Evaluating Automatic Parallelization in SUIF

This paper presents the results of an experiment to measure empirically the remaining opportunities for exploiting loop-level parallelism that are missed by the Stanford SUIF compiler, a state-of-t...