Simultaneous multithreaded vector architecture: merging ILP and DLP for high performance | Publicación