History

A quick history lesson. The 80x86 family started with the 8088 and 8086 (16-bit) processors. These were 16-bit processors as the general- purpose register was 16 bits in size . The 8088, however, had an 8-bit databus. The general-purpose registers had two addressable components , i.e., AX, with the upper 8 bits as AH and the lower 8 bits as AL. Along came the 80186 and 80286. With these came the first foundations of segment:offset addressing. These had a 16-bit segment and a 16-bit offset, which were used to generate a 20-bit address. Thus, there were 12 bits of overlap between the two registers. The segment was used to address a 16-byte block within 1 MB of memory.

BaseAddress = (Segment — 16) + offset

The 80386 was the first 32-bit processor to break the 1 MB memory barrier . Additional instructions were added, and the segment:offset registers became selector:offsets. There then became a differentiation between Real Mode using 16-bit addressing and Protected Mode, which used the new 32-bit technology. Then the processor clone wars started between manufacturers. The 80486 came along with additional instructions. The Pentium series came next with the introduction of multimedia instructions known as SIMD with the MMX registers. There were different flavors, different instructions it was very convoluted and confusing. AMD introduced 3DNow! This is about when the book you are now reading was started. Then came Katmai by Intel, now known as the SSE new instruction set. AMD countered, Intel countered, back and forth. Other manufacturers either sunk at battle or pushed onto the shores of an embedded island. Then the technology was forked by Intel (but we are ignoring the Itanium series in this book). Somewhere along the way AMD and Intel each came out with 64-bit processors and seem to have reached a tentative truce.

Incidentally, the FPU (floating-point unit) was an optional second coprocessor originally known as the 8087 and subsequently the 80187, 80287, 80387, and then the 80487. Back then, floating-point was emulated in software as an alternative if no hardware was present. The FPU grew up with the CPU and eventually become an internal component of the 80486DX CPU. The 80486SX still had to rely upon a coprocessor.

History 

Rumor or fact? There are lots of stories of processors, faulty FPU chips, math bugs , etc. One story has to do with the 80486SX. Rumor has it that if the Intel 80486DX's FPU failed testing or cheaper SX processors were in short supply, then connections to the FPU were zapped within the 80486DX to transform it into an 80486SX processor. ( Rumor or fact? )

This is as deep into Real Mode as this book is going to get. If you need additional information, please investigate specific documentation from the 80x86 processor manufacturer for your embedded application as Real Mode is fading into obscurity. The same applies to the need for selector manipulation as it can get pretty complicated, but this is only needed by operating system programmers for pre-64 bit processors, and that is not what this book is about. It is written to help those programmers but primarily written for application developers for 32- and 64-bit operating systems. System-level programming pretty much requires its own book about this size.

Many interesting things occurred during the introduction of the Protected and Private Modes. C compilers had to be selected for various code vs. data models.

Model	Code	Data
Huge	32-bit	32-bit
Small	32-bit	16-bit
Large	16-bit	32-bit
Tiny	16-bit	16-bit