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.