Chris Fowler
Chris Fowler, a Cambridge Computer HPC/AI research scientist, is transforming porous media fluid flow modeling with groundbreaking simulations on the HiPerGator supercomputer.
A century-old mathematical theory—first proposed by meteorologist Lewis Fry Richardson in 1922—is being redefined through the combined power of advanced theory and cutting-edge computational science.
Fowler has spent six years developing the Lattice Boltzmann Porous Medium (LBPM-UNC) simulator, capable of capturing microscale physics of fluid flows through porous materials. This work supports the revolutionary Thermodynamically Constrained Averaging Theory (TCAT) developed by Drs. Cass T. Miller and William G. Gray at the University of North Carolina at Chapel Hill.
In a recent hero-run at the University of Florida, Fowler harnessed 512 GPUs across HiPerGator’s 64 DGX A100 SuperPOD nodes, achieving a remarkable 90% weak scaling performance. This means the computation was 512 times larger than what’s possible on a single GPU with minimal efficiency loss—a significant technical achievement.
The cross-institution collaboration included UF Distinguished Professor Dr. Rafael Muñoz-Carpena, UF Information Technology staff, and NVIDIA’s AI Technology Center, with NVIDIA senior solutions architects Mark Hill and Kristopher Keipert providing critical guidance.
This breakthrough has far-reaching implications for carbon capture, climate prediction, environmental remediation, and medical applications like tumor treatment modeling. The work demonstrates how theoretical advances, when paired with computational resources like HiPerGator, can solve complex scientific challenges with real-world impact.
“It’s only through the joint use of theory, computation, and experimental approaches that such advancements are possible,” notes Dr. Miller, emphasizing the crucial role of supercomputing resources in enabling this scientific progress.
Read more in this UFIT Research Computing article.