Hardware parallelism vs. software parallelism

In this paper, we explore the rationale for multicore parallelism and instead argue that a better use of transistors is to use reconfigurable hardware cores. The difficulty in achieving software parallelism means that new ways of exploiting the silicon real estate need to be explored. Hardware implementations can often expose much finer grained parallelism than possible with software implementations. We discuss some of the challenges from a design and system support perspective.