Introduction

Abstract

The book is organized as follows. The rest of this chapter describes the task of instruction selection (and why it is needed), establishes a common taxonomy, discusses common issues when comparing existing methods, and briefly covers the first publications on code generation. Chapters 2 through 5 then examine a fundamental principle to instruction selection: macro expansion, tree covering, DAG covering, and graph covering. Chapter 6 ends the book with conclusions through which I attempt to identify directions for future research