Amestoy, Patrick R.; Duff, Iain S.; Vömel, Christof Task scheduling in an asynchronous distributed memory multifrontal solver. (English) Zbl 1075.65039 SIAM J. Matrix Anal. Appl. 26, No. 2, 544-565 (2005). Summary: We describe the improvements to the task scheduling for MUMPS, an asynchronous distributed memory direct solver for sparse linear systems. In the new approach, we determine, during the analysis of the matrix, candidate processes for the tasks that will be dynamically scheduled during the subsequent factorization. This approach significantly improves the scalability of the solver in terms of execution time and storage. By comparison with the previous version of MUMPS, we demonstrate the efficiency and the scalability of the new algorithm on up to 512 processors. Our test cases include matrices from regular three-dimensional grids and irregular grids from real-life applications. Cited in 5 Documents MSC: 65F05 Direct numerical methods for linear systems and matrix inversion 65F35 Numerical computation of matrix norms, conditioning, scaling 65F50 Computational methods for sparse matrices 65Y20 Complexity and performance of numerical algorithms Keywords:sparse linear systems; high performance computing; MUMPS; multifrontal Gaussian elimination; distributed memory code; task scheduling; numerical examples Software:Scotch; SPARSPAK; SparseMatrix; ScaLAPACK; METIS; MUMPS; PaStiX PDFBibTeX XMLCite \textit{P. R. Amestoy} et al., SIAM J. Matrix Anal. Appl. 26, No. 2, 544--565 (2005; Zbl 1075.65039) Full Text: DOI