Abstract

elations and finding natural loops. This work is supported in part by an DARPA/NSF infrastructure grant (NDA904-97-C-0225) and an NSF Young Investigator award (CCR-9457779). We also gratefully acknowledge generous support of this research by Advanced Micro Devices, Digital Equipment, Hewlett-Packard, International Business Machines, Intel, and Microsoft. 1 Transformation. The library supports transformations of the CFG that change the underlying program. Many compiler optimizations are essentially CFG manipulations, rearranging and/or duplicating the basic blocks of the program, but leaving the internals of the basic blocks the same. Examples include loop peeling, loop unrolling, and static correlated branch prediction [9]. To support such tasks, the data structures of the CFG are actively linked to the underlying program representation, so that changing graph connections or duplicating CFG nodes results in corresponding changes to the program being c

Citations