ffflfiffi !# " $&%('*)+,%('-)fi.+

Computational Grids have become an important and popular computing platform for both scientific and commer-cial distributed computing communities. However, users of such systems typically find achievement of application ex-ecution performance remains challenging. Although Grid infrastructures such as Legion and Globus provide basic re-source selection functionality, work allocation functional-ity, and scheduling mechanisms, applications must interpret system performance information in terms of their own re-quirements in order to develop performance-efficient sched-ules. We describe a new high-performance scheduler that in-corporates dynamic system information, application re-quirements, and a detailed performance model in order to create performance efficient schedules. While the sched-uler is designed to provide improved performance for a magneto hydrodynamics simulation in the Legion Compu-tational Grid infrastructure, the design is generalizable to other systems and other data-parallel, iterative codes. We describe the adaptive performance model, resource selec-tion strategies, and scheduling policies employed by the scheduler. We demonstrate the improvement in application performance achieved by the scheduler in dedicated and shared Legion environments