Benchmarking MILC code with OpenMP and MPI

A trend in high performance computers that is becoming increasingly popular is the use of symmetric multiprocessing (SMP) rather than the older paradi gm of MPP. MPI codes that ran and scaled well on MPP machines can often be run on an SR;IP machine using the vendor's version of MPI. However, this ap proach may not make optimal use of the (expensive) SMP hardware. More signi ficantly, there are machines like Blue Horizon, an IBM SP with 8-way. SMP n odes at the San Diego Supercomputer Center that carl only support 4 MPI pro cesses per node (with the current switch). On such a machine it is imperati ve to be able to use OpenMP parallelism on the node, and MPI between nodes. We describe the challenges of converting MILC MPI code to using a second l evel of OpenMP parallelism, and benchmarks on IBM and Sun computers.