PanguLU: A Scalable Regular Two-Dimensional Block-Cyclic Sparse Direct Solver on Distributed Heterogeneous Systems
Xu Fu, Bingbin Zhang, T. Wang, Wenhao Li, Yuechen Lu, Enxin Yi, Jianqi Zhao, X. Q. Geng, Fangying Li, Jingwen Zhang, Zhou Jin, Weifeng Liu
Abstract
Sparse direct solvers play a vital role in large-scale high performance computing in science and engineering. Existing distributed sparse direct methods employ multifrontal/supernodal patterns to aggregate columns of nearly identical forms and to exploit dense basic linear algebra subprograms (BLAS) for computation. However, such a data layout may bring more unevenness when the structure of the input matrix is not ideal, and using dense BLAS may waste many floating-point operations on zero fill-ins.
Topics & Concepts
Computer scienceSolverSparse matrixScalabilityExploitBlock (permutation group theory)ComputationParallel computingComputational scienceScale (ratio)Matrix (chemical analysis)Theoretical computer scienceAlgorithmProgramming languageMathematicsDatabaseGaussianQuantum mechanicsComposite materialMaterials scienceGeometryPhysicsComputer securityMatrix Theory and AlgorithmsElectromagnetic Scattering and AnalysisTensor decomposition and applications