Abstract
In LAPACK there are two types of subroutines for solving problems with symmetric matrices: routines for full and packed storage. The performance of full format is much better as it allows the usage of Level 2 and 3 BLAS whereas the memory requirement of the packed format is about 50% of full. We propose a new storage layout which combines the advantages of both algorithms: its factorization performance is better than that of full storage layout, and its memory requirement is percentage-wise slightly larger than packed storage.
Our new algorithms, called DBSSV, DBSTRF, and DBSTRS are now part of ESSL[9]. On three recent IBM RS/6000 platforms, Power3, Power2 and PowerPC 604e DBSTRF outperforms LAPACK’s DSYTRF by about 20%, and DBSTRS, with 100 RHS, outperforms LAPACK’s DSYTRS by more than 100%. These performance results are decidedly unfair to our new algorithms: we compare against Level 3 algorithms as opposed to Level 2 packed algorithms.
This research is supported by the UNI•C collaboration with the IBM T.J. Watson Research Center at Yorktown Heights. The research of the last author was supported by Grants MM-707 and I-702 of the Bulgarian Ministry of Education and Science.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
B. Andersen, F. Gustavson, and J. Waśniewski, A recursive Formulation of the Cholesky Factorization Operating on a Matrix in Packed Storage Form, University of Tennessee, Knoxville, TN, Computer Science Dept. Technical ReportCS-00-441, May 2000, also LAPACK Working Note number 146 (http://www.netlib.org/lapack/lawns/lawn146.ps), and submitted to the Transaction of thematical Software (TOMS) of the ACM.
E. Anderson, Z. Bai, C. Bischof, J. Demmel, J. Dongarra, J. Du Croz, A. Greenbaum, S. Hammarling, A. McKenney, S. Ostrouchov, and D. Sorensen, LAPACK Users’ Guide Release 3.0, SIAM, Philadelphia, 1999, (http://www.netlib.org/lapack/lug/lapack lug.html).
E. Elmroth and F. Gustavson, Applying Recursion to Serial and Parallel QR Factorization Leads to Better Performance, IBM Journal of Research and Development, 44, #4, 605–624, 2000
G. H. Golub and C. F. Van Loan, Matrix Computations, 3rd edition, The John Hopkins University Press, Baltimore, 1996.
F. Gustavson, Recursion leads to automatic variable blocking for dense linearalgebra algorithms, IBM J. Res. Develop., 41 (1997), pp. 737–755.
F. Gustavson, A. Karaivanov, J. Waśniewski, and P. Yalamov, Recursive formulation of algorithms for symmetric indefinite linear systems (manuscript).
F. Gustavson, A. Henriksson, I. Jonsson, B. Kågström, and P. Ling, Recursive Blocked Data Formats and BLAS’s for Dense Linear Algebra Algorithms, in Applied Parallel Computing, B. Kågström et. al., eds., Lecture Notes in Computers Science, v. 1541, Springer, 1998, pp. 120–128.
N. J. Higham, Accuracy and Stability of Numerical Algorithms, SIAM, Philadelphia, 1996.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2001 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Gustavson, F., Karaivanov, A., Marinova, M., Waśniewski, J., Yalamov, P. (2001). A Fast Minimal Storage Symmetric Indefinite Solver. In: Sørevik, T., Manne, F., Gebremedhin, A.H., Moe, R. (eds) Applied Parallel Computing. New Paradigms for HPC in Industry and Academia. PARA 2000. Lecture Notes in Computer Science, vol 1947. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-70734-4_14
Download citation
DOI: https://doi.org/10.1007/3-540-70734-4_14
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-41729-3
Online ISBN: 978-3-540-70734-9
eBook Packages: Springer Book Archive