Research and ACM SRC Posters Archive

Memory-Efficient CFD Based on MPS: Effective One-Billion-Cell Resolution on a Single Node

Poster Type: Research Posters

Author: Junya Onishi (RIKEN Center for Computational Science (R-CCS)), Ayato Takii (Kobe University, Japan; RIKEN Center for Computational Science (R-CCS)), Sangwon Kim (RIKEN Center for Computational Science (R-CCS)), Younghwa Cho (Hokkaido University, Japan), Makoto Tsubokura (Kobe University, Japan; RIKEN Center for Computational Science (R-CCS))

Supervisor:

Abstract: We investigate matrix product states (MPS), a tensor-network compression method, as a memory-efficient representation of flow variables. A three-dimensional incompressible Navier-Stokes solver is implemented entirely in MPS form and is applied to canonical flow problems. Results show substantial memory savings and the ability to perform a $1024^3$ simulation on a single GPU. Performance analysis revealed new bottlenecks, particularly bond-dimension growth during nonlinear operations, suggesting novel optimization strategies are needed to fully realize MPS-based CFD at extreme scales.

Best Poster Finalist (BP): no
Poster: PDF
Poster Summary: PDF

Back to Poster Archive Listing