In the last few years, changes have had a profound influence on the problems addressed by compiler designers. First, the proliferation in machine architectures has necessitated the fine-tuning of compiler back ends to exploit features such as multiple memory banks, pipelines and clustered architectures. These features provide potential for compilers to play a vital role in the drive for improved performance. Areas in which such possibilities can be exploited are, for example, speculative execution, software pipelining, instruction scheduling and dynamic compilation. Another trend that is continuously explored is the generation of parts of the compiler from specifications. Such techniques are well established for the front end, but still in the experimental stages for the problems of code optimization and code generation. Optimizers in compilers are relying on newer intermediate forms like static single assignment (SSA) form. The growing use of object-oriented languages has brought into focus the need for optimizations that are special to such languages. Algorithms for program slicing and shape analysis have made it possible to gather detailed information about a program that can be used in tasks such as debugging, testing, scheduling and parallelization.

The Compiler Design Handbook: Optimizations and Machine Code Generation addresses some of these issues listed that are of central concern to compiler designers. However, it does not attempt to be an encyclopedia of optimization and code generation. Several topics could not be included mainly because of space limitations. Notable among omissions are just-in-time (JIT) compilation, implementation of parallel languages, interpreters, intermediate languages, compilation of hardware description languages, optimization for memory hierarchies, constraint programming languages and their compilation, functional programming languages and their compilation, code generation using rewriting systems, peephole optimizers and runtime organizations. The topics included are of relevance mainly to the compilation of imperative languages.

The material presented in this volume is addressed to readers having a sound background in traditional compiler design. In particular, we assume that the reader is familiar with parsing, semantic analysis, intermediate code generation, elements of data flowanalysis, code improvement techniques and principles of machine code generation. We hope that the material presented here can be useful to graduate students, compiler designers in industry and researchers in the area.