Optimizing C With Assembly Code

( Learn how to squeeze faster performance from existing programs. ( The diskette included with the book has replacements for common functions in the standard C library. Using this code, instead of the library that comes with commercial compilers, will make programs run faster. ( The diskette also provides a utility, TACHO.EXE, a "tachometer" that assesses the cycle times of executable program files. Optimizing C with Assembly Code shows how to make C programs run faster by strategically inserting brief inline assembly code. Many of the explanations about cycle times and pipeline stalls have never been published anywhere; not even by Intel. Although the book targets MS-DOS and Windows 3.x running on 486s, many of the improvements are also relevant for OS/2 or other 32-bit systems. Most of these optimizations are useful for 386s. The authors warn the reader if they are not useful for 386s. Additionally, the authors evaluate the optimizers of the three most popular C compilers. Borland C++, Microsoft Visual C++, & Symantec C++ are critically compared. These compilers are used as test compilers throughout the book, but the processing rules focus on methods common to all C compilers. The assembly routines in this book can improve processing by 15% to 50% over the commercial compiler libraries. Programmers who know how to profile their programs and provide efficient processing, and are looking for the last iota of speed in execution. No assembler experience is necessary, but the authors assume that the reader has a reference text for looking up instruction definitions, register names, etc. (One of the Appendices provides 80486 assembler instruction mnemonics and cycle times).