Presentation
GPU Acceleration of Extreme Scale Pseudo-Spectral Simulations of Turbulence Using Asynchronism
SessionComputational Fluid Dynamics
Event Type
Paper
TP
Accelerators
Algorithms
Deep Learning
GPUs
MPI
OpenMP
Performance
Scalable Computing
Simulation
BSP Finalist
TimeTuesday, 19 November 201911:30am - 12pm
Location405-406-407
DescriptionThis paper presents new advances in GPU-driven Fourier pseudo-spectral numerical algorithms, which allow the simulation of turbulent fluid flow at problem sizes which exceed the current state of the art. In contrast to several massively parallel petascale systems, the dense nodes of Summit, Sierra, and expected exascale machines can be exploited with coarser MPI decompositions which result in improved MPI all-to-all scaling. An asynchronous batching strategy, combined with the fast hardware connection between the large CPU memory and the fast GPUs allows effective use of the GPUs on problem sizes which are too large to reside in GPU memory. Communication performance is further improved by a hybrid MPI+OpenMP approach. Favorable performance is obtained up to a 18432^3 problem size on 3072 nodes of Summit, with a GPU to CPU speedup of 4.7 for a 12288^3 problem size (the largest problem size previously published in turbulence literature).
Download PDF
Archive