Talk:Matrix multiplication

From Rosetta Code

Hints about optimization[edit]

The task looks easy, but between two implementations of this basic function, speed may differ by several orders of magnitude. There are several ways to optimize a matrix product (optimizing cache usage by loop order and block product, transposing, using SIMD processor instructions, OpenMP...). Here is a lecture I like at MIT OpenCourseWare: Matrix Multiply: A Case Study. In real life, one would use an optimized BLAS library like what is found in ATLAS or Intel MKL. Arbautjc (talk) 23:05, 14 August 2016 (UTC)

That depends on the size and structure of the matrices, and also the application, and available resources, and on the people involved, for a variety of reasons. --Rdm (talk) 01:20, 15 August 2016 (UTC)