Evan Bollig, Florida State University

GPU Acceleration of CRM for Global Climate Simulation

Abstract: Clouds play an essential role in Earth's climate, but to this day are a major source of uncertainty in global climate simulations due to limits on resolution inferred by computational cost. To address this concern, and decrease compute times, we are porting a Cloud Resolving Model (CRM) from NCAR's Community Climate System Model (CCSM v3.0) to run on GPU accelerators. We present the current status of this ongoing effort and discuss successes and failures in running CRM on GPUs.

Presentation slides (pdf)