Efficient parallelisation using combined loop and data transformations
Michael O’Boyle, Peter M. W. Knijnenburg · International Conference on Parallel Architectures and Compilation Techniques · 1999
This paper attempts to minimize parallelization overhead on distributed shared memory machines, such as the SGi Origin 2000, by the combination of non-singular loop and data transformations. We show that conflicting requirements on a loop transformation may be resolved by using a data transformation and vice-versa. We develop optimization criteria for locality, synchronization and communication and show that neither loop nor data transformations can be solely used for efficient parallelization. This leads to the development of a novel global optimization heuristic which is applied to 3 SPEC kernels where it is shown to outperform techniques solely based on loop or data transformations and to give significant improvement over an existing state-of-the- art commercial auto-parallelizer.