Github user srowen commented on the pull request: https://github.com/apache/spark/pull/131#issuecomment-37493745 Yes I like where this is going. I had started on a similar change, that used gemm instead, since it can compute C <- A*B + C. In fact I had wanted to add an extra API method in jblas to let us call dgemm directly with the same matrix twice, to efficiently compute AtA without transposing A literally. (I submitted a few other small jblas optimizations/fixed that help here, so, the idea of depending on a next jblas release would be of more use than just this method.) Still would have to see what Mikio thinks. What do you think about going that way?
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---