This paper describes and evaluates the profile-based optimizations in the Compaq C compiler tool chain for Alpha. The optimizations include superblock formation, inlining, commando loop optimization, register allocation, code layout, and switch statement optimization. The optimizations either are extensions of classical optimizations or are restructuring transformations that enable classical optimizations. Profile-based optimization is highly effective, achieving a 17% speedup over aggressive classical optimization on the SPECInt95 benchmarks. Inlining contributes the most performance and code layout, superblock formation, and loop restructuring are also important.
Robert S. Cohn, P. Geoffrey Lowney