GTC 2020: Accelerating Chemistry Modules in Atmospheric Models Using GPUs
After clicking “Watch Now” you will be prompted to login or join.
Click “Watch Now” to login or join the NVIDIA Developer Program.
Accelerating Chemistry Modules in Atmospheric Models Using GPUs
Christian Guzman-Ruiz, Barcelona Supercomputing Center
We'll explore a novel solution to speed up chemistry modules in atmospheric models. The Multiscale Online Nonhydrostatic AtmospheRe CHemistry model (MONARCH), a chemical weather prediction system developed by the Barcelona Supercomputing Center, is used as our test bed. The model implements a new flexible treatment for gas- and aerosol-phase chemical processes, Chemistry Across Multiple Phases (CAMP), that allows multiple chemical processes (such as gas- and aerosol-phase chemical reactions, emissions, deposition, photolysis, and mass-transfer) to be solved simultaneously as a single system. We'll discuss innovative ways to speed up the CAMP module, including reducing memory accesses, multiple-cell chemistry solving, adaptation of the most time-consuming functions to GPUs, and heterogeneous computation approaches for CPU/GPU execution. We'll compare the optimized model to state-of-the art chemistry solvers commonly used in the earth sciences community (for example, EBI and KPP).