Abstract
Sequential execution of conditional branches in non-numerical code limits the exploitation of instruction-level parallelism (ILP). In order to cope with this limiation, exploitation of parallelism must be extended to concurrent execution of data and branches in a single cycle.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have