Distributed One-Stage Hessenberg-Triangular Reduction with Wavefront Scheduling
نویسندگان
چکیده
A novel parallel formulation of Hessenberg-triangular reduction of a regular matrix pair on distributed memory computers is presented. The formulation is based on a sequential cache-blocked algorithm by Kågström, Kressner, E.S. Quintana-Ortí, and G. QuintanaOrtí (2008). A static scheduling algorithm is proposed that addresses the problem of underutilized processes caused by two-sided updates of matrix pairs based on sequences of rotations. Experiments using up to 961 processes demonstrate that the new formulation is an improvement of the state of the art and also identify factors that limit its scalability.
منابع مشابه
Parallel Reduction of a Block Hessenberg-Triangular Matrix Pair to Hessenberg-Triangular Form—Algorithm Design and Performance Results
The design, implementation and performance of a parallel algorithm for reduction of a matrix pair in block upper Hessenberg-Triangular form (Hr, T ) to upper Hessenberg-triangular form (H, T ) is presented. This reduction is the second stage in a two-stage reduction of a regular matrix pair (A, B) to upper Hessenberg-Triangular from. The desired upper Hessenberg-triangular form is computed usin...
متن کاملBlocked Algorithms for Reduction of a Regular Matrix Pair to Generalized Schur Form
This contribution considers the problem of transforming a regular matrix pair (A;B) to generalized Schur form. The focus is on blocked algorithms for the reduction process that typically includes two major steps. The rst is a two-stage reduction of a regular matrix pair (A;B) to condensed form (H;T ) using orthogonal transformations Q and Z such that H = QAZ is upper Hessenberg and T = QBZ is u...
متن کاملEfficient Algorithm for Simultaneous Reduction to the m-Hessenberg–Triangular–Triangular Form
This paper proposes an efficient algorithm for simultaneous reduction of three matrices. The algorithm is a blocked version of the algorithm described by Miminis and Page (1982) which reduces A to the m-Hessenberg form, and B and E to the triangular form. The m-Hessenberg– triangular–triangular form of matrices A, B and E is specially suitable for solving multiple shifted systems. Such shifted ...
متن کاملParallel Algorithms and Library Software for the Generalized Eigenvalue Problem on Distributed Memory Computer Systems
We present and discuss algorithms and library software for solving the generalized non-symmetric eigenvalue problem (GNEP) on high performance computing (HPC) platforms with distributed memory. Such problems occur frequently in computational science and engineering, and our contributions make it possible to solve GNEPs fast and accurate in parallel using state-of-the-art HPC systems. A generali...
متن کاملParallel Reduction from Block Hessenberg to Hessenberg using MPI
In many scientific applications, eigenvalues of a matrix have to be computed. By first reducing a matrix from fully dense to Hessenberg form, eigenvalue computations with the QR algorithm become more efficient. Previous work on shared memory architectures has shown that the Hessenberg reduction is in some cases most efficient when performed in two stages: First reduce the matrix to block Hessen...
متن کامل