Automated Construction of High Performance Distributed Programs in LuNA System

Akhmed-Zaki, Darkhan; Lebedev, Danil; Malyshkin, Victor

doi:10.1007/978-3-030-25636-4_1

Cited by 11 publications

(2 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To speed up scientific calculations, various technologies are used, such as Open Multi-Processing (OpenMP) (Iryanto & Gunawan, 2017;Werneck et al, 2016), Message Passing Interface (MPI) (Akhmed-Zaki et al, 2019;Pannala et al, 2003), Open Computing Language (OpenCL) (Khramchenkov & Khramchenkov, 2018), Compute Unified Device Architecture (CUDA) (McClure et al, 2014;Zaza et al, 2016) and fragmented programming (Akhmed-Zaki et al, 2019;Malyshkin & Perepelkin, 2011).…”

Section: Public Interest Statementmentioning

confidence: 99%

“…A parallel implementation of the considered problem is realized by the MPI programming interface, which is the main tool in the development of parallel programs for supercomputers. In the future, a project to develop an active knowledge management system to solve the multicomponent fluid flow problem is planned (Akhmed-Zaki et al, 2019). The result of this project will be a system that will automatically choose the a solution method, one of which is the method described in this article, for practical problems.…”

Section: Public Interest Statementmentioning

confidence: 99%

See 1 more Smart Citation

GMRES based numerical simulation and parallel implementation of multicomponent multiphase flow in porous media

et al. 2020

View full text Add to dashboard Cite

This article considered the numerical simulation of multicomponent multiphase flow in porous media. The resulting system of nonlinear equations linearized by the Newton-Raphson method and solved with the iterative Generalized minimal residual method (GMRES) algorithm. To achieve better convergence, we used the ILU(0) preconditioner to the GMRES algorithm. As a result, we used a completely implicit scheme called the Newton-ILU0-GMRES algorithm to solve the problem of interest. Based on the obtained sequential algorithm, we implemented a parallel algorithm using Message Passing Interface (MPI) technology. Additionally, we made comparisons between the parallel program of the presented algorithm and the parallel program using the ready-made Portable Extensible Toolkit for Scientific Computation (PETSc) library. We developed an MPI parallel algorithm and tested it on the MVS-10P supercomputer of the Interdepartmental Supercomputer Center of the Russian Academy of Sciences.

show abstract