arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 0707.2347cs.MS
收藏

Strassen-Winograd矩阵乘法算法的内存有效调度

Memory efficient scheduling of Strassen-Winograd's matrix multiplication algorithm

Brice Boyer, Jean-Guillaume Dumas, Clément Pernet, Wei Zhou

摘要:

我们提出了几个新的时间表Strassen-Winograd的矩阵乘法算法,他们减少了额外的内存分配要求,通过三种不同的手段:通过引入一些预添加,通过对输入矩阵,或通过使用第一递归级的经典乘法。特别是,我们展示了两个完全到位的时间表:一个具有相同数量的操作,如果输入矩阵可以被覆盖;另一个,稍微增加常数的复杂性的前导项,如果输入矩阵是只读的。这些时间表中的许多已经发现了一个基于卵石游戏的穷举搜索算法的实现。

英文摘要:

We propose several new schedules for Strassen-Winograd's matrix multiplication algorithm, they reduce the extra memory allocation requirements by three different means: by introducing a few pre-additions, by overwriting the input matrices, or by using a first recursive level of classical multiplication. In particular, we show two fully in-place schedules: one having the same number of operations, if the input matrices can be overwritten; the other one, slightly increasing the constant of the leading term of the complexity, if the input matrices are read-only. Many of these schedules have been found by an implementation of an exhaustive search algorithm based on a pebble game.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑