Available Instruction-Level Parallelism for Superscalar and Superpipelined Machines