A fine grained approach to scheduling asynchronous multiprocessors

A new approach is given for scheduling a sequential instruction stream for execution `in parallel´ on asynchronous multiprocessors. Schedules are constructed by a careful balancing of execution and communication costs at the level of individual instructions, and their data dependencies. Three methods are used to evaluate the authors approach. First, several existing methods are extended to the fine grained situation and then compared to their approach using simulated executions. In each instance, their method is found to provide significantly shorter schedules. Second, the simulations are used to examine the effects of various architectural considerations on the executions of the schedules. Their approach provides significant speedup in a wide-range of situations. Third, schedules produced by their approach are executed on a two-processor Data General shared memory multiprocessor system, indicating a strong correlation between their simulations and the actual executions. Their results establish that fine grained parallelism can be exploited in a substantial manner when scheduling a sequential instruction stream for execution `in parallel´ on asynchronous multiprocessors