• Choose a category
• All
• Classic-Fiction

By Root 1389 0

that could operate directly on memory locations. For instance, you could add two numbers together and store the result back into memory with a single command. The Philco 2000, an early transistorized machine (1959), had an operation that could repeat a sequence of instructions until the contents of a counter was decremented to zero — very much like a do-loop. These were complex operations, even by today’s standards. However, both machines had a limited amount of memory — 32-K words. The less memory your program took up, the more you had available for data, and the less likely that you would have to resort to overlaying portions of the program on top of one another.

Complex instructions saved time, too. Almost every large computer following the IBM 704 had a memory system that was slower than its central processing unit (CPU). When a single instruction can perform several operations, the overall number of instructions retrieved from memory can be reduced. Minimizing the number of instructions was particularly important because, with few exceptions, the machines of the late 1950s were very sequential; not until the current instruction was completed did the computer initiate the process of going out to memory to get the next instruction.[80] By contrast, modern machines form something of a bucket brigade — passing instructions in from memory and figuring out what they do on the way — so there are fewer gaps in processing.

If the designers of early machines had had very fast and abundant instruction memory, sophisticated compilers, and the wherewithal to build the instruction “bucket brigade” — cheaply — they might have chosen to create machines with simple instruction sets. At the time, however, technological choices indicated that instructions should be powerful and thrifty with memory.

Beliefs About Complex Instruction Sets

So, given that the lot was cast in favor of complex instruction sets, computer architects had license to experiment with matching them to the intended purposes of the machines. For instance, the do-loop instruction on the Philco 2000 looked like a good companion for procedural languages like FORTRAN. Machine designers assumed that compiler writers could generate object programs using these powerful machine instructions, or possibly that the compiler could be eliminated, and that the machine could execute source code directly in hardware.

You can imagine how these ideas set the tone for product marketing. Up until the early 1980s, it was common practice to equate a bigger instruction set with a more powerful computer. When clock speeds were increasing by multiples, no increase in instruction set complexity could fetter a new model of computer enough so that there wasn’t still a tremendous net increase in speed. CISC machines kept getting faster, in spite of the increased operation complexity.

As it turned out, assembly language programmers used the complicated machine instructions, but compilers generally did not. It was difficult enough to get a compiler to recognize when a complicated instruction could be used, but the real problem was one of optimizations: verbatim translation of source constructs isn’t very efficient. An optimizing compiler works by simplifying and eliminating redundant computations. After a pass through an optimizing compiler, opportunities to use the complicated instructions tend to disappear.

Fundamental of RISC*

A RISC machine could have been built in 1960. (In fact, Seymour Cray built one in 1964 — the CDC 6600.) However, given the same costs of components, technical barriers, and even expectations for how computers would be used, you would probably still have chosen a CISC design — even with the benefit of hindsight.

The exact inspiration that led to developing high performance RISC microprocessors in the 1980s is a subject of some debate. Regardless of the motivation of the RISC designers, there were several obvious pressures that affected the development of RISC:

The number of transistors that could fit on a single chip was increasing. It was clear that one would