Automatic Insertion of Low Power Annotations in RTL for Pipelined Microprocessors

Provided by: edaa
Topic:
Format: PDF
The authors propose a new technique for low power microprocessor design, a fine-grained clock gating scheme implemented at the RTL or the architectural level which utilizes the program structure of the model. Their algorithm automatically identifies fine-grained blocks of circuit which are not used on any given cycle during the execution of a particular instruction, and shuts them down. This scheme of slicing the circuit based on the instruction being executed is termed instruction-driven slicing. All prior approaches towards analyzing and optimizing RTL and architectural models for lower power dissipation suffer from modeling the power dissipation for very specific hardware structures.

Find By Topic