Reducing the branch penalty in pipelined processors | IEEE Journals & Magazine | IEEE Xplore