Parallelizing iterative loops with conditional branching

Abstract

This paper considers automatic restructuring of loops with branching for parallel processing, especially a class of loops termed conditional cyclic loops. A cyclic loop possesses a dependence cycle caused by branching across loop iterations, which makes it difficult to parallelize. In general, parallel execution of a cyclic loop provides little benefit due to the need of solving a full-order nonlinear Boolean recurrence relation. However, the Boolean recurrence in practice is often of simpler forms. With the simpler forms, the number of possible predicate values of branching is reduced drastically compared to a general cyclic loop, These simple forms of cyclic loops found in practice can be parallelized for O(p/ log p) speedup with p processors. >