In 1981 a new processor chip was introduced by Intel, the 802868(often just called the 286). This offers an external data bus of 16 bits and a wider address bus of 24 bits. The 24 bits of the address bus potentially gives access to 224bytes of memory, that is, to 16 Mbyte, and the resulting memory above the 1 Mbyte region became known as Extended Memory. The 80286 chip is fully compatible with the 8086 and 8088 chips and it adds a new mode of operation calledProtected Mode, which enables access to this extended memory. The original mode of operation then became known asReal Mode.
IBM developed the AT(Advanced Technology) architecture, based on this new chip, because it was fully compatible with the 8088 and all software that ran on the earlier systems should be able to run on the AT. The new chip was much faster than the 8088, and the original AT, which first ran at 6 MHz, was some five times faster than the 8088 running at 4.77 MHz. In practice, the AT systems using the 80286 were seen, from a user viewpoint, to be little more than very fast XTs, though there were some other important changes in the architecture. For example, combined 8 bit and 16 bit ISA expansion slots were introduced to take advantage of the new 16 bit data bus, without losing the legacy 8 bit systems. The IRQ and DMA channel allocations set by the system were changed and an increased range of both IRQ and DMA channels was made available. However, because the operating system and much of the available software was unable to use the new protected mode of the chip, useful advantage could not be taken of the additional memory access that the system could provide.
The 80386 Chip
When Intel introduced the 80386 (often just called the 386) in 1985, it rightly promised a vast improvement in the performance of PCs. The chip can operate at clock speeds from 16 MHz up to 33 MHz. With its 32 bit address bus it provides an address space of 4 Gbyte, and with a 32 bit internal and a 32 bit external data bus it can transfer 4 bytes of data at a time. It is fully backward compatible with the 8086, 8088 and the 80286 processors and it adds yet a further mode to the protected and real modes of its predecessor, that ofVirtual 8086 Mode, which is sometimes called Virtual Real Mode. This new mode allows a control program to create what are in effect a series of “virtual” 8086 systems on the one PC, each of which is operating in a protected mode which prevents it from interfering with the others.
Many variations of the 386 chip exist. The original 80386 became known as the 386DX when Intel introduced a cheaper version of the chip known as the 386SX. Whilst retaining the internal 32 bit data bus, the external data bus on the 386SX was reduced to 16 bits. This move took advantage of the large number of 16 bit ISA
8 Before this Intel had produced the 80186 and 80188 chips. These were very similar to their 8086 and 8088 predecessors, but incorporated on the same chip several others of the 8086 family of components. This provided for a reduced chip count over the original 8086 family.
cards, which by that time were in common use, and 386SX-based systems rapidly became popular entry level computers. A coprocessor, used to perform floating point arithmetic in hardware, thus improving the performance of scientific and engineering calculations, was produced for the 80386, and this was designated the 80387.
The 80486 Chip
The 80486, which is normally called the 486, is essentially an 80386 with an enhanced 80387 coprocessor built onto the chip. It also incorporates enhanced memory management, 8 kbyte of internalcachememory and a more efficient design. A 486 will run, for the same clock speed, about twice as fast as the equivalent 386, and the chip is fully compatible with the 8086, 8088, 80286 and 80386 processors. Initial chips ran at a clock rate of 25 MHz, but later ones were available in 33 MHz and 50 MHz versions. The 486 effectively created a revolution. Before its introduction,Graphical User Interfaces(GUIs), such as those used by Microsoft Windows, had not achieved widespread, popularity largely because of performance limitations, often charac- terized by the appearance of the dreaded hourglass icon. The 486 changed that, providing a performance that made GUIs more acceptable.
Many variations of the 486 chip also exist. The original 80486 chip became known as the 486DX when the cheaper 486SX was introduced. This 486SX is essentially a 486DX without the onboard maths coprocessor, and the early 486SXs were simply 486DXs with their maths coprocessor element disabled, although the chip later became established as a design in its own right. Intel also produced the 487SX as a separate maths coprocessor for those who wished subsequently to upgrade their 486SX systems. Motherboards of upgradeable systems were fitted with an extra socket into which this 487SX could be fitted. In practice, the 487SX is a fully functional 25 MHz 486DX, and when it is installed in its socket it completely disables the original 486SX via a new signal from one of its pins, and takes over all processor functions from the 486SX.
The big advance, based on this same idea, was the introduction of theOverDrive processor. This was before the widespread use of ZIF sockets, and the exercise of prising out a 169 pin processor chip and replacing it with another was not a recom- mended practice. Instead, the idea of fitting a new processor chip into a separate upgrade socket which automatically disabled the existing processor seemed like a very sensible strategy. The 486DX2 overdrive processor uses the same pinout as the 487SX and therefore can be fitted into the same upgrade socket. The 486DX2 achieves clock speed doubling by running at twice the speed of the motherboard clock. For a 33 MHz clock this means that the 486DX2 runs internally at 66 MHz. In a similar manner, the 486DX4 triples the clock speed.
None of these chips is being marketed today, although forensic computing analysts may still come across systems using them. The separate OverDrive socket idea has also now been dropped in favour of the ZIF socket. The original OverDrive socket is now called Socket 1 (see Table 4.1), and the current upgrade strategy is to remove the processor chip from its ZIF socket and replace it with the upgrade rather than to add a second chip which disables the first.