Multi-GPU parallel computation of unsteady incompressible flows using kinetically reduced local Navier–Stokes equations

Hashimoto, Tomohisa; Yasuda, T.; Tanno, Itaru; Tanaka, Yoshihiro; Morinishi, Koji; Satofuka, Nobuyuki

doi:10.1016/j.compfluid.2018.03.028

Cited by 17 publications

(12 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Compared to fine-grained parallelism in a kernel with multiple threads, concurrent kernel execution is coarse-grain parallelism. Streams are often used to hide the overhead of data transfer communications from CPU to GPU or vice versa and computations [29,36]. Moreover, for kernels between which there is no data dependency, they can be calculated at the same time through streams to improve efficiency.…”

Section: Concurrent Kernel Executionmentioning

confidence: 99%

Optimization and acceleration of flow simulations for CFD on CPU/GPU architecture

Jiang

Zhou

et al. 2019

J Braz. Soc. Mech. Sci. Eng.

View full text Add to dashboard Cite

With the increasing requirement of high computational power in computational fluid dynamics (CFD) field, the graphic processing units (GPUs) with great floating-point computing capability play more important roles. This work explores the porting of an Euler solver from central processing units (CPUs) to three different CPU/GPU heterogeneous hardware platforms using MUSCL and NND schemes, and then the computational acceleration of one-dimensional (1D) Riemann problem and two-dimensional (2D) flow past a forward-facing step is investigated. Based on hardware structures, memory models and programming methods, the working manner of heterogeneous systems was firstly introduced in this paper. Subsequently, three different heterogeneous methods employed in the current study were presented in detail, while porting all parts of the solver loop to GPU possessed the best performance among them. Several optimization strategies suitable for the solver were adopted to achieve substantial execution speedups, while using shared memory on GPU was relatively rarely reported in CFD literature. Finally, the simulation of 1D Riemann verified the reliability of the modified codes on GPU, demonstrating strong ability in capturing discontinuities of both schemes. The two cases with their 1D computational domains discretized into 10,000 cells both realized a speedup exceeding 25, compared to that executed on a single-core CPU. In simulation of the 2D step flow, we came to the highest speedups of 260 for MUSCL scheme with 800 × 400 mesh size and 144 for NND scheme with 400 × 200 computational domain, respectively.

show abstract

Section: Concurrent Kernel Executionmentioning

confidence: 99%

Optimization and acceleration of flow simulations for CFD on CPU/GPU architecture

Jiang

Zhou

et al. 2019

J Braz. Soc. Mech. Sci. Eng.

View full text Add to dashboard Cite

show abstract

“…The lattice-Boltzmann (LB) method, which solves the Boltzmann transport equation on a discretised phase space [6], is also closely related to the AC method [28,53,5]. All these methods are explicit in time and local in space and thus particularly amenable to massively parallel GPU-based simulations and have low memory requirements [24,25,26,29,41].…”

Section: Introductionmentioning

confidence: 99%

“…The KRLNS equations [30,9,24,25,26], the EDAC method [15,16,29] and GP equation [61] have been applied for different viscous incompressible flows. Kajzer and Pozorski [29] used the EDAC method to perform direct numerical simulations of a turbulent channel flow at the friction Reynolds number 180 and 395 on a collocated grid system.…”

Section: Introductionmentioning

confidence: 99%

Analysis of artificial pressure equations in numerical simulations of a turbulent channel flow

Dupuy

Toutant

Bataille

2020

Journal of Computational Physics

View full text Add to dashboard Cite

Recently, several methods have been proposed to simulate incompressible fluid flows using an artificial pressure evolution equation, avoiding the resolution of a Poisson equation. These methods can be seen as various levels of approximation of the compressible Navier-Stokes equation in the low Mach number limit. We study the simulation of incompressible wall-bounded flows using several artificial pressure equations in order to determine the most relevant approximations. The simulations are stable using a finite difference method in a staggered grid system, even without diffusive term, and converge to the incompressible solution, both in direct numerical simulations and for coarser meshes, to be used in large-eddy simulations. A pressure equation with a convective and a diffusive term produces a more accurate solution than a compressible solver or methods involving more approximations. This suggests that it is near to an optimal level of approximation. The presence of a convective term in the pressure evolution equation is in particular crucial for the accuracy of the method. The rate of convergence of the solution in terms of artificial Mach number is studied numerically and validates the theoretical quadratic convergence rate. We demonstrate that this property can be used to accelerate the rate of convergence using an extrapolation in terms of artificial Mach number. Since the approach is based on an explicit and local system of equations, the numerical procedure is massively parallelisable and has low memory requirements.

show abstract

“…The latter applied a GPU to accelerate the numerical simulation of steady flows around a hypersonic aircraft, which shows that the GPU acceleration has a great application prospect in aerospace engineering [20]. After that, the studies and applications of the multi-GPU acceleration, as well as CPU-GPU collaborative computing, further improved the speed of numerical simulations, and relevant work can be seen in [23][24][25][26]. As with the parallel computing method on CPU, data dependencies need to be paid much attention for the GPU implementation, and moreover, due to the particularity of the GPU architecture, some additional factors should be considered when writing a suitable GPU program.…”

Section: Introductionmentioning

confidence: 99%

GPU acceleration of a 2D compressible Euler solver on CUDA-based block-structured Cartesian meshes

Feng

Liang

Liu

et al. 2020

J Braz. Soc. Mech. Sci. Eng.

View full text Add to dashboard Cite

A novel block-structured Cartesian mesh method is developed that is well designed for the graphics processing unit (GPU) acceleration of flow simulations. The size of the mesh block is set based on the thread hierarchy in Compute Unified Device Architecture which is an easy-to-use GPU programming model. The mesh method is implemented in a finite-volume compressible flow solver, where the two-dimension steady Euler equations are solved by using the AUSM + scheme in spatial discretization and third-order total-variation-diminishing Runge-Kutta method in time discretization. For the GPU implementation, the redundant computation, data structure reorganization and Structure-of-Arrays (SoA) mesh layout are utilized to reduce the frequency and data size for information transfer between host and device. Two test cases, the supersonic flow over a circular cylinder and a NACA0012 airfoil, are presented to validate numerical approaches and valuate computational performance. Additionally, numerical experiments are carried out on a test case at different mesh sizes and execution configurations to investigate properties of the proposed method.

show abstract

Multi-GPU parallel computation of unsteady incompressible flows using kinetically reduced local Navier–Stokes equations

Cited by 17 publications

References 14 publications

Optimization and acceleration of flow simulations for CFD on CPU/GPU architecture

Optimization and acceleration of flow simulations for CFD on CPU/GPU architecture

Analysis of artificial pressure equations in numerical simulations of a turbulent channel flow

GPU acceleration of a 2D compressible Euler solver on CUDA-based block-structured Cartesian meshes

Contact Info

Product

Resources

About