Bidiagonalization and R-Bidiagonalization: Parallel Tiled Algorithms, Critical Paths and Distributed-Memory Implementation

Abstract : We study tiled algorithms for going from a " full " matrix to a condensed " band bidiagonal " form using orthogonal transformations: (i) the tiled bidiagonalization algorithm BIDIAG, which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm R-BIDIAG, which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both BIDIAG and R-BIDIAG, we use four main types of reduction trees, namely FLATTS, FLATTT, GREEDY, and a newly introduced auto-adaptive tree, AUTO. We provide a study of critical path lengths for these tiled algorithms, which shows that (i) R-BIDIAG has a shorter critical path length than BIDIAG for tall and skinny matrices, and (ii) GREEDY based schemes are much better than earlier proposed algorithms with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts.
Complete list of metadatas

https://hal.inria.fr/hal-01484113
Contributor : Mathieu Faverge <>
Submitted on : Monday, March 6, 2017 - 6:24:24 PM
Last modification on : Wednesday, November 20, 2019 - 3:03:57 AM
Long-term archiving on: Wednesday, June 7, 2017 - 4:06:35 PM

File

ipdps-final.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01484113, version 1

Citation

Mathieu Faverge, Julien Langou, Yves Robert, Jack Dongarra. Bidiagonalization and R-Bidiagonalization: Parallel Tiled Algorithms, Critical Paths and Distributed-Memory Implementation. IPDPS'17 - 31st IEEE International Parallel and Distributed Processing Symposium , May 2017, Orlando, United States. ⟨hal-01484113⟩

Share

Metrics

Record views

929

Files downloads

299