FluTAS: A GPU-accelerated finite difference code for multiphase flows

Abstract

We present the Fluid Transport Accelerated Solver, FluTAS, a scalable GPU code for multiphase flows with thermal effects. The code solves the incompressible Navier-Stokes equation for two-fluid systems, with a direct FFTbased Poisson solver for the pressure equation. The interface between the two fluids is represented with the Volume of Fluid (VoF) method, which is mass conserving and well suited for complex flows thanks to its capacity of handling topological changes. The energy equation is explicitly solved and coupled with the momentum equation through the Boussinesq approximation. The code is conceived in a modular fashion so that different numerical methods can be used independently, the existing routines can be modified, and new ones can be included in a straightforward and sustainable manner. FluTAS is written in modern Fortran and parallelized using hybrid * M.C-E. and N.S. contributed equally to this work.

Multiple-GPU accelerated high-order gas-kinetic scheme for direct numerical simulation of compressible turbulence

“…For the case with 512 3 cells, the computational time of CPU code with supercomputer using 1024 Intel Xeon E5-2692 cores is approximately 3 times longer than that of GPU code using 8 Tesla V100 GPUs. It can be inferred that the efficiency of GPU code with 8 Tesla V100 GPUs approximately equals to that of MPI code with 3000 supercomputer cores, which agree well with the multiple-GPU accelerated finite difference code for multiphase flows [26]. These comparisons shows the excellent performance of multiple-GPU accelerated HGKS for large-scale turbulence simulation.…”

“…Meanwhile, GPU is a form of hardware acceleration, which is originally developed for graphics manipulation and is extremely efficient at processing large amounts of data in parallel. Currently, GPU has gained significant popularity in high performance scientific computing [24,25,26]. In this paper, to accelerate the computation, the WENO based HGKS will be implemented with single GPU using CUDA.…”

“…Since these units have a parallel computation capability inherently, they can provide fast and low cost solutions to high performance computing (HPC). In recent years, GPUs have gained significant popularity in computational fluid dynamics [24,25,26] as a cheaper, more efficient, and more accessible alternative to large-scale HPC systems with CPUs. Especially, to numerically study the turbulent physics in much higher Reynolds number turbulence, the extreme-scale DNS in turbulent flows has been implemented using multiple GPUs, i.e., incompressible isotropic turbulence up to 18432 3 resolution [27] and turbulent pipe flow up to Re τ ≈ 6000 [28].…”