Efficient distribution analysis via graph contraction

Alignment and distribution of array data should be managed by optimizing compilers for parallel computers, but current approaches to the distribution problem formulate it as an NP-complete graph optimization problem. The graphs arising in applications are large and difficult to optimize. Some earlier results on methods that use graph contraction to reduce the size of a distribution problem are improved. An experiment using 7 example programs that show these contraction operations to be effective in practice is reported on. Between 70% and 99% reductions in problem size were obtained, the larger number being more typical, without loss of solution quality.