Implementation Analysis of Fast Matrix Multiplication Algorithms on Shared Memory Computers
Elisa Francomano, Alessandra Macaluso, Marián Vajteršic · 1995
The paper presents analysis of matrix multiplication algorithms from the point of view of their efficient implementation on shared memory machines. The algorithms of Winograd and Strassen have been analyzed and implemented considering the memory accesses, the stride and data transfer overheads. A particular attention is paid to the parallel implementation of the modified recurrence-free variant of the Strassen's algorithm.