A performance analysis of a mimetic finite difference scheme for acoustic wave propagation on GPU platforms

Abstract

, J. A performance analysis of a mimetic finite differencescheme for acoustic wave propagation on GPU platforms. "Concurrency and computation. Practice and experience", 1 Febrer 2017, vol. 29, núm. 4, p. 1-18 SUMMARYRealistic applications of numerical modeling of acoustic wave dynamics usually demand high performance computing because of the large size of study domains and demanding accuracy requirements on simulation results. Forward modeling of seismic motion on a given subsurface geological structure is by itself a good example of such applications, and when used as a component of seismic inversion tools or as a guide for the design of seismic surveys, its computational cost increases enormously. In the case of finite difference methods (or any other volumen-discretization scheme), memory and computing demands rise with grid refinement which may be neccesary to reduce errors on numerical wave patterns and better capture target physical devices. In this work, we present several implementations of a mimetic finite difference method for the simulation of acoustic wave propagation on highly dense staggered grids. These implementations evolve as different optimization strategies are employed starting from appropriate setting of compilation flags, code vectorization by using SSE and AVX instructions, CPU parallelization by exploiting the OpenMP framework, to the final code parallelization on GPU platforms. We present and discuss the increasing processing speed-up of this mimetic scheme achieved by the gradual implementation and testing of all these performance optimizations. In terms of simulation times, the performance of our GPU parallel implementations is consistently better than the best CPU version.Received . . . KEY WORDS: Acoustic waves; mimetic finite differences; SIMD extensions; GPU; CUDA programming * Correspondence to: C/ Jordi Girona 1.3, Edf. 08034, This is the peer reviewed version of the following article: Otero, B., Frances, J., Rodriguez-Cruz, R., Rojas, O., Solano, F., Guevara-Jordan, J. A performance analysis of a mimetic finite difference scheme for acoustic wave propagation on GPU platforms. "Concurrency and computation.

Alternating direction implicit time integrations for finite difference acoustic wave propagation: Parallelization and convergence

“…Main reasons are their formulation on structured meshes and their low data dependency among grid sub-domains. Particularly, implementations on Graphics Processing Units (GPU) have become useful and powerful wave simulation tools in 2-D and 3-D, given that realistic applications may result computationally intensive even in two spatial dimensions [31], [33], [39], [47], [49], [54]. Conventional ADI methods and CFD schemes addressing for high computational efficiency must rely on fast SEL solvers.…”