Many instructions involve the arithmetic and logic unit (ALU). This works in conjunction with the General
Purpose Registers - temporary storage areas which can be loaded from memory or written to memory. A typical ALU instruction might be to add the contents of a memory location to a general purpose register. The ALU also alters the bits in the Status Register (SR) as each instruction is executed; this holds information on the result of the previous instruction. Typically, the SR has bits to indicate a zero result, an overflow, a carry and so forth. The control unit uses the information in the SR to execute conditional instructions such as ‘jump to address 7410 if the previous instruction overflowed’.
This is about all there is as far as a very basic processor is concerned and just about any operation can be carried out using sequences of simple instructions like those described.
The invention and transition
Credit for the invention of the microprocessor is given to Intel. These first microprocessors were the 4004 and were released in 1971. This single chip matched the performance of the room size computer ENIAC from the fifties (Wyant and Hammerstrom, 19). This chip could only support a four bit bus. These four bits only offered the possibility of coding 16 symbols (2^4=16). Sixteen symbols were enough for digits 1-9 and then some operators. This limited the 4004 to calculator usage. The 4004 ran at 108 kHz which is 1/10 of 1 MHz (Rosch, 66). The smallest feature on the chip measured 10 microns and contained 2300
The next generation of Intel chips used an 8 bit data bus. The first member of this generation was released
in 1972 and was called the 8008. This chip was the same as the 4004, but it had 4 more bits on each register. This chip had enough bits to code 256 symbols (2^8=256). This number is easily enough to encode our alphabet, numerals, punctuation marks, etc. The 8008 also ran a little faster than the 4004 with its speedy clock of 200 kHz. The 8008 contained 3500 transistors and had line widths 10 microns. Both chips had a MIPS of 0.06 (Rosch, 66).
The next member of the Intel family was born in 1974 and was called the 8080. This chipped was intended to handle byte sized data (8 bit). The 8080 contained 6000 transistors and had 6 micron technology. This chip performed at 0.65 MIPS and had an internal clock speed of 2 MHz. This was one of the first chips to have the capabilities of running a small computer (Rosch, 66).
In June of 1978 the 8086 family was released by Intel. These chips used 16 bit registers. The fastest chip in this series ran at 10 MHz and could execute.75 MIPS. This chip forced engineers of the time to begin developing fully 16 bit devices, which were more expensive than their 8-bit brethren. Because of this, the 8086 family was considered ahead of it's time (Rosch, 67-68).
A year later Intel introduced the 8080. This chip was a step backwards in chip evolution with its 8 bit data bus. The 8080 could process.64 MIPS with its 6000 transistors. The 8080 used 6 micron technology. This chip is worth mentioning primarily because IBM chose to use it in its first personal computer. IBM was able to use the 8088 with existing 8 bit hardware, which was more cost effective. Later IBM began using the 8086 in its newer systems (Rosch, 68).
In 1982 Intel released the 80286. The 286 family was available in clock speeds of 8, 10, and 12 MHz that could execute 1.2, 1.5, and 1.66 MIPS respectively. The 80286 contained 134,000 transistors with 1.5 micron technology. These chips all used a 16 bit data bus and were used by IBM in it's AT models. This was also the first chip to use virtual memory, or using disk space as RAM (Random Access Memory). To allow full downward compatibility the 286 was designed to have two operating modes.
These modes are real and protected mode. Real mode mimics the operation of an 8086. Protected mode allows multiple applications to be run simultaneously and not interfere with each other (Rosch, 70-71).
The next member to the Intel family was added in November 1985 and was the 80386. These chips are offered in speeds of 16, 20, 25, 33 MHz and can process 5.5, 6.5, 8.5, and 11.4 MIPS respectively. The number of transistors in the 80386 is 275,000 with 1.5 micron technology. The 386 family doubled the register size to 32 bits. Also the 386 uses 16 bytes of prefetch cache that the chip uses to store the next few instructions.
The 386 has three models which are called the 386DX, 386SX, and the 386SL. The 386DX was the original and most powerful. The 386SX is a more economical sibling to the DX. It is basically scaled down, less powerful DX. Also the SX uses a 16 bit data bus. The SL also uses 16 bit buses but it includes power saving features targeted at notebook usage. The SL uses 1.0 micron technology and contains 855,000 transistors (Rosch, 72-78).
The 80486 family was introduced in April 1989 and became a "better 386" (Rosch, 78). The 486 was originally released in a DX model with speeds of 25, 33, and 50 MHz that processed 20, 27, and 41 MIPS respectively. The DX also contains a math coprocessor or floating point unit that helps speed up math operations. The 486DX uses a 32 bit bus and contains 1,200,000 transistors. It uses 1.0 micron technology in the 25 and 33 MHz models, but in the 50 MHz model uses 0.8. The next to be released was the 486SX. The SX was designed to cut cost at the price of not having a math coprocessor. As a result the SX will not perform as well as the DX in math intensive operations. The SX contains 1,185,000 transistors and uses the same technology as the DX. The SX is available in 16, 20, 25, and 33 MHz models that process 13, 16.5, 20, and 27 MIPS respectively.
To add the power of a FPU (Floating Point Unit) to the SX Intel released the OverDrive upgrade processors in March 1992. The first, the 486DX2, incorporated clock doubling technology. These chips operate at double the bus speed. These chips are available in 50 and 66 MHz models that can process 41 and 54 MIPS respectively.
The 50 MHz model was designed to replace the 25 MHz 486SX and the 66 MHz model was for the 33 MHz 486SX. The OverDrive chips contain 1.2 million transistors. The next to be released was the SL model which was, like the 386SL, targeted at laptop usage. The SL contains 1.4 million transistors and can process 15.4, 19, and 25 MIPS while running at 20, 25, and 33 MHz respectively. The 486DX4 was the next OverDrive chip to be released. It contains clock tripling technology.
The DX4 can turn a 33 and 25 MHz 486's into DX4-100 and DX4-75 respectively. These chips can process 60 and 81 MIPS running at 75 and 100 MHz respectively. The DX4 uses 0.6 micron technology (Rosch 84-85).
The next addition to the Intel family was the Pentium. The Pentium was originally released in a 60 MHz model that operated at 5 volts. This chip contains 3,100,000 transistors and can process 100 MIPS. The next to be released was the 66 MHz model. It uses the same technology but is a 3.3 volt chip and can process 112 MIPS. Currently the Pentium is available in 66, 75, 90, 100, 120, 133, 150, and 166 MHz models. Beyond the 75 all Pentiums use 0.6 micron technology. A 180 MHz is slated for future release. The Pentium family is, like all of Intel's chips, uses CISC technology. Also they use pipelining, superscalar architecture, and branch prediction logic. A Pentium OverDrive is also available for upgrading 486 systems to Pentium technology. The Pentium OverDrive is available in a 63 and 83 MHz version (Rosch, 85-87).
After the Pentium, the only more advanced chip Intel has for personal use is the Pentium Pro. This chip has only been available for a short time and is targeted at workstation and server usage. It will only run Windows NT and native 32 bit software at an increased speed. When using 16 bit software, the less powerful Pentium will outperform its larger sibling. The Pentium Pro also contains 256K (256,000 bytes) of on chip cache memory.
The only certainty in the future of processors is constant improvement. One prediction for the future is called Moore's Law. This prediction is named after Intel cofounder Gordon Moore who presented it in 1965. The law states the transistor densities will double every two years. Line width is also continuing to shrink and is estimated to be at 0.2 microns by the turn of the century. When all is considered the future of computers is very exciting (Wyant and Hammerstrom, 184-185).
Trends of development
The control unit coordinates and controls all the other parts of the computer system. The control unit even oversees the operations of the input and output devices. The arithmetic-logic unit does the actual processing by performing mathematical operations and logical operations, such as making comparisons. In a microprocessor, the control unit and the arithmetic-logic unit are mounted on a single silicon chip. Large computer systems, as well as newer workstations and network servers, frequently contain more than one central processing unit. Multiple CPUs enable the computer to execute more than one instruction, or process more than one program, at the same time. This capability is known as multiprocessing.
Processor manufacturers must carefully consider compatibility when introducing new models. In particular, manufacturers must decide whether to make the new chip downwardly compatible with previous models. A downwardly compatible chip can run the programs designed to run with the earlier chip(s). To introduce a microprocessor that is not downwardly compatible with previous models is very risky. People may not buy a computer that cannot run the programs they already own. Manufacturers learned this lesson with early mainframe computers.
For this reason, the microprocessors used in today's personal computers are descendants of older processor designs. Two brands predominate: Intel and Motorola (although Digital Equipment Corporation is also producing some impressive microprocessors). Intel microprocessors, including the Pentium chip, are downwardly compatible with chips dating all the way back to Ted Hoff's 4004, the world's first processor. The Motorola 68040 series processor is downwardly compatible with the 68000 (dating from the early 1980s), the 68020, and the 68030.
CPU performance is evaluated by the number of operations that the processor can carry out in one second.
Today's fastest processors can carry out many millions of operations per second! A processor's speed is determined by two major factors: bus width and clock speed.
A data bus connects the CPU and memory and provides a pathway to the computer's peripherals. A microprocessor has both an internal data bus and an external data bus. Sometimes the internal data bus is wider than the external data bus. The internal data bus operates only within the microprocessor itself; the external bus regulates communication with the rest of the computer. For example, the CPU used in the original IBM PC had a 16-bit internal data bus and an 8-bit external data bus. The use of a narrower external bus enables designers to use inexpensive, existing peripherals, such as disk drives and memory chips. However, this design is a compromise that results in a substantial performance penalty.
Bus width is not the only design factor that affects a computer's speed. The system clock regulates the CPU's processing functions by emitting a pulse at regular intervals. The clock speed is the number of times that the system clock pulses in one second. Clock speed is usually measured in millions of pulses, or cycles. One million cycles is a megahertz. Clock speed alone is not an adequate gauge of a processor's performance. A 32-bit chip can process data much more rapidly than a chip hobbled by a 16-bit external data bus, even if the clock speed is the same. The number of operations per clock tick, or cycle, also affects performance. Most computers perform one operation per cycle. The Pentium and Pentium Pro chips, however, use a superscalar architecture that permits more than one instruction to be executed each clock tick.
CISC stands for complex instruction set computer. A CISC chip, such as the Motorola 68040 or the Intel Pentium, provides programmers with many instructions, and the processing circuitry includes many special-purpose circuits that carry out these instructions at high speed. Because the chip provides so many processing tools, CISC designs make the programmer's job easier. CISC chips, however, are complex and expensive to produce, and they run hot because they consume so much current.
For this reason, RISC chips are less complex, less expensive to produce, and more efficient in power usage. The drawback of the RISC design is that the computer must combine or repeat operations to complete many processing operations. (For example, you can eliminate multiplication circuitry by repeated addition.) RISC chips also place extra demands on programmers, who must consider how to get complex results by combining simple instructions. But careful tests show that this design results in faster processing than the CISC chips. An example of a RISC chip (with certain CISC compromises) is the PowerPC processor, which was developed jointly by Apple, IBM, and Motorola.
Computer designers argue about which design is best, but the marketplace will decide. Which will win--CISC or RISC? CISC chips are still in production because so many people use CISC software. (RISC chips can run CISC programs only under software emulation, which slows performance.) RISC chips, such as the PowerPC, may become popular if enough native applications become available. (Bohr, M.
1998)
Even so, the distinction between CISC and RISC may become meaningless. CISC manufacturers are employing many RISC design features that enable the chips to carry out more than one instruction at a time. These features include superscalar architecture (a design that enables the computer to process more than one instruction at a time), pipelining (a design that provides two or more processing pathways that can be used simultaneously), and branch prediction (a module that tries to predict the most effective way to route an instruction through the microprocessor). In the meantime, RISC manufacturers are finding that they must include some CISC design components to ensure compatibility.
The distinctions among the Intel 486, Pentium, and Pentium Pro chips are much clearer when you understand the terminology. The 486DX chip operates at a maximum clock speed of 100 megahertz (the 486DX4-100) and has a 32-bit architecture (internal data bus, external data bus, and address bus.) The
Pentium chip has a 32-bit internal data bus and address bus but also has a 64-bit external data bus. The Pentium Pro has clock speeds up to 200MHz. Though the Intel processor can now reach speeds higher then this.
The instructions to start the computer are stored in read-only memory chips, which are not volatile. Read-only memory chips are manufactured with instructions stored permanently on them. The instructions to start the computer are on a special chip known as a ROM BIOS chip (Basic Input/Output System).
Some ROM chips are manufactured with instructions that are appropriate for a specific end user. These specially programmed ROM chips are PROM (Programmable Read-Only Memory) chips. Once the chip is programmed, the contents cannot be altered. Newer chips, EPROM (Erasable PROM) chips, can be removed from the computer, erased using a special device, and reprogrammed. The newest type of
ROM chips, EEPROM (Electrically Erasable PROM) chips, can be altered electrically using special programs, without being removed from the computer.
No processor could function without high-speed memory, where the processor can store the programs and data it is using. (When an advertisement states that a computer system has 4M of RAM, the ad is referring
High speed memory is what this advertisement is referring to.) Think of memory as a temporary scratch pad that the processor uses while carrying out its operations. Storage devices, such as disk drives, store and retrieve data too slowly for this purpose.
Memory has many different names. It is called random-access memory--or just RAM--as well as primary memory. And sometimes, just to confuse things further, memory is called primary storage. This storage is in contrast to storage devices that are referred to as secondary storage, such as disks. If one Increases the memory, performance of the system shall improve. For today's Microsoft Windows and Macintosh applications, 8M of RAM is the absolute minimum, and 16M is much better. Many programs run much more quickly with 16M, which is large enough to enable most of the program's instructions to be kept in memory. With 8M, the program must access instructions from secondary storage, which is much slower.
The increased transistor count was also used to expand microprocessor capabilities. The world's first microprocessor was four bits wide. Today, 64-bit microprocessors have emerged, and will be the norm in just a few years. Making microprocessors wider, means that a lot more transistors have to be used, as
registers and Arithmetic and Logic units (ALUs) have to increase in width.
Better support for operating systems was included. Architects used the increased transistor count to add
Memory management units (MMUs), support for virtual memory and memory protection. More circuitry was added still to support multiprocessing, the cooperation between several microprocessors to
complete the same task. (Patt Y. et al 1997)
Computer applications have also become more complex. At first, microprocessors only had to deal with integers. Later on, floating point units were added. Today, microprocessors have to cope with so-called natural media types: 3D graphics, audio and video. More transistors have been utilized to support these new data types. The most famous example is Intel's MMX (Multimedia Extensions) technology. Similar extensions can be found in all modern architectures such as the MIPS (MIPS Digital Media eXtensions), Sun Sparc (Visual Instruction Set), PowerPC (AltiVec) e.t.c. Microprocessors, also called central processing units (CPUs), are frequently described as the "brains" of a computer, because they act as the central control for the processing of data in personal computers (PCs) and other computers. Chipsets perform logic functions in computers based on Intel processors. Motherboards combine Intel microprocessors and chipsets to form the basic subsystem of a PC. Because it's part of every one of your computer's functions, it takes a fast processor to make a fast PC.
Ever since this first processor was introduced the market has done nothing but soared to unbelievable highs. The first processor common in personal computers was the 8088.This processor was introduced in June of 1978. It could be purchased in three different clock speeds starting at style 5 Megahertz and going up to 10
Megahertz. This CPU had 29,000 transistors. The 80286 and 80386 processors then evolved. The 386 was the first processor to be introduced in the DX, SX, and SL versions. Next came the 80486 processors of which there were even more choices here. The first 486 processor had 1,200,000 transistors and the latest have 1.4 million transistors. There speeds varied any where from 16 MHz on the first ones to 100 MHz on the most recent 486 processors. Some of which are still in use in homes all around the country.
In March 1993, Intel invented the Pentium Processor. It ran at clock speeds of 60 & 66 Mhz. These first Pentium processors had 3.1 million transistors, and had a 32-bit data path.
The P7 (originally referred to a 64-bit 80x86 which was dropped in favour of the IA-64) was first released as the Pentium 4 in December 2000.
The SPARC and ARM processors are a mere page in the great book of processor history. There will be many new and extremely different processors in the near future. A tremendous amount of time and money have been put into the making and improving of the processor. The improving and investment of billions of pounds are continually going toward the cause of elaborating the processors. The evolution of the processor will continue to evolve for the better until the time when a much faster and more efficient electronic device is invented. This is turn will create a whole new and powerful generation of computers. Hopefully this report has given the reader some insight into the world of these two RISC processors and how they compare to that of CISC architectures developed by Intel and Motrola.