You are here: MIMS > EPrints
MIMS EPrints

2007.94: Using Mixed Precision for Sparse Matrix Computations to Enhance the Performance while Achieving 64-bit Accuracy

2007.94: Alfredo Buttari, Jack Dongarra, Jakub Kurzak, Piotr Luszczek and Stanimire Tomov (2007) Using Mixed Precision for Sparse Matrix Computations to Enhance the Performance while Achieving 64-bit Accuracy.

Full text available as:

PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
317 Kb

Abstract

By using a combination of 32-bit and 64-bit floating point arithmetic the performance of many sparse linear algebra algorithms can be significantly enhanced while maintaining the 64-bit accuracy of the resulting solution. These ideas can be applied to sparse multifrontal and supernodal direct techniques, and sparse iterative techniques such as Krylov subspace methods. The approach presented here can apply not only to conventional processors but also to exotic technologies such as Field Programmable Gate Arrays (FPGA), Graphical Processing Units (GPU), and the Cell BE processor.

Item Type:MIMS Preprint
Additional Information:

Appears also as Technical Report UT-CS-06-584, Department of Computer Science, University of Tennessee, Knoxville, TN, USA, November 2006 and LAPACK Working Note 180

Subjects:MSC 2000 > 65 Numerical analysis
MSC 2000 > 68 Computer science
MIMS number:2007.94
Deposited By:Miss Louise Stait
Deposited On:03 July 2007

Download Statistics: last 4 weeks
Repository Staff Only: edit this item