System and method for executing compute-intensive database user-defined programs on an attached high-performance parallel computer

摘要: The invention pertains to a system and method for dispatching executing the compute-intensive parts of workflow database queries on an attached high-performance, parallel computing platform. performance overhead moving required data results between platform high-performance where workload is executed amortized in several ways, example, by exploiting fine-grained parallelism superior hardware speeding up calculations, using in-memory structures cache sets sequence time-lagged same data, so that these can be processed without further transfer overheads, replicating within multiple independent target set simultaneously partitions A specific embodiment this was used deploying bio-informatics application involving gene protein matching Smith-Waterman algorithm connected via Ethernet local area network supercomputer.