Parallelization of Diagnostics for Climate Model Development

The parallelization of the diagnostics forclimate research has been an important goal in the performance testing andimprovement of the diagnostics for the Department of Energy’s (DOE’s) AcceleratedClimate Modeling for Energy (ACME) project [1]. The primary mission of the ACMEproject is to build and test the next-generation Earth system model for currentand future generations of computing systems operated by the DOE office ofscience computing facilities, including the envisioned exascale systemsforeseen in the early part of the next decade. As part of the underpinningworkflow environment, a diagnostics, model metrics, and intercomparison Pythonframework, called UVC Metrics was created to aid in testing and productionexecution of the model. This framework builds on common methods and similarmetrics to accommodate and diagnose individual component models, such asatmosphere, land, ocean, sea ice, and land ice. This paper reports on initialparallelization of UVC Metrics for the atmosphere model component using twopopular frameworks: MPI and SPARK. A timing study is presented to assess theperformance of each method in which significant improvement was achieved forboth frameworks despite I/O contentions with NFS. The advantages anddisadvantages of each framework are also presented.