Check out my first novel, midnight's simulacra!
Winograd's Algorithm: Difference between revisions
From dankwiki
(Created page with 'Category: Computer Science Eponyms') |
No edit summary |
||
| Line 1: | Line 1: | ||
A matrix multiplication algorithm which replaces one recursive matrix multiplication with 15 matrix additions (compared to [[Strassen's Algorithm|Strassen's]] 18), yielding O(n<sup>2.81</sup>) asymptotic performance. It's used by [http://www.ics.uci.edu/~fastmm/FMM-Reference/reference.html D'Alberto and Nicolau] in their hybrid-daptive matrix multiplication library, and Craig Douglas in his [http://www.mgnet.org/~douglas/ccd-free-software.html GIMMW]. | |||
==See Also== | |||
* The [[Coppersmith-Winograd Algorithm]] for matrix multiplication | |||
* [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.39.6887 "GEMMW: A Portable Level 3 Blas Winograd Variant Of Strassen's Matrix-Matrix Multiply Algorithm"] from the January 1994 Journal of Computational Physics. | |||
[[Category: Computer Science Eponyms]] | [[Category: Computer Science Eponyms]] | ||
Latest revision as of 00:44, 30 September 2009
A matrix multiplication algorithm which replaces one recursive matrix multiplication with 15 matrix additions (compared to Strassen's 18), yielding O(n2.81) asymptotic performance. It's used by D'Alberto and Nicolau in their hybrid-daptive matrix multiplication library, and Craig Douglas in his GIMMW.
See Also
- The Coppersmith-Winograd Algorithm for matrix multiplication
- "GEMMW: A Portable Level 3 Blas Winograd Variant Of Strassen's Matrix-Matrix Multiply Algorithm" from the January 1994 Journal of Computational Physics.
