This article will discuss some of the processors of the past and how they have evolved, starting with the 8086 and working forward to the 80486 processors. Along the way, the characteristics—data bus, address bus, and speed—of each processor will be identified.

I will also identify any new architecture changes that are introduced with a specific processor (for example, a built-in math co-processor). Once a change like this is implemented, all subsequent processors follow.


8086
In 1978, Intel introduced the first major processor for personal computers—the 8086—which had a 16-bit data bus, 16-bit registers, and a 20-bit address bus. A 20-bit address bus meant that the 8086 could access 1MB of RAM. The speed of the 8086 ranged from 4.77 MHz to around 10 MHz, which is extremely slow considering today’s standards (approximately 1.5 GHz).

8088
The 8086 chip was too expensive for PC manufacturers to use in their systems and still sell the system at a reasonable price to their customers. So Intel introduced the cheaper 8088 chip a year after the 8086. Like the 8086, the 8088 processor had 16-bit registers and a 20-bit address bus (which meant it could access 1MB of RAM).

However, the data bus was decreased from 16-bits to 8-bits. The 8088 ran at the same speed as the 8086, keeping its speeds at 4.77 MHz and 8 MHz.

The 8086/8088 did not include a built-in cache, nor did it have a built-in math co-processor.

If you wanted to add a math co-processor to your system, you would purchase an 8087 chip to sit on the system board beside the CPU. The 8087 chip was specifically designed as the math co-processor for the 8086 and 8088 processors.

80286
In 1982, Intel produced the 80286 chip, which had 16-bit registers and a 16-bit data bus, and ran at speeds ranging between 6 MHz and 20 MHz. Other than the speed increase, these characteristics matched that of the 8086—this time, however, the market was there.

The 80286 also increased the size of the address bus to 24-bits, which meant that it could access up to 16MB of RAM. Like the 8086/8088, the 80286 processor did not contain its own internal cache, nor did it include a math co-processor. Processors prior to the 80286 chip ran in real mode, while the 80286 processor introduced what is known as protected mode. The following sections compare real mode with protected
mode.

Real mode
Real mode meant that the processor accessed memory as a whole and dealt with it as a single entity. In other words, real-mode processors did not have any multitasking capabilities—the capacity to divide memory up into multiple parts and run a different application (or task) in each part, switching back and forth between them.

Protected mode
Protected-mode processors support the dividing up of system memory into different parts and assigning a different application to each part of memory. Therefore, protected-mode processors support multitasking and multitasking operating systems, such as Windows 95/98, Windows NT, and Windows 2000.

Protected-mode processors also support virtual memory, which is the process of using hard disk space as emulated memory. This means you could have 2MB of RAM while the system is also using 10MB of hard disk space as “pretend” RAM. In this case—as far as the applications that are running are concerned—there is
12MB of RAM.

80386DX
In 1985, Intel released its first 32-bit processor, the 80386DX, which had a 32-bit data bus, a 32-bit address bus, and 32-bit registers. The 32-bit address bus meant that the 80386DX processor could access 4 gigabytes (GB) of RAM, which is an amazing improvement over previous processors (unfortunately, most people can’t afford to purchase 4 GB of RAM).

The speed of the 80386DX processor ranged between 16 MHz and 40 MHz. The 80386DX contains no built-in cache, and the math co-processor (the 80387 chip) has to be purchased separately. Once again, the math co-processor would be inserted on the system board in the math co-processor socket.

80386SX
Three years after the 80386DX chip was out, Intel released the 80386SX, which was a lower-end 386 chip. The 80386SX was a 16-bit processor, meaning it had a 16-bit data bus. It could also only access 16MB of RAM, so the address bus had been reduced to 24-bit. The speed of the 80386SX processor ranged from 16 MHz to 33 MHz. The registers were maintained as 32-bit registers.

Although both flavors of the 80386 chips support real and protected mode, they have taken this support to the next level. These chips enable on-the-fly switching between the two modes, whereas the 80286 processor had to be reset before it could switch from one mode to another.

The major difference between the 80386DX processor and the 80386SX processor is that the DX flavor is a 32-bit processor, while the SX flavor is a 16-bit processor.

When you compare the characteristics of the 80286 and the 80386SX chip, you realize that the 80286 chip is competitive with the 80386SX chip. In reality, an 80386SX chip is nothing more than a glorified 80286 chip with a bigger price tag.

80486DX
In 1989, major advancements were made in the performance of the computer system when Intel released the 80486DX chip. This chip had a 32-bit data bus, a 32-bit address bus (4GB of RAM), and 32-bit registers.

The 80486DX chip introduced two major advancements in CPU technology. The first was the idea of integrating cache directly into the chip. The 80486DX had 8 kilobytes (K) of built-in cache, or what is called L1 cache. The second major advancement was that the math co-processor was integrated inside the 80486DX chip. Now, instead of buying a math co-processor chip and inserting it onto the system board,
the chip was integrated and working as long as it was enabled in the system BIOS.

The speed of the 80486DX ranged from 20 MHz to 50 MHz (20 MHz, 25 MHz, and 33 MHz were the most popular speeds). After the original 486s, a second generation of the 80486DX arrived that were marketed as 80486DX2-50, 80486DX2-66, and 80486DX4-100. The following sections discuss the DX2 and DX4 model processors.

DX2
The “80486DX” portion of “80486DX2-50” means that the processor is the DX flavor of the 80486. The “2” after the “DX” implies “clock double,” a term indicating that the CPU is working at twice the speed of the system board. In our example, the CPU works at a speed of 50 MHz, while the system board runs at 25 MHz. So, as information travels out of the CPU and hits the system board, the data slows down to half the speed.

The same could be said for the 80486DX2-66. The CPU works at a speed of 66 MHz, while the system board runs at 33 MHz.

DX4
The DX4 model works exactly the same way as the clock double, only the clock double is actually a clock triple. In other words, the CPU works at three times the speed of the system board. So, why call it a “DX4” if it’s really a clock triple? Because one extra enhancement, other than the clock triple, was added to the DX4: 8K more of L1 cache to the chip. The idea is that our clock triple plus the extra 8K of cache memory gives us a DX4 chip: (clock)3 plus 8(K) equals (DX)4.

All 80486 chips—except the DX4 chips, which have 16K of L1 cache—have 8K of L1 cache. Also, all processors created after the 80486 chip use L1 cache, though they may differ in the actual amount. This becomes one of the selling points of the different processors.

80486SX
Two years after the success of the 80486DX chip, Intel decided to market a lowerend 80486 chip. This new chip, released in 1991, was called the 80486SX.

What did the 80486SX chip have that made it so special? Or maybe a better question to ask is: What didn’t it have? The 80486SX chip was a full-blown 80486DX with the integrated math co-processor disabled. This time, Intel was trying to attract customers who could not justify the price of the more functional chip, so Intel simply downgraded the 80486DX, marketing it as a different chip and selling it at a lower price.

Because the 80486SX chip did not have an integrated and functioning math co-processor, there was a place on the system board to add a math co-processor chip (the 80487SX).

Suppose the 80486DX chip sold for $900 and the 80486SX chip sold for $700.

Assume also that I’m not using the computer for a lot of large mathematical calculations or graphics applications. In this case, I don’t really need a math co-processor integrated into my CPU. I could save myself $200 by buying the system with the 80486SX chip. However, as time goes on, I may start using the computer more and more, playing with a lot of graphics applications. My best friend (let’s call him Dan),
tells me that I would get a performance increase by purchasing a math co-processor for my system. It just so happens that the math co-processor that I have to purchase, the 80487SX, is on sale for $400. I purchase the 80487SX chip and place it in the empty socket by my CPU on the system board When I place the 80487SX chip on the system board in the math co-processor socket, it disables the 80486SX entirely so that my system is just using the 80487SX chip. So, I spent $700 on an 80486SX chip (that gets disabled) and $400 on an 80487SX (which is really the 80486DX) chip—a total of $1100—when I could have made the original purchase of $900 and obtained the same system in the long run.

Popular Intel Processors

In this section, we will provide an overview of some new terms, such as voltage, transistors, and sockets. We will also discuss the Pentium class processors and their characteristics, including data bus, address bus, registers, and any new technologies.

Characteristics, Voltage
CPU voltage and transistor integration One important CPU characteristic that you have to watch for when upgrading your processor is the voltage the processor requires. The voltage is the power the processor
draws from the main system board, which it receives originally from the power supply.

A processor is designed to run at a certain voltage. You need to ensure that the system board you are placing the processor into is providing that voltage. If a system board supports more than one voltage, you can change a jumper on the system board—which will then control the voltage used by the processor.

Processors are made up of thousands, even millions, of transistors. A transistor acts as a switch, either permitting or prohibiting the flow of current. If current is allowed to flow through the transistor, then some form of result is generated. If the current is not allowed to flow through the transistor, a different result is generated. How does the switch get turned on to allow the flow of current? The answer is from an input device, such as the keyboard. The action of pressing keys on a keyboard sends a positive charge to the transistor to turn on the switch.

Table 2-1 lists some of the popular processors, their required voltage, and the number of transistors used to build the logic of that particular CPU.

Table 2-1 lists some of the popular processors, their required voltage, and the number of transistors used to build the logic of that particular CPU.

Socket
Intel decided to develop a new standard for upgrading a processor on system boards, beginning with the 80486 chips and continuing with the Pentium class processors.

This standard was processor sockets. A processor socket is a socket designed to hold a specific processor chip with the appropriate number of pins.

This enabled Intel to develop new chips with compatibility in mind. They could design a new chip for an old socket so that customers could update their computers by dropping the new processor in the compatible socket.

Popular Pentium processors supported mainly Socket 5 with 320 pins or Socket 7 with 321 pins. Thus, to add a Pentium processor to a system board, you would have to find out what socket existed on that board, then purchase a CPU that would fit in that socket. You would also have to remember to match the voltage of the board to the voltage required by the CPU. Figure 2-5 will help you identify a CPU socket in your system.

Table 2-2 lists the different types of sockets and the processors that support them. It also shows the number of pins associated with the different types of sockets.

Pages: 1 2