Using Mixed Precision for Sparse Matrix Computations to Enhance the Performance while Achieving 64-bit Accuracy
University of Tennessee, Knoxville
ACM Trans. Math. Softw., Vol. 34, No. 4. (2008), pp. 1-22
@article{buttari2008using,
title={Using mixed precision for sparse matrix computations to enhance the performance while achieving 64-bit accuracy},
author={Buttari, A. and Dongarra, J. and Kurzak, J. and Luszczek, P. and Tomov, S.},
journal={ACM Transactions on Mathematical Software (TOMS)},
volume={34},
number={4},
pages={1–22},
issn={0098-3500},
year={2008},
publisher={ACM}
}
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.
December 12, 2010 by hgpu
