| タイトル | Minimizing inner product data dependencies in conjugate gradient iteration |
| 本文（外部サイト） | http://hdl.handle.net/2060/19830026339 |
| 著者（英） | Vanrosendale, J. |
| 著者所属（英） | NASA Langley Research Center |
| 発行日 | 1983-07-01 |
| 言語 | eng |
| 内容記述 | The amount of concurrency available in conjugate gradient iteration is limited by the summations required in the inner product computations. The inner product of two vectors of length N requires time c log(N), if N or more processors are available. This paper describes an algebraic restructuring of the conjugate gradient algorithm which minimizes data dependencies due to inner product calculations. After an initial start up, the new algorithm can perform a conjugate gradient iteration in time c*log(log(N)). |
| NASA分類 | COMPUTER PROGRAMMING AND SOFTWARE |
| レポートNO | 83N34610
REPT-83-36
NASA-CR-172178
NAS 1.26:172178 |
| 権利 | No Copyright |
| URI | https://repository.exst.jaxa.jp/dspace/handle/a-is/161518 |
