论文信息 - Implementing Iterative Solvers for Irregular Sparse Matrix Problems in High Performance Fortran

Implementing Iterative Solvers for Irregular Sparse Matrix Problems in High Performance Fortran

Writing efficient iterative solvers for irregular, sparse matrices in HPF is hard. The locality in the computations is unclear, and for efficiency we use storage schemes that obscure any structure in the matrix. Moreover, the limited capabilities of HPF to distribute and align data structures make it hard to implement the desired distributions, or to indicate these such that the compiler recognizes the efficient implementation.

Eric de Sturler | Damian Loher

[1] Y. Saad,et al. GMRES: a generalized minimal residual algorithm for solving nonsymmetric linear systems , 1986 .

[2] E. de Sturler,et al. Incomplete block LU preconditioners on slightly overlapping subdomains for a massively parallel computer , 1995 .

[3] Alex Pothen,et al. PARTITIONING SPARSE MATRICES WITH EIGENVECTORS OF GRAPHS* , 1990 .

[4] E. Sturler,et al. Communication Cost Reduction for Krylov Methods on Parallel Computers , 1994, HPCN.

[5] Horst D. Simon,et al. Fast multilevel implementation of recursive spectral bisection for partitioning unstructured problems , 1994, Concurr. Pract. Exp..

[6] E. De Sturler. Iterative methods on distributed memory computers , 1994 .

[7] Rice UniversityCORPORATE,et al. High performance Fortran language specification , 1993 .

[8] Frédéric Nataf,et al. Domain Decomposition Methods for Fluid Dynamics , 1995 .

[9] Giovanni P. Galdi,et al. Navier-Stokes Equations and Related Nonlinear Problems , 1995 .

[10] H. V. D. Vorst,et al. Reducing the effect of global communication in GMRES( m ) and CG on parallel distributed memory computers , 1995 .