The goal of this project is to develop compilation techniques to support efficient machine-independent programming of high-performance multiprocessors through implementation of COSMIC, the "Communication-Optimizing, Shared-Memory Integrated Compiler." To achieve high performance, COSMIC will apply optimizations for locality, synchronization, communication, and memory system effects. COSMIC follows two basic guidelines. First, it uses compilation techniques for message-passing machines to retain most of the benefits of explicit messages. Second, it exploits architectural and operating system support available on shared-memory multiprocessors to improve flexibility and performance. A novel characteristic of COSMIC will be its ability to take advantage of the multiple coherence protocols and hybrid message-passing support found in software Distributed -Shared-Memory (DSM) systems and Flexible-Shared-Memory system (FSM) machines. To evaluate the impact on the performance of scientific applications, COSMIC will be tested on programs from benchmark suites such as Perfect, NAS, SPEC, and SADIE, as well as representative applications (e.g., CHARMM, MOLDYN, DSMC, EULER, NRL Flame) from fields such as computational chemistry, computational fluid dynamics, and computational combustion.
