RECONFIGURING THE CARRY LOOK-AHEAD ADDER USING APPLICATION BEHAVIOR IN EMBEDDED PROCESSORS

摘要

In this work we introduce a reconfigurable carry look-ahead adder (or simply RCLA). RCLA is an adder that enhances performance by avoiding aggressive usage of a 32-bit conventional Cany Look-Ahead adder for add operations using small operands. Instead, for small size operands, RCLA performs its computations using just a portion of its resources and therefore in less number of cycles. We evaluate RCLA using a subset of MiBench embedded benchmarks. Our evaluations show that RCLA improves average performance by 5%.