Performance predictions of multilevel communication optimal LU and QR factorizations on hierarchical platforms

Laura Grigori 1, 2 Mathias Jacquelin 3 Amal Khabou 4
2 ALPINES - Algorithms and parallel tools for integrated numerical simulations
LJLL - Laboratoire Jacques-Louis Lions, Inria Paris-Rocquencourt, Institut National des Sciences Mathématiques et de leurs Interactions
4 POSTALE - Performance Optimization by Software Transformation and Algorithms & Librairies Enhancement
LRI - Laboratoire de Recherche en Informatique, Inria Saclay - Ile de France
Abstract : In this paper we study the performance of two classical dense linear algebra algorithms, the LU and the QR factorizations, on multilevel hierarchical platforms. We note that we focus on multilevel QR factorization, and give a brief description of the multilevel LU factorization. We first introduce a performance model called Hierarchical Cluster Platform (Hcp), encapsulating the characteristics of such platforms. The focus is set on reducing the communication requirements of studied algorithms at each level of the hierarchy. Lower bounds on communication are therefore extended with respect to the Hcp model. We then present a multilevel QR factorization algorithm tailored for those platforms, and provide a detailed performance analysis. We also provide a set of performance predictions showing the need for such hierarchical algorithms on large platforms.
Complete list of metadatas

https://hal.inria.fr/hal-01112909
Contributor : Laura Grigori <>
Submitted on : Tuesday, February 3, 2015 - 7:17:25 PM
Last modification on : Sunday, March 31, 2019 - 1:24:38 AM

Links full text

Identifiers

Citation

Laura Grigori, Mathias Jacquelin, Amal Khabou. Performance predictions of multilevel communication optimal LU and QR factorizations on hierarchical platforms. 29th International Supercomputing Conference, ISC 2014, Jun 2014, Leipzig, Germany. ⟨10.1007/978-3-319-07518-1_5⟩. ⟨hal-01112909⟩

Share

Metrics

Record views

318