The Automated Crystal Runtime System: A Framework

There exists substantial data level parallelism in scientific problems. The Crystal/ACRE(Automated Crystal Runtime Environment) runtime system is an attempt to obtain parallel, implementations for scientific computations, particularly those where the data dependencies are manifest only at runtime. This can preclude compiler based detection of certain types of parallelism. The automated system is structured as follows: An appropriate level of granularity is first selected for the computations. A directed acyclic graph representation of the program is generated on which various aggregation techniques may be employed in order to generate efficient schedules. These schedules are then mapped onto the largest machine. We describe some initial results from experiments conducted on the Intel Hypercube and the Encore Multimax that indicate the usefulness of our approach. Using the runtime system, it will be relatively easy to program different applications and study the performance implications of the various parameters. When the performance data is available, we would like to develop mathematical models that describe the relationships between the various important parameters in the system.