Preface
Purpose
This manual describes the floatingpoint software and hardware for the SPARC®, Intel®, and PowerPC^{TM} (processorbased system) architectures. It is primarily a reference manual designed to accompany Sun^{TM} language products.
Certain aspects of the IEEE Standard for Binary FloatingPoint Arithmetic are discussed in this manual. To learn about IEEE arithmetic, see the 18page Standard. See Appendix G, "References," on page 237, for a brief bibliography on IEEE arithmetic.
This guide describes the floatingpoint environment supported through a combination of software and hardware by the Solaris operating system running on SPARC, Intel, and PowerPC systems. This guide refers to compiler products that are components of the Sun WorkShop Compilers^{TM} 4.2 release.
Operating Environments
Sun WorkShop Compilers 4.2, described here, run under Solaris 2.x operating environment for SPARC, Intel, and PowerPC systems.
Audience
This manual is written for those who develop, maintain, and port mathematical and scientific applications or benchmarks. Before using this manual, you should be familiar with the programming language used (FORTRAN, C, etc.), dbx (the sourcelevel debugger), and the operating system commands and concepts.
Organization
 Chapter 1, "Introduction," introduces the floatingpoint environment.
 Chapter 2, "IEEE Arithmetic," describes the IEEE arithmetic model, IEEE formats, and underflow.
 Chapter 3, "The Math Libraries," describes the mathematics libraries provided with Sun WorkShop Compilers 4.2.
 Chapter 4, "Exceptions and Exception Handling," describes exceptions and shows how to detect, locate, and handle them.
 Appendix A, "Examples," contains example programs.
 Appendix B, "SPARC Behavior and Implementation," describes the floatingpoint hardware options for SPARC workstations.
 Appendix C, "Intel Behavior and Implementation," lists Intel and SPARC compatibility issues related to the floatingpoint units used in Intel systems.
 Appendix D, "PowerPC Behavior and Implementation," discusses features and compatibility issues specific to the floatingpoint arithmetic implemented in PowerPC systems.
 Appendix E, "What Every Computer Scientist Should Know About Floating Point Arithmetic," is an edited reprint of a paper by David Goldberg that is a tutorial on floatingpoint arithmetic.
 Appendix F, "Standards Compliance," discusses standards compliance.
 Appendix G, "References," includes a list of references and related documentation.
 "Glossary," contains a definition of terms.
The examples in this manual are in C and FORTRAN, but the concepts apply for either compiler on a SPARC, Intel, or PowerPC system.
The David Goldberg paper is now Appendix E, "What Every Computer Scientist Should Know About Floating Point Arithmetic," in this guide.
Typographic Convention
The following table describes the typographic conventions used in this book.
Shell Prompts in Command Examples
The following table shows the default system prompt and superuser prompt for the C shell, Bourne shell, and Korn shell.
