Accelerating Matrix Operations with Improved Deeply Pipelined Vector Reduction | IEEE Journals & Magazine | IEEE Xplore