This chapter discusses the differences in system
architecture among PC- compatible systems and also explains
memory structure and use. In addition, the chapter discusses
how to obtain the documentation necessary for maintaining and
upgrading your computer.
Types of Systems
Many types of PC-compatible systems are on the market
today. Most systems are similar, but a few important
differences in system architecture have become more apparent
as operating environments (such as Windows and OS/2) have
increased in popularity. Operating systems such as OS/2 1.x
and Windows 3.1 require at least a 286 CPU platform on which
to run. OS/2 2.x, 3.x (Warp), and Windows 95 will run on a 386
system, and Windows NT 4.x requires at least a 486 CPU to run.
Knowing and understanding the differences among these hardware
platforms will enable you to plan, install, and use modern
operating systems and applications in order for you to use the
hardware optimally.
All PC-compatible systems can be broken down into two basic
system types, or classes, of hardware:
- 8-bit (PC/XT-class) systems
- 16/32/64-bit (AT-class) systems
The term PC stands for personal computer,
XT stands for an eXTended PC, and AT
stands for an Advanced Technology PC. The terms
PC, XT, and AT as used here are taken
from the original IBM systems of those names. The XT basically
was a PC system that included a hard disk for storage in
addition to the floppy drive(s) found in the basic PC system.
These systems had an 8-bit 8088 processor and an 8-bit
Industry Standard Architecture (ISA) Bus for system expansion.
The bus is the name given to expansion slots in which
additional plug-in circuit boards can be installed. The 8-bit
designation comes from the fact that the ISA Bus found in the
PC/XT class systems can send or receive only 8 bits of data in
a single cycle. The data in an 8-bit bus is sent along eight
wires simultaneously, in parallel.
More advanced systems are said to be AT-class, which
indicates that they follow certain standards and follow the
basic design first set forth in the original IBM AT system.
AT is the designation IBM applied to systems that first
included more advanced 16-bit (and later, 32- and 64-bit)
processors and expansion slots. AT-class systems must have any
processor that is compatible with Intel 286 or higher
processors (including the 386, 486, Pentium, Pentium Pro, and
Pentium II processors) and must have a 16-bit or greater
system bus. The system bus architecture is central to the AT
system design.
The first AT-class systems had a 16-bit version of the ISA
Bus, which is an extension of the original 8-bit ISA Bus found
in the PC/XT-class systems. Eventually, several expansion slot
or bus designs were developed for AT-class systems, including
those in the following list:
- 16-bit ISA Bus
- 16/32-bit Extended ISA (EISA) Bus
- 16/32-bit PS/2 Micro Channel Architecture (MCA)
Bus
- 16-bit PC-Card (PCMCIA) Bus
- 32-bit Cardbus (PCMCIA) Bus
- 32-bit VESA Local (VL) Bus
- 32/64-bit Peripheral Component Interconnect (PCI)
Bus
- Accelerated Graphics Port (AGP)
A system with any of these types of expansion slots is by
definition an AT-class system, regardless of the actual Intel
or Intel-compatible processor used. AT-type systems with 386
or higher processors have special capabilities not found in
the first generation of 286-based ATs. The 386 and higher
systems have distinct capabilities regarding memory
addressing, memory management, and possible 32- or 64-bit wide
access to data. Most systems with 386DX or higher chips also
have 32-bit bus architectures to take full advantage of the
32-bit data transfer capabilities of the processor.
Most PC systems today incorporate 16-bit ISA slots for
backward compatibility and lower function adapters, and PCI
slots for truly high performance adapters. Most portable
systems use PC-Card and Cardbus slots in the portable unit, as
well as ISA and PCI slots in optional docking stations.
Chapter 5, "Bus Slots and I/O Cards," contains a great deal
of in-depth information on these and other PC system buses,
including technical information such as pinouts, performance
specifications, and bus operation and theory.
Table 2.1 summarizes the primary differences between the
older 8-bit (PC/XT) systems and a modern AT system. This
information distinguishes between these systems and includes
all IBM and compatible models.
Table 2.1 Differences Between PC/XT and AT
Systems
System Attributes |
(8-bit) PC/XT Type |
(16/32/64-bit) AT Type |
Supported processors |
All x86 or x88 |
286 or higher |
Processor modes |
Real |
Real/Protected/Virtual Real |
Software supported |
16-bit only |
16 or 32-bit |
Expansion slot width |
8-bit |
16/32/64-bit |
Slot type |
ISA only |
ISA, EISA, MCA, PC-Card, Cardbus, VL-Bus,
PCI |
Hardware interrupts |
8 (6 usable) |
16 (11 usable) |
DMA channels |
4 (3 usable*!8 (7 usable) |
|
Maximum RAM |
1M |
16M/4G or more |
Floppy controller speed |
250 Kbit/sec |
250/300/500/1,000 Kbit/sec |
Standard boot drive |
360K or 720K |
1.2M/1.44M/2.88M |
Keyboard interface |
Unidirectional |
Bi-directional |
CMOS memory/clock |
None standard |
MC146818 compatible |
Serial-port UART |
8250B |
16450/16550A |
The easiest way to identify a PC/XT (8-bit) system would be
by the 8-bit ISA expansion slots. No matter what processor or
other features the system had, if all of the slots are 8-bit
ISA, then the system would be a PC/XT. AT (16-bit plus)
systems can be similarly identified by having 16-bit or
greater slots of any type. These could be ISA, EISA, MCA,
PCCard (formerly PCMCIA), Cardbus, VL-Bus, or PCI. Using this
information, you can properly categorize virtually any system
as a PC/XT type or an AT type. There really have been no PC/XT
type (8-bit) systems manufactured for many years. Unless you
are in a computer museum, virtually every system you would
encounter today is based on the AT type design!
For more information on the other architectural differences
between these types of systems, consult the various sections
of the book that deal with each system component.
Documentation
One of the biggest problems in troubleshooting, servicing,
or upgrading a system is having proper documentation. I
believe that good documentation is critical for system support
and future upgrade capability. Because it can be a problem
getting documentation on systems or components that are older,
the time to acquire documentation is when the system or
components are new.
There are several types of documentation available to cover
a given system:
- System-level documentation. The system-specific
manual(s) put together by the system manufacturer or
assembler. Some companies break this down further into
Operations, Technical Reference, and Service
manuals.
- Component-level documentation. The specific OEM
(Original Equipment Manufacturer) manuals for each major
component such as the motherboard, video card, hard disk,
floppy drive, CD-ROM drive, modem, network card, SCSI
adapter, and so on.
- Chip- and chipset-level documentation. The most
specific and technical manuals which cover items such as the
processor, motherboard chipset, super I/O chip, BIOS, memory
modules, video chipset, and various disk controller, SCSI
bus interface, network interface, and other chips used
throughout the system.
The system- and component-level documentation is essential
for even the most basic troubleshooting and upgrading tasks.
More technical literature such as the chip- and chipset-level
documentation is probably necessary only for software and
hardware developers who have more special requirements.
However, if you are like me and really want to know as much
about a system as possible, then you will find as I do that
having the chip- and chipset-level documentation can give you
insights and information about a system you simply can't get
otherwise. This section will examine all of this documentation
and, most importantly, explain how to get it!
Basic System Documentation
When you purchase a complete system, it should include a
basic set of documenta- tion. What you actually get will vary
widely depending on what type of system you get and who put it
together.
Name-brand manufacturers such as IBM, Compaq,
Hewlett-Packard, Toshiba, Packard Bell, and others will almost
certainly include custom manuals they have developed
specifically for each system they sell. For those types of
systems which use proprietary components, you should contact
the manufacturer for their specific documentation.
Companies who assemble or build systems out of industry
standard components may either produce their own
documentation, or simply include the documentation that is
included with the components they install in their systems.
Most of the larger system assemblers such as Gateway, Dell,
Micron, Midwest Micro, and others will also have their own
custom-produced documentation for the main system unit, and
may even have custom manuals for many of the individual system
components.
This type of documentation is useful for people setting up
a system for the first time or for performing simple upgrades,
but often lacks the detailed technical reference information
needed by somebody who might be troubleshooting the system or
upgrading it beyond what the manufacturer or assembler had
originally intended. In that case, you are better off with any
of the OEM component manuals which are available directly from
the component or peripheral manufacturers themselves.
Most of the smaller system assemblers will forego any
custom-produced system documentation and simply include the
component level manuals for the components they are including
in the assembled system. For example, if an Asus motherboard
and STB video card were included in a particular system, then
the manuals from Asus and STB which originally came with those
products would be included with the assembled system.
Getting Documentation from an Assembler
Some system assemblers like to keep the component
documentation and not include it with the systems they
build. This forces the purchaser of the systems to go back
to the assembler for any support or technical information,
and also tends to make the purchaser believe that the
assembler actually manufactured the system rather than
simply assembled it using off-the-shelf components. I would
not recommend purchasing from any system assembler who did
not include all of the documentation for the individual
system components they are installing.
The standard manuals included with most system components
and peripherals contain basic instructions for system setup,
operation, testing, relocation, and option installation. Some
sort of basic diagnostics disk (sometimes called a
Diagnostics and Setup or Reference Disk)
normally is included with a system as well. Often the
diagnostics are simply a custom labeled version of a commonly
available commercial diagnostic program.
TIP: Most system vendors and equipment
manufacturers have jumper settings and manuals available on
their Web sites in downloadable form. Appendix A contains a
list of vendors and their Web sites.
Component and Peripheral
Documentation
It is a well-known fact that many systems sold today are
not really manufactured as a custom unit by a single company
but instead are assembled out of standard off-the-shelf
components that are available on the open market. In fact, I
normally recommend that people purchase exactly that type of
system, because all of the components conform to known
standards and can easily be replaced or upgraded later.
Even proprietary manufactured systems such as IBM, Compaq,
Hewlett-Packard, Packard Bell, and others use at least some
off-the-shelf standard components (disk drives, for example).
To fully document a system, I recommend you take an inventory
of the standard components used, and collect all of the OEM
documentation or product manuals for them.
This process is simple; when I am supporting a given
system, I first disassemble it and write down all of the
information on each of the components inside. Sometimes you
will need to do a little more investigating or even ask the
company who assembled the system exactly what components they
included. Most components such as hard disks, CD-ROM drives,
video cards, sound cards, network cards, and more are pretty
easy to identify. Somewhere on the device or card there should
be a label indicating at least the manufacturer and usually
also the model number. From this, you can look up the
manufacturer in the vendor listing in Appendix A of this book.
Using that information, you can contact the company via
telephone, fax, or Internet Web site to obtain the complete
documentation on their products.
Motherboards can be tricky to identify because not all
manufacturers mark them clearly. In that case, you are best
off contacting the company who sold you the computer to ask
them exactly what motherboard you have. Don't be afraid to ask
the company exactly what motherboard or other components they
are installing in the systems they sell. If they can't or
won't answer, you may be better off purchasing from a
different company in the future. If the company who sold the
system is no longer available or cannot help, check the
paperwork that came with the system. Sometimes there are clues
in the original paperwork that might indicate what motherboard
your system includes. Most of the popular motherboard
manufacturers are listed in the vendor list in Appendix A.
You will also find a listing in the vendor list for
MicroHouse. They sell a reference manual called the
Encyclopedia of Main Boards, which has information on
hundreds of different PC-compatible motherboards. As a last
resort, you might be able to identify your board by comparing
it to those shown in their book. In their motherboard
encyclopedia, you will find information about the jumper and
switch settings of most PC boards on the market.
As an example, one system I worked on is a Gateway P6
(Pentium Pro) 200MHz system which included the following
industry standard components:
Motherboard: |
Intel VS440FX "Venus" |
Video Card: |
STB Velocity 3D |
Hard Disk: |
Quantum Fireball TM3840A |
Floppy Drive: |
Panasonic JU-256 |
CD-ROM Drive: |
Mitsumi FX120 |
Many of the larger system assemblers like Gateway and Dell
have been using Intel motherboards. Most motherboards today
use Intel processors and chipsets, but some people may not be
aware that Intel makes complete PC motherboards as well. Even
so, Gateway did not include the actual Intel motherboard or
other component manuals but instead included their own custom
manuals for the motherboard, video card, hard disk, and CD-ROM
drive. By contacting the individual companies directly over
the Internet and via the telephone, I was able to obtain more
detailed documentation on all of these products. Many times
the OEM documentation or product manuals can be downloaded
directly from the respective companies' Web sites, often in
the form of Adobe Acrobat .PDF files, which you can read with
the Acrobat reader available for free for downloading from
Adobe.
NOTE: Not all of these manuals are available
online, and even if they are, it is still nice to have the
printed manuals or datasheets in your documentation library
for future reference.
Chip and Chipset Documentation
If you really want the ultimate in documentation for your
system, I highly recommend getting the documentation for the
various chips and chipsets in your system. This would include
specific manuals for each of the major chip-level components
in the system--such as the processor, motherboard chipset,
BIOS, super I/O chipset, and so on. Before you can get this
documentation, you must first identify all of the relevant
chips and chipsets in your system.
The process is relatively simple. Look at the documentation
for each major component, especially the motherboard. The OEM
motherboard documentation should tell you which chipset is
used on the board, which processors are supported, and which
super I/O chip is used. From the OEM documentation you have on
your system components, you should be able to find out what
the following major chip and chipsets are:
Processor |
Super I/O Chip |
Motherboard Chipset |
Video Chipset |
ROM BIOS |
|
If your motherboard has an integrated video card, then the
video chipset type will be listed there also. If you have a
separate video card, then look in the video card manual and it
should clearly identify the video chipset used.
The most important chips you will want to identify are on
the motherboard. The first thing you would want to identify is
the processor. This should be relatively easy; most PC systems
use Intel processors. A small percentage of systems use AMD or
Cyrix processors, or versions of these processors sold under
other names. The documentation that comes with the system will
normally identify which brand of processor you have, and which
model and speed it is.
If you aren't sure what processor is in the system,
software programs such as the MSD (Microsoft Diagnostics)
program are included with Windows, or a system diagnostic you
purchase such as the Norton Utilities.
Normally the processor is the largest chip on the
motherboard, and can often be identified by simply reading the
information stamped on it. In some cases, the processor will
have a heat sink or fan attached to the top, in which case you
may have to either remove the heat sink or fan to read the
information stamped on top of the chip, or simply remove the
entire processor and heat sink or fan assembly to read the
information stamped on the bottom of the chip.
The chipset is difficult for software to determine, so you
will either have to find out which chipset you have from the
motherboard documentation or by first-hand inspection. The
motherboard chipset normally consists of several large chips
on the board; there are chipsets that use anywhere from one to
six chips. Generally each chip in the set will have a part
number stamped on it, but the chipset itself will be named
after the main chip. The Pentium Pro board I mentioned had two
chips labeled 82441FX and 82442FX; these are both a part of
what Intel calls the 440FX chipset.
Finding out the manufacturer of the motherboard BIOS is
easy; that is normally found in the motherboard manual. It is
also displayed, along with the exact version number you have,
every time you power the system on. Most systems today use an
AMI, Award, or Phoenix BIOS, but there are several other
manufacturers as well.
Virtually all motherboards built in the '90s and later
include a special interface chip called a super I/O chip. This
is a single chip device that normally includes the following
components:
- Primary and secondary IDE host adapters
- Floppy controller
- Two serial ports
- One parallel port
Some do not include the IDE host adapters, such as the
National Semiconductor 87308 used in the Pentium Pro
motherboard I have been referring to. That particular chip
also includes an 8042-style keyboard controller and a
MC146818-style real-time clock with nonvolatile CMOS RAM.
Other super I/O chips may include a game (joystick) interface
as well. Obtaining the documentation for your particular super
I/O chip will of course tell you exactly what it's
capabilities are.
Another important chipset in a system is the video
chipset. This is normally found on the video card, or on
the motherboard if the motherboard has built-in video. The OEM
video card or motherboard documentation should tell you
exactly which video chipset you have. If not, then you can use
free software such as MSD or commercial programs such as
Norton Utilities to identify which chipset you have without
even opening up the case. A last resort would be to open the
system and read the part number right off of the video
chipset, which is usually the largest chip on the video
card.
Using as an example a Pentium Pro 200MHz system as an
example, I found it contained the following main chip and
chipset components:
Processor: |
Intel Pentium Pro |
Motherboard Chipset: |
Intel 440FX "Natoma" |
ROM BIOS: |
AMI |
Super I/O Chip: |
National Semiconductor PC87308 |
Video Chipset: |
S3 Inc. ViRGE/VX |
Note that this particular motherboard did not have the
video integrated, so the video chipset was on the video
card.
From this documentation, I learned about the capability to
increase the clock multiplier setting on the motherboard to an
otherwise undocumented 3.5x, which resulted in running the
Pentium Pro 200 chip at 233MHz! I was also able to get more
information on the various serial, parallel, and disk
controllers contained in the super I/O chip, and learned more
about the advanced CMOS settings in the BIOS Setup
routines.
For example, I often get questions about the Advanced CMOS
settings. Most people assume that these settings would be
described in their ROM BIOS documentation because the
ROM-based CMOS Setup program in their system controls these
settings. If you contact the BIOS manufacturer or read the
BIOS documentation, you will quickly find out that the ROM
BIOS manufacturer knows little or nothing about these
settings. In fact, these settings actually have little or
nothing to do with the particular ROM BIOS used, and
everything to do with the particular motherboard
chipset used. You can find descriptions of all these
settings in the documentation for your motherboard chipset,
which can be obtained from the chipset manufacturer.
Manufacturer-Specific
Documentation
If your system is from a name-brand manufacturer--such as
IBM, Compaq, Hewlett-Packard, Toshiba, and others--then there
may be a wealth of information available in
manufacturer-specific manuals and documentation. Because of
the specific nature of the information in these types of
manuals, you most likely will have to obtain it from the
manufacturer of the system.
The process of obtaining other manufacturers' manuals may
(or may not) be easy. Most large companies run responsible
service and support operations that provide technical
documentation. Other companies either do not have or are
unwilling to part with such documentation, in an effort to
protect their service departments (and their dealers' service
departments) from competition. Contact the manufacturer
directly; the manufacturer can direct you to the correct
department so that you can inquire about this information.
Information on how to contact most PC manufacturers can be
found in the vendor listing in Appendix A of this book.
Warranties and Service Contracts
Extended warranties are a more recent trend in the computer
industry. With the current fierce competition among hardware
vendors, a good warranty is one way for a specific
manufacturer to stand out from the crowd. Although most
companies offer a one-year warranty on their systems, others
offer longer warranty periods, such as three years or
more.
In addition to extended-length warranties, some
manufacturers offer free or nearly free on-site service during
the warranty period. Many highly competitive mail-order
outfits offer service such as this for little or no extra
cost.
TIP: Most companies offer extended-length
warranties and free or low-cost on-site service. If your
system is "mission-critical," meaning it absolutely must be
functioning all the time (such as with a network file,
database, or application server ), you might want to
consider an on-site service contract. Such contracts are
usually overkill for a standard PC.
In most normal cases, service contracts are not worth the
price. In the retail computer environment, a service contract
is often a way for a dealer or vendor to add income to a sale.
Most annual service contracts add 5 to 10 percent of the cost
of the system. A service contract for a $5,000 system, for
example, costs $250 to $500 per year. Just as in the auto
industry, salespeople in the computer industry are trained to
vigorously sell service contracts. Extra cost warranties or
service contracts are largely unnecessary except in special
situations.
NOTE: Retail computer stores, electronics
stores, and PC distributors make almost no money on the home
PC you buy. Stiff price competition has taken away any
profit they might make. However, add-ons--such as service
contracts, surge suppressors, mouse pads, and so on--are
very high-margin items. Stores make their money by selling
the extras, not the computer itself.
The high prices of service contracts also might affect the
quality of service you receive. Technicians could try to make
their work seem more complex than it actually is to make you
believe that the contract's price is justified. For example, a
service technician might replace your hard disk or entire
motherboard with a spare when all you need is low-level
formatting for the hard disk or a simple fix for the
motherboard such as a single memory chip. A "defective" drive,
for example, probably is just returned to the shop for
low-level formatting. Eventually, it ends up in somebody
else's system. Replacing a part is faster and leaves the
impression that your expensive service contract is worth the
price because you get a "new" part. You might be much less
impressed with your expensive service contract if the service
people visit, do a simple troubleshooting procedure, and then
replace a single $2 cable or spend 15 minutes reformatting the
hard disk.
With some basic troubleshooting skills, simple tools, and a
few spare parts, you can eliminate the need for most of these
expensive service contracts. Unfortunately, some companies
practice deceptive servicing procedures to justify the
expensive service contracts they offer. Users are made to
believe that these types of component failures are the norm,
and they have a mistaken impression about the overall
reliability of today's systems.
TIP: If you have many systems, you can justify
carrying a spare-parts inventory, which can also eliminate
the need for a service contract. For less than what a
service contract costs for five to 10 systems, you often can
buy a complete spare system each year. Protecting yourself
with extra equipment rather than service contracts is
practical if you have more than 10 computers of the same
make or model. For extremely time-sensitive applications,
you might be wise to buy a second system along with the
primary unit--such as in a network file-server application.
Only you can make the appropriate cost-justification
analysis to decide whether you need a service contract or a
spare system.
In some instances, buying a service contract can be
justified and beneficial. If you have a system that must
function at all times and is so expensive that you cannot buy
a complete spare system, or for a system in a remote location
far away from a centralized service operation, you might be
wise to invest in a good service contract that provides timely
repairs. Before contracting for service, you should consider
your options carefully. These sources either supply or
authorize service contracts:
- Manufacturers
- Dealers or vendors
- Third parties
Although most users take the manufacturer or dealer
service, sometimes a third party tries harder to close the
deal; for example, it sometimes includes all the equipment
installed, even aftermarket items the dealers or manufacturers
don't offer. In other cases, a manufacturer might not have its
own service organization; instead, it makes a deal with a
major third-party nationwide service company to provide
authorized service.
After you select an organization, several levels of service
often are available. Starting with the most expensive, these
levels of service typically include:
- Four-hour on-site response
- Next-day on-site response
- Carry-in/ship-in, or "depot," service
The actual menu varies from manufacturer to manufacturer.
For example, IBM offers only a full 24-hours-a-day,
7-days-a-week, on-site service contract. IBM claims that a
technician is dispatched usually within four hours of your
call. For older systems, but not the PS/2, IBM also offers a
courier or carry-in service contract. Warranty work, normally
a customer carry-in depot arrangement, can be upgraded to a
full on-site contract for only $40. After the first-year $40
contract upgrade expires, you can continue the full on-site
service contract for standard rates.
If you have ever bought a service contract, you may be
surprised by the pricing. In most cases, the price will be so
high that you will only be able to justify it for
mission-critical systems such as file servers.
|