💾 Archived View for gemini.spam.works › mirrors › textfiles › uploads › intel.txt captured on 2023-01-29 at 12:54:32.
⬅️ Previous capture (2020-10-31)
-=-=-=-=-=-=-
THE INTEL MICROPROCESSOR ======================== This article contains a few, very mildly edited pages from the Exam Cram Server+ book by Deborah Haralson and Jeff Haralson (ISBN 1-58880-106-3). This document was typed up by Jonathan Smith. If you are interested in learning more, I suggest you buy the book And now, on with the article... BOOK: Exam Cram Server+ AUTHORS: Deborah Haralson Jeff Haralson ISBN: 1-58880-106-3 PAGES: 71 - 78 DATE: 4:30 PM 8/20/2002 Intel 8080 ---------- The 8080 was first commercfially produced in 1974. It was the processor in the Altair 8800 home computers. The Altair name comes from a Star Trek planet, which let the computer where it has never gone before. This computer was placed on the market by the MITS calculator company and was sold in kit form. Bill Gates and Paul Allen wrote the BASIC programming language for MITS, which was used on the Altair computer, and it started Bill's meteoric rise to his present level of success. Intel 8086 and 8088 ------------------- The Intel 8086 was released on June 6, 1978. It was Intel's first 16-bit microprocessor. It was designed to be fast and powerful. The 8086 boasted a 4.77 to 10 MHz clock speed, 16- bit registers, and an external 16-bit data bus. It also had a 20-bit address bus to address 1MB of RAM. When IBM entered the computer business, the 8086 was considered too expensive, and oddly enough, too powerful to fulfill "Big Blue's" vision of a computer. The result was that Intel essentially "dumbed down" the 8086 processor and called it the 8088. At this point, some of the factors that affected the overall speed of a computer can really be identified. The 8086 and the 8088 could run the exact same program and could be set to the exact same clock speed. The 16-bit data bus would let the 8086 import and export the data to the chip at twice the speed of the 8088. Intel 80286 ----------- The Intel 80286, commonly called the 286, started as a 6 MHz speed demon. Soon Intel pushed the envelope by upping the clock speed to 10, 12.5, and eventually 20 MHz. To add to the power, the 286 came with a 24-bit address path that allowed for a whopping 16MB of addressable memory. This chip was the heart of the IBM AT computer. The 286 also offered two modes of operations: real mode and protected mode. These two modes of operation made the package operate like two separate chips in one package. Real mode operations essentially acted like an 8086 microprocessor. The computer could run the 8088 and 8086 programs without requiring any modifcation to the programs. Every system instruction was available and fully functional when the 286 was operating in real mode. This level of compatibility came at a price. The 80286 microprocessor essentially lobotomized itself, cutting back the addressable memory to 1MB. This made the old programs, not to mention the owner of the aformentioned programs, happy, but limited the full flexibility and functionality of the machine. Protected mode operations allowed specially programmed applications many advantages over the older real mode programs. Protected mode applications were "protected" in that they were limited to writing only to their assigned memory. This helped to prevent applications from stealing or overwriting areas of memory that was originally designated to a separate application. The protected mode helped add stability and reliability to a machine that was raplidly becoming indispensable in the business environment. The protected mode of operations also heralded in the invention of virtual memory, which allowed applications access to far more than the 16MB of physical memory. This meant that with up to 1GB of memory, larger applications coul be loaded simultaneously and would be protected from fighting each other for memory. Bill Gates' MS-DOS had a very tough time addressing this huge amount of available storage, so the 286 saw the rise of additional operating systems like Microsoft Windows, SCO Unix, and IBM's OS/2. The more powerful, true multitasking operating systems were still in the development stage, and the 286 was obsolete by the time some of the "beefier" operating systems like Windows 95, Windows NT, and Windows 2000 were introduced. One major disadvantage to the two modes in the 80286 microprocessor was that the computer needed to be rebooted to change the operational mode. This led to a gradual shift from real mode applications to those natively written to take advantage of the many features of the 286. Intel 80386 ----------- Intel changed the world of computing when it introduced the 80386 microprocessor in June 16, 1985. The 80386 was the first true 32-bit processor. All of the support circutry, the external bus, the address bus, and the registers were a full 32 bits in width. The chip originally shipped with a clock speed of 12 or 16 MHz and could address 4GB of physical RAM and 65TB of virtual memory. One of the biggest advantages of the 386 was that it could switch between real and protected modes of operation without requiring a reboot. The 386 microprocessor introduced an additional operation mode known as virtual real mode. This made it possible for the 386 processor to host virtual machines that enabled more than one application to actually run at the same time. Intel eventually produced 80386 processors that could run at 25 and 33 MHz. Clone manufacturers weren't far behind as chief Intel rival, AMD, produced an 80386-compatible that ran at an unprecedented 40 MHz. Intel charged a pretty menny for the flagship of their processor fleet. The big 80386 was out of reach for many businesses, so Intel decided to offer a smaller version that was designed to increase sales without dropping the price of the newest kid on the block. In June of 1988, Intel released the 386SX. 386DX and 386SX --------------- The 386DX (Dual word eXternal) was the full-blown 80386 in its entire 32-bit glory. The 386SX (Single word eXternal) was created by disabling half of the 32 bits of the data bus to a single word or two bytes. This drop it to the 16-bit, 286 levels. Intel also dropped the memory addressing to a miserly 24-bits, which limited the SX machines to only 16MB of RAM. The one advantage of the 386SX was that it could be used to upgrade existing 16-bit motherboards. This extended the life of users' machines and also temporarily ensured that motherboard manufacturers had a few more months to unload their inventory before it all became paperweights. It was in the 80386-80486 years that Microsoft first produced the Windows operating system, and IBM released OS/2. NetWare ruled the server market, and it was only toward the end of 1990 that Microsoft started beta testing their first NOS, which would be known as Windows NT. Meanwhile, Unix was being used in corporate environments since the 8088 became available. Intel 80486 ----------- The 486 processor was roughly twice as fast as a 386 running at the same clock speed. The four factors that contributed to this fact were: - The 486 only took two cycles to execute an instruction, whereas the 386 took 4.5 cycles. This meant that the 486 could perform 2.5 operations for every one that the 386 could crank out. - Intel moved the cache on the chip creating the first Level 1 cache available for mass production. This built-in cache had an average hit ration of 90 to 95 percent. This meant that when the processor needed data from memory, it was right there most of the time. This convenience almost eliminated the need for any wait states on the main CPU. - Built-in math coprocessors were included in most versions of the 486. The coprocessor that greatly enhanced the mathematical prowess of the older machines was brought onboard the main processing chip. This meant no more trips to the bus to request information. - Memory could be accessed using burst-mode memory cycles. This meant that the normal two-cycle memory access would proposition the circutry for the next data transfer. Once the inital 32 bits were transferred in two cycles, the next 12 bytes could be had with only one clock cycle for each 32 bits, or 4 bytes. That meant up to 16 bytes of additional data could be transferred in as little as five cycles instead of the normal 8+ cycles required by the 386. The first 80486DX processor was introduced by Intel on April 10, 1998, and the first systems started to appear the following year. The inital chip ran at 25 MHz, but remember, this outperformed a compatible 80386 50 MHz system. 486SX ----- Intel once again put the brakes on its premier processor to provide consumers with a cheaper, but slower, version of its flagship processor. The company accomplished this by removing the math coprocessor. Intel initially just turned the coprocessor off. This thrilled many hardware hackers as they discovered they could turn it back on and get a more expensive chip at a bargain. Intel fought back by actually removing the circuitry. Intel did offer an upgrade from the 486SX to the 486DX by installing the OverDrive processor to restore the missing circuits, but at the cost of reduced proximity to the main CPU circuitry. clock Doubling -------------- Intel answered the increasing demands of speed by introducing the DX2 and DX4 processors. These machines ran at clock speeds of up to 100 MHz by taking the CPU and configuring it to run internally at double the external clock speed. This would take a normal 486DX running at 33 MHz and produce a 66 MHz speed demon. The DX4 would triple the external clock, making the same 33 MHz clock runs the chip at 100 MHz. Keep in mind that this only affected the internal processes. The internal register transfers and computer would run at 100 MHz, but external calls for memory would only run at the normal 33 MHz rate. The DX4 also offered increasing amounts of internal onboard cache. Pentium Processors ------------------ Because you cannot copyright numbers, Intel decided to move to Latin to name its processors. Pentium is derived from the Latin word for five as the Pentium is the new 80586 family of microprocessors. The first, or Pentium I, was introduced on October 19, 1992, with the first chip shipping on March 22 the following year. By this time, many corporations were migrating to client/server solutions that lived outside of the mainframe enviromnents with which most people were familiar. Windows NT Server was in full swing and competing vigorously with NetWare for an increasing market share. Unix had become more popular, however, not as mainstream as the other two network operating systems. The Pentium was the first chip to offer superscalar technology. This meant that the Pentium featured twin data paths, which allowed the processor to execute two instructions at the same time. These two busses were called U and V. The U pipeline was designed to process the entire instruction set of the Pentium. The V pipeline offered a subset of instructions. This allowed the processor to divide the instructional tasks into discrete operations that were then shared between the two pipelines. Software that is designed to take advantage of the multiple paths is known as multithreaded applications. Windows NT was one of the first operating systems to take advantage of this technology. The Pentium also boasted a 64-bit data bus with a built-in math coprocessor. These machines also come with twin internal 8KB caches. One cache is used for data and the other is designed to store programming instructions. The first generation of Pentium processors did not double the external clock, whereas the econd generation had a clock multiplier of 1.5 or 3 depending on the model. Thus, the Pentiums ran from 33 to 200 MHz. Pentium Pro ----------- In 1995, the Pentium Pro was introduced. This processor was aimed at the higher-end servers and workstations. The Pentium Pro added a second memory-caching chip. This was the first Pentium that supported the Reduced Instrction Set Computer (RISC) processor. All of the other Pentiums only ran the Complex Instruction Set Computer (CISC) processor. Beacause a Pentium Pro was really a Pentium at heart, it also incorporated a CISC to RISC translator. Additional features like three-way superscalar execution and dynamic execution made this chip a powerhouse. Pentium MMX ----------- More and more applications began to rely on high-speed graphics. This manipulation was taxing even the fastest Pentium processors. To address this "need for speed" Intel released the third generation Pentium in 1997, which included a pipelined Multi-Media extension (MMX) chip. This provided a dedicated processor designed for Single Instruction Multiple Data (SIMD), which lets a single command operate on different sets of data. The SIMD feature of the MMX chip transformed the graphics industry, particularly in the games arena. In order to have the "latest and greatest," you had to have MMX. To upgrade to MMX, you needed to have a Pentium motherboard that matched the "socket 7" specification. The 233 MHz MMX essentially ended the original Pentium line in June of 1997. Pentium II ---------- In 1997, the first Pentium with MMX built-in to the chip was introduced. This was the Pentium II. Thie chip added several features that revolutionized the processor market. These features included: - 512KB L2 cache in addition to the separate L1 caches that the Pentium already supported. - Error Correction Coding (ECC) of the L2 Cache bus, which ensured that the data being transported was checked and correct. - Single-Edge connector (SEC) packaging, which protected the internal circuitry and made the installation much easier and less prone to error. - Multiple branch prediction, which "looked-ahead" in the processing cycle and prefetched information along decision branches. This meant that when a decision was reached, the requested result was already loaded and ready. - Dual Independent Bus (DIB) design, which isolated the system bus from the cache bus. It allowed complete synchronization between the CPU and the various caches without being interrupted by any system calls. - Speculative execution, which executed operations based on "best-guess" results. This kept the pipes full and boosted system performance. - Data-flow analysis where the processor actually reorders the order of execution to ensure speedy processing. This required the CPU to actually analyze the instructional relationships between the different commands and avoid breaking the order too radically. Intel Celeron ------------- The consumer market was finally introduced to the speed increase of the microprocessor with the SX designation. The Pentium II was oriented more towards the business market, as the businesses were the only ones with the pockets deep enough to purchase Intel's powerhouse. To keep the revenue stream alive and provide the home consumer with more power, Intel introduced the Celeron. The Celeron was a pretty decent processor in comparison to other cloned processors on the market. It supported the MMX technology that was such a huge success in the home entertainment market. It also provided some of the Pentium II's sexier features, like multiple branch prediction, data-flow analysis, and speculative execution. Xeon ---- The Xeon first became available in 1998 and soon became "the chip" for high-end servers. This chip supported a clock speed that ranged from 450 MHz to 1 GHz. The multiprocessing capabilites made this the prime choice for application servers where raw processing power is king. The Xeon designation was applied to the Pentium II and Pentium III chips to establish a CPU that was specifically designed for the server environment. Pentium III ----------- The Pentium III added streaming SIMD enhancements for mathematical processing and graphics rendering. It also provided additional support for power control, which allowed the processor to conserve power and generate less heat when in an idle state. Pentium 4 --------- The Intel Pentium 4 was designed for pure speed. The first Pentium boasted two pipelines, but the Pentium 4 has 20. It is a part of the Intel Netburst microarchitecture. Additional features include: - 400 MHz system bus that delivers 3.2GB per second to the memory controller. - Internal Arithmetic Logic Units that run at twice the clock speed. - Advanced dynamic execution that enhances the prediction of execution flow. - 128-bit, 144 instruction, streaming SIMD design for crisp, and more imporant, really fast graphics.