
processing. More RAM in workstations allows the use of larger and more complex software
applications. In a NetWare server computer, additional RAM is used for file and directory caching
(file caching is the process of storing often-used disk information in memory). Because memory is
more than 100 times faster than disk access time, the amount of memory available for file
caching directly affects the performance of a NetWare server computer.
The latest development in RAM is extended data out (EDO) RAM. This type of RAM is faster
than older RAM and is now the standard RAM used in Pentium and Pentium Pro microcomputers.
Notes. (1) A special type of RAM, Video RAM or VRAM, is used on graphics cards. (2)
When adding RAM to a microcomputer, be sure you add the correct type of RAM for that
microcomputer. Check the specifications in the User’s Guide that came with the PC.

ROM. Read-only memory (ROM), as its name suggests, cannot be changed. On most
microcomputer systems, ROM is used to store boot instructions and control such basic hardware
functions as the inputting of data from the keyboard or access to the disk drive. Because they
cannot be changed, instructions stored in ROM are referred to as firmware. Because ROM is
slower than RAM, most 80386 and later microprocessors allow moving the contents of ROM into
RAM during booting, a process known as RAM shadowing. RAM shadowing can significantly
increase the speed of such hardware-oriented operations as accessing the screen and keyboard.

CMOS. The original IBM PC bus contained switches that were used to set configuration options
such as memory capacity, disk drives, and video. Today’s system boards contain a built-in setup
program that is used to store this configuration information in a special memory type called
CMOS. CMOS (complementary metal oxide semiconductor) memory uses very little power,
and its contents can be maintained with a small on-board battery when the computer’s power is
off. The CMOS battery is recharged whenever the system is powered. If you add a new disk drive
or more memory, you will need to run a setup program to update your computer’s CMOS
configuration. Many CMOS setup programs are built into the ROM of the system board and can
be executed by pressing a special key sequence (such as the Escape key) while the computer
system is initially booting. Some computers need to be booted with a special disk in order to
change the CMOS configuration settings.
Note. Be aware that the CMOS battery can completely discharge when a computer is
turned off for an extended period of time, causing loss of configuration information.

Cache. Cache memory is very high-speed memory made of chips called SRAM (static RAM).
Most RAM consists of relatively inexpensive chips called DRAM (dynamic RAM). Although
inexpensive, DRAM bears a hidden cost; it needs a special clock cycle to maintain its memory
contents. Because of this extra refresh cycle, DRAM is slower than SRAM because it requires
wait states when used with processors running at speeds above 20 MHz. SRAM’s speed
advantage over DRAM makes it more suitable for caching the most recently used memory
locations. It increases the speed of processing by allowing the processor to access data or
instructions without using the wait states. High-speed (33 MHz and above) computers typically
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
need and use 128-256 KB of cache memory to improve their performance. Intel 80486, Pentium,
and Pentium Pro microprocessors use built-in cache. The 80486 has 8 or 16 KB of cache
memory, the Pentium has 16 KB, and the Pentium Pro has 16 KB built in to the microprocessor
chip. This is known as the level 1 (L1) cache (primary cache). Additional cache, known as the
level 2 (L2) cache (secondary cache), can often be installed on the system board to increase
system performance. Typically, 128-256 KB of L2 cache is used with the 80486 and Pentium. The
Pentium Pro uses a built-in L2 cache of 256-512 KB.
4.2 Single In-Line Memory Modules
Most RAM is currently provided on small memory cards called single in-line memory modules (SIMMs).
SIMMs are arranged on the system board in banks. A bank can contain from one to four SIMM sockets,
and a computer’s system board contains several memory banks. Memory is added in banks by filling all
SIMM sockets in the bank with the same type of SIMM chip. The number of SIMM banks determines the
maximum amount of memory that can be placed on the system board as well as the ease of memory
expansion. If a memory board does not contain enough SIMM banks, you can replace existing SIMMs
with SIMMs of higher capacity in order to expand the computer’s memory.
SIMMs contain 1, 4, 8, 16 or 32 MB of RAM and are supplied in either 30-pin or 72-pin models: 30-pin
SIMMs supply 8 bits to the data bus per module; and 72-pin SIMMs supply 32 bits per module. When 30pin SIMMs are used with an 80386SX processor, each bank must contain two SIMMs (8 bits X 2 SIMMs =
16-bit data bus width). If 30-pin SIMMs are used in an 80486 system that has a 32-bit data bus, each
bank must contain 4 SIMMs (8 bits X 4 SIMMs = 32-bit data bus width). Make sure to choose system
boards that use 72-pin SIMMs when you purchase computers based on 80486, Pentium, or Pentium Pro
processors.
As mentioned previously, each memory bank must be filled with the same type of SIMM, because SIMMs
of different capacities cannot be mixed with a bank. Assume, for example, you are using an 80386SX
processor that has a 16-bit data bus with two SIMM slots per bank and that your computer currently has
four 1-MB SIMMs in bank one for a total of 4 MB of RAM. You want to add more memory, so you will
need to install SIMMs in matching pairs. You need to install at least two more 1-MB SIMMs, giving you a
total of 6 MB. You should not place one 1-MB SIMM in bank two for a total capacity of 5 MB. The next
step up would be to add two 4-MB SIMMs in bank two for a total capacity of 10 MB.
In addition to obtaining the correct capacity for the SIMMs, you need to make sure that the SIMMs are
fast enough to keep up with the clock speed used for the memory banks. SIMMs that are too slow will
cause the computer to crash. The speed of the memory chips is measured in nanoseconds (billionths of a
second). The speed of most SIMMs ranges between 60 and 80 nanoseconds. When adding SIMMs to a
computer, you should check the system’s manual to verify the appropriate chip speeds.
Finally, you need to match the type of RAM used in the PC. If the PC was purchased with EDO RAM, you
should use EDO RAM when you add memory.
Note. The Pentium Pro also supports dual in-line memory modules (DIMMs). This is a combination of two
SIMMs, which are read alternatively in memory access cycles. This results in an effective data bus of 128
bits over the Pentium Pro’s 64-bit data bus.
4.3 Memory Usage
Windows 95 automates memory management that you previously had to specify by statements in the
CONFIG.SYS and AUTOEXEC.BAT files. The same memory models and terms, however, still apply,
because Windows 95 must still support old DOS applications. DOS was designed to run on an 8088
processor in real mode and is therefore limited to managing 1 MB (1,024 KB) of RAM. The first 640 KB of
this 1-MB memory area is referred to as conventional memory and is used by DOS to run software
applications. The memory between 640 KB and 1 MB is called upper memory and is reserved for
hardware use. For example, part of upper memory is used by your video card to store data displayed on
the screen. The network administrator might need to use this memory area when configuring certain
NICs. The memory above 1 MB is called extended memory and is available to microprocessors running
in either protected or virtual mode. Because DOS was not designed to use extended memory, it requires
Local area network (LAN)
Topic 2 – Computer Hardware
Revision 0
Nov 2000
special driver software, called the extended memory system (XMS), to access this memory. An example
of an XMS memory manager is the HIMEM.SYS driver, which provides access to the extended memory
necessary to load the DOS operating system. Operating systems that do not rely on DOS—Windows NT
workstation, UNIX, OS/2, and NetWare—can access extended memory directly without the need for
special drivers.
You might hear the term expanded memory in connection with IBM-compatible computers. Expanded
memory is an older technology that replaces RAM chips on a separate expansion card that is then added
to the IBM-compatible system. Expanded memory was originally designed as a combined effort by Lotus,
Microsoft, and Intel to provide a method for running large programs and worksheets with DOS on 8088based computer systems. Because expanded memory is placed on a separate expansion card, it requires
special software called an expanded memory system (EMS) to swap information or program
instructions stored on the card into page frames located within the upper memory area. This swapping
process causes computers using expanded memory to run more slowly that computers using extended
memory. MS-DOS includes a special EMM386.EXE memory driver that can be used to make extended
memory act like an expanded memory card in order to support applications written to use expanded
memory. Today, application and system software is designed to use extended rather than expanded
memory. You might still encounter IBM-compatible computers that are configured to use expanded
memory in order to support older applications.
5.0 Storage Systems
Advances in disk storage systems have been as important to the development of microcomputer systems
as the improvements made to processors and memory. Instructions and data need to be retrieved from
disks and placed in RAM before the processor chip can act on them. Therefore, both the speed and
capacity of disk storage are critical to the performance of a computer system. Consider a NetWare
server’s primary purpose for a moment. Its major function involves the shared use of its hard disk drives.
The NetWare operating system is specifically designed to maximize the performance and reliability of its
disk storage system.