The Art of Assembly Language

Overview

This chapter discusses the implementation of a CPU's instruction set. Although the choice of a given instruction set (for example, the 80x86 instruction set) is usually beyond the control of a software engineer, understanding the choices a hardware design engineer has to make when designing a CPU's instruction set can definitely help you write better code.

CPU instruction sets contain several trade-offs based on assumptions that computer architects make concerning the way software engineers write code. If the machine instructions you choose match the assumptions the CPU's designers have made, then your code will probably run faster and require fewer machine resources. Conversely, if your code violates the assumptions the hardware engineers have made, chances are pretty good it will not perform as well as it otherwise could.

Although studying the instruction set may seem like a task suited only to assembly language programmers, even high-level language programmers should understand the design of their CPU's instruction set. After all, every high-level language statement maps to some sequence of machine instructions. Indeed, studying generic instruction set design is probably more important to high-level language programmers than it is to assembly programmers (who should study the specific instruction set they are using), as the general concepts are portable across architectures. Therefore, even if you don't ever intend to write software using assembly language, it's important to understand how the underlying machine instructions work and how they were designed in the first place.