A technique to eliminate redundant inter-processor communication on parallelizing compiler TINPAR

Optimizing inter-processor(PE) communication is crucial for parallelizing c ompilers for message-passing parallel machines to achieve high performance. In this paper, we; propose a technique to eliminate redundant inter-PE mes sages. This technique utilizes data-flow analysis to find a definition poin t that corresponds to a use point where the definition and the use occur in different PEs. If several read accesses occurred in the same PE use the da ta defined at the same definition point in another PE, redundant inter-PE m essages are eliminated as Follows: only one inter-PE communication is perfo rmed for the earliest read access and the previously received data are used for the following read. In order to guarantee the consistency of the data, a valid flag and a sent nag are provided for each chunk of received data. The control of these flags is equivalent to the coherence control by the se lf invalidation on a compiler aided cache coherence scheme.