Overview

This book is a comprehensive introduction to all the components of a high-performance parallel linear algebra library, as well as a guide to the PLAPACK infrastructure. PLAPACK is a library infrastructure for the parallel implementation of linear algebra algorithms and applications on distributed memory supercomputers such as the Intel Paragon, IBM SP2, Cray T3D/T3E, SGI PowerChallenge, and Convex Exemplar. This infrastructure allows library developers, scientists, and engineers to exploit a natural approach to encoding so-called blocked algorithms, which achieve high performance by operating on submatrices and subvectors. This feature, as well as the use of an alternative, more application-centric approach to data distribution, sets PLAPACK apart from other parallel linear algebra libraries, allowing for strong performance and significanltly less programming by the user. This book is a comprehensive introduction to all the components of a high-performance parallel linear algebra library, as well as a guide to the PLAPACK infrastructure. Scientific and Engineering Computation series

Full Product Details

Author:   Robert van de Geijn (University of Texas at Austin)
Publisher:   MIT Press Ltd
Imprint:   MIT Press
Dimensions:   Width: 20.30cm , Height: 1.30cm , Length: 22.90cm
Weight:   0.386kg
ISBN:  

9780262720267

ISBN 10:   0262720264
Pages:   214
Publication Date:   03 April 1997
Recommended Age:   From 18
Audience:   College/higher education ,  Professional and scholarly ,  Undergraduate ,  Postgraduate, Research & Scholarly
Format:   Paperback
Publisher's Status:   No Longer Our Product
Availability:   Out of stock  
The supplier is temporarily out of stock of this item. It will be ordered for you on backorder and shipped when it becomes available.

Table of Contents

Part 1 Introduction: why a new infrastructure? natural description of linear algebra algorithms; physically based matrix distribution; redistributing and duplicating matrices and vectors; implementation of basic matrix-vector operations (preview); basic linear algebra subprograms; message-passing interface; parallel sparse linear algebra; FORTRAN interface; availability. Part 2 Templates and linear algebra objects: initializing PLAPACK; distribution templates; linear algebra objects; example; return values; more operations and information. Part 3 Advanced linear algebra object manipulation: creating views into objects;, splitting of linear algebra objects; shifting of linear algebra objects; determining where to split; creating objects 'conformal to' other objects; annotating object orientation; casting object types; more operations and information. Part 4 Application program interface: introduction; API-activation; opening and closing an object; accessing a vector; accessing a matrix; completion and synchronization; examples; more operations and information. Part 5 Data duplication and consolidation: copy; reduce; pipeline computation and communication; a building block approach to implementing copy and reduce; more operations and information. Part 6 Vector-vector operations: copy; swap; scaling a vector (object); scaled vector (object) addition; inner product of vectors; norms of vectors; maximum absolute value in vector; example - parallelizing inner product; example - parallelizing 'axpy' for vector objects; more operations and information. Part 7 Matrix-vector operations: general matrix-vector multiplication; symmetric matrix-vector multiplication; triangular matrix-vector multiplication; triangular solve; Rank-1 update; symmetric Rank-1 update; symmetric Rank-2 update; example - parallelizing matrix-vector multiplication; example parallelizing Rank-1 update; more operations and information. Part 8 Matrix-matrix operations: general matrix-matrix multiplication; symmetric matrix-matrix multiplication; symmetric Rank-k update; symmetric Rank-2k update; triangular matrix-matrix multiplication; triangular solve with multiple right-hand-sides; example - parallelizing matrix-matrix multiplication; queering algorithmic blocking size; more operations and information. Part 9 Application of the infrastructure: Cholesky factorization; right-looking variant; left-looking variant; more operations and information. Summaries: PLAPACK routines and their arguments; BLAS related routines.

Reviews

Author Information

Robert van de Geijn is a Professor in the Department in Computer Science and the Texas Institute for Computational Engineering and Sciences at the University of Texas at Austin.

Tab Content 6

Author Website:  

Countries Available

All regions