Evolution of Computer Processors
Intel 8086 and 8088:
In 1978, Intel introduced the 8086 microprocessor, which had a clock speed of 4.77MHz. The clock speed of a processor is the speed at which the CPU operates. Clock speeds are rated in megahertz or millions of electronic cycles per second. A computer rated at 5MHz has five million processing cycles per second. The more cycles per second computer support, the more instructions it can execute. Remember that most instructions, because they include many processing steps, including memory transfers, ALU operations, etc., require more than a single CPU cycle to complete.
The 8086 was capable of running about 0.33 MIPS (millions of instructions per second). MIPS is a standard used to measure the processing power of a processor. The 8086 processor, which included 29,000 transistors was a 16-bit processor (its data bus was 16-its wide) and had an address bus of 20-bits. It could address 1MB (megabyte) of memory, which was an incredible amount at the time. The 8086 was not a popular choice for PCs, but it did create a baseline for all future Intel 80×86 processors.
Computer manufacturers largely skipped over the next Intel microprocessor generation, the 80186, to adopt the Intel 80286. IBM chose the 80286 for its next PC release, the PC AT. The 286 processor, as it was commonly known, was released in early 1982 with a 6MHz clock speed, 134,000 transistors, and nearly three times the power of the 8086 with 0.9 MIPS. The 286 was able to address 16MB of memory with its address bus expanded to 24 bits. Its data bus was also backwards compatible with the 8086’s original 16 bits. Later 286 versions had clock speeds of 10MHz and 12MHz.
Intel 80386, 80386DX, and 80386SX:
Intel released the 16MHz 80386, commonly called the 386, in 1985. The 386 microprocessor was a full 32-bit processor packaged in a 132-pin PGA package. The 386 had 275,000 transistors and had the clock speed to support over 5 MIPS. While it had a 32-bit mode, which meant it was able to move data in bytes, 16-bit words, or 32-bit double words (or dwords). Two features provided by the 386 were improved virtual memory capabilities that allowed large amounts of memory to be temporarily stored on the hard disk and instruction pipelining, a process that preloads and pre-evaluates complex instructions, which results in faster processing speeds. The Intel 386 had versions with clock speeds ranging from 16MHz to 33MHz. However, other manufacturers, specifically AMD and Cyrix had competing processor versions with speeds up to 40MHz.
The first of the 386 processors introduced by Intel was the 386DX. The 386 was a true 32-bit processor and included 32-bit internal registers, a 32-bit internal data bus, and a 32-bit external data bus. It was built with a new technology called VLSI (Very Large Scale Integration) with 275,000 transistors. The 386 used less power than its predecessors, including the 8086, because it was constructed of CMOS (Complementary Metal Oxide Semiconductor) materials, a way of manufacturing transistors that reduces the amount of power required when idle.
The 386DX could address up to 4 gigabits of system memory, but its built-in virtual memory management (VMM) system allowed VMM-enabled software to access a virtual memory store of the equivalent of 64 terabytes (TB) of memory (a terabyte is a trillion bytes of memory).
Intel also released a lower-cost version of the 386 processor called the 386SX. The primary difference between the 386SX and the 386DX was that the SX model had only a 16-bit external data bus and a 24-bit address bus, which made it backwards compatible with the 286 processor. The 386SX was released to fill a market need, which was a lower-priced processor with the power of the 386DX at the cost of the 286.
Intel released the 386SL, which was a 20MHz processor, in 1990. The SL version was similar to the 386SX, but it was specifically designed for portable computers, featuring improved power management functions.
Intel 80486DX and SX:
Processors did not break the one million-transistor barrier until Intel released the 25MHz 486DX microprocessor in early 1989. This processor had over 1.2 million transistors and generated 20 MIPS. This processor also introduced several innovations, including the inclusion of processor cache (Level 1 cache) on the processor chip, the introduction of burst-mode memory access, and for the first time, an integrated math coprocessor. Before the 486DX, if a user wished to speed up the math functions on a PC, a separate math coprocessor had to be installed. The 486 was packaged in a 168-pin Ceramic Pin Grid Assembly (CPGA) package that required a processor-mounted fan to cool it. Processors before the 486DX had relied on the system fan in the PC’s power supply for cooling.
The next model of the 486 was released in 1992 as the 80486DX2. The ‘2’ designation referred to a technique called overclocking that allows a processor’s clock speed to be doubled. The DX2 was first released as a 50MHz version, which doubled the DX’s 25MHz bus speed, and was followed by a 66MHz version (33MHz bus times 2). The 486DX4 was a product of overclocking as well. The 486 25MHz and 33MHz processors were overclocked to produce triple their normal clock speeds. This resulted in the DX4 processor being available with 75MHz (25MHz times 3) and 100MHz (approximately 33MHz times 3) clock speeds.
Intel did not have much competition until Advanced Micro Devices (AMD) released its 75MHz 5×86 microprocessor. The TheAMD5x86 processors were compatible with 486 motherboards but had similar power to the early Pentium processors.
The Cyrix 5×86, also known as the M1SE, was intended to compete with the Intel 486 with which it was socket compatible. Like the AMD 5×86, the Cyrix processor was able to compete with early Pentium processors.
Although it was known in its early development as the 80586 in 1992 by Intel had discovered that model numbers. So, instead of the number, they used a trademark name, Pentium, for their next processor. This new processor included many new features, including separate 8-bit caches for data and instructions and a very fast FPU. The Pentium kept the 32-bit address bus of the 486 but added a 64-bit data bus. It also included superscalar architecture, which is a processor technology that allows more than one instruction to be executed in a single clock cycle. The clock speeds of the Pentium processor ranged from 60MHz to 200MHz.
The next version of the Pentium processor was the Pentium MMX, which had clock speeds from 166MHz to 233MHz. This version of the Pentium processor added MMX (Multimedia Extensions) technology to the Pentium along with some improved internal clock speeds. MMX technology is a set of instructions that uses matrix math (another meaning for MMX) to support graphic compression and decompression algorithms (such as JPEG, GIF, and MPEG) and 3D graphic renderings. MMX allows the FPU to act on several pieces of data simultaneously through a process called SIMD (single instruction multiple data).
Cyrix 6×86 Processors:
Cyrix, which is now VIA Cyrix, produced a family of Pentium clone processors that were designated as the 6×86-P series. The ‘P’ value in the model name was a performance rating indicator. The 6×86-P200 indicated that the Cyrix processor with that model number had the performance equivalent of a Pentium 200MHz processor. Cyrix produced models ranging from its 6×86-P120 to the 6×86-P200. The 6×86-P series had overheating problems as well as some incompatibility issues, which prompted Cyrix to produce a low-power, low-temperature version, the 6x86L.
Intel Pentium Pro:
The Pentium Pro, the next in the Pentium line, was developed as a network server processor. To support the demands put on a network server, the Pentium Pro has 1 megabit of advanced second level (L2) cache. The Pentium Pro 200MHz version was specially designed to support 32-bit network operating systems, such as Windows NT, and to be used in configurations of one, two, or four processors.
The Pentium II:
The Pentium II is the Intel Pentium Pro with MMX technology added. The PII, as it is commonly referred to, is available with clock speeds of 233MHz, 266MHz, and 300MHz. It is excellent for multimedia reproduction that requires support for full-motion video and 3D images.
Developed from use in desktop and portable computers, the Celeron microprocessor is the low-cost model of the Pentium II processor series. It features two choices for mountings, the Pentium II’s Slot 1 and a socket style named after the number of pins in use, Socket 370, shown in Figure 3-16. The Celeron is released in versions with
clock speeds of 333MHz to 500MHz, with newer models, built on the Pentium III core, to offer clock speeds of 566MHz or faster.
It depicts the Pentium II Xeon processor, which is the successor to the Pentium Pro as a network server processor. To enhance its ability as a network server microprocessor, the Xeon features a range of L2 cache size choices, ranging from 512K, 1MB, and 2MB. The Xeon is capable of addressing and caching up to 64GB of memory with its 36-bit memory address bus. The PII Xeon can be configured with four to eight CPUs in one server.
Developed to compete with the Pentium MMX, theAMDK6 outperforms it in speed and price. It is available in 166MHz, 200MHz, 233MHz, 266MHz versions, and a 300MHz model that mounts in a Super 7 socket.
Cyrix 6x86MX and Cyrix III:
Also known as the MII, the Cyrix 6x86MX processor contains an MMX instruction set. The 6x86MXhas a performance rating (PR) of P-166 to P-366. Cyrix, which is now owned by VIA Technologies, now also offers the 6×86 in a P-433 version as well. The Cyrix III microprocessor runs at clock speeds of 433, 466, 500, and 533MHz. It supports Intel’s MMX and 3DNow, the AMD equivalent of multimedia extensions. The Cyrix III processor, technically the VIA Cyrix III, is the equivalent of an Intel Pentium II Celeron processor.
AMD K6-2 and K6-III Processors:
To compete with theMMXtechnology of the Pentium processors, the AMD K6-2 processor, has an added 3DNow, a set of 3D graphic instructions that extend the MMX instructions that have already been incorporated into the K6 design. The K6-2 processors are available with clock speeds from 266MHz to 550MHz. A newer model of the AMD K6 line is the K6-2+, which has an additional L2 cache on the processor chip and some new power control features. The K6-III processor features 256K of L2 cache and clock speeds from 400MHz to 600MHz. A newer model, the K6-III+ includes 1MB of cache and runs at the clock speeds of the K6-III.
Intel Pentium III:
The Pentium III processor features 9.5 million transistors, a 32K L1 cache, and 512K of L2 cache. The Pentium III is available with clock speeds of 450MHz to 1GHz and is packaged in a second-generation single-edge connector package called the SECC2. The SECC2 package, which fits into the Slot 1 bus, is designed to conduct and remove heat better than earlier single-edge packages.
The honour of having the new powerhouse processor may have moved to AMD with the release of its 1GHz AMD Athlon. The Athlon boasts 22 million transistors, support for Intel’s MMX and AMD’s 3DNow, and improved FPU functions. It also has the power to decode more instructions simultaneously than the Pentium III, with 256KB of L2 cache and 128KB of L1 cache on the chip. The Athlon is plug-compatible with the Slot 1 connector, but it is designed for AMD’s Slot A bus, which runs at bus speeds of 200MHz to 400MHz. A derivative of the Athlon is the AMD Duron. The TheAMDDuron processor is designed for general computing, including business, home user, and portable applications. The Duron processor is available at clock speeds of 600MHz, 650MHz, and 700MHz.
Intel Pentium 4:
The latest of the Intel processors is the Intel Pentium 4 which is available with processor speeds of 1.3 to 1.5 GHz (gigahertz). The Pentium 4 uses a new proprietary micro-architecture called Net-Burst, which features a 400 MHz system bus, advanced on-board caching, enhanced floating-point math and multimedia support, and hyper pipelining technology. The Pentium 4 also supports the use of dual-channel RDRAM.