Asynchronous DSP for low-power energy-efficient embedded systems

Authors:
Yiqiong Shi;Bah-Hwee Gwee;Joseph Chang
Affiliations:
School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore;School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore;School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore
Venue:
Microprocessors & Microsystems
Year:
2011

Citing 7
Cited 0

Unifying synchronous/asynchronous state machine synthesis

ICCAD '93 Proceedings of the 1993 IEEE/ACM international conference on Computer-aided design
System-level power optimization: techniques and tools

ACM Transactions on Design Automation of Electronic Systems (TODAES)
A Low-Power, Low-Noise, Configurable Self-Timed DSP

ASYNC '98 Proceedings of the 4th International Symposium on Advanced Research in Asynchronous Circuits and Systems
An Asynchronous Low-Power 80C51 Microcontroller

ASYNC '98 Proceedings of the 4th International Symposium on Advanced Research in Asynchronous Circuits and Systems
The Design of an Asynchronous TinyRISCTM TR4101 Microprocessor Core

ASYNC '98 Proceedings of the 4th International Symposium on Advanced Research in Asynchronous Circuits and Systems
ARM996HS: The First Licensable, Clockless 32-Bit Processor Core

IEEE Micro
Principles of Asynchronous Circuit Design: A Systems Perspective

Principles of Asynchronous Circuit Design: A Systems Perspective

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents the design of an asynchronous DSP that is code compatible with the Motorola DSP56000, with the objective of low power and high energy efficiency to extend the lifespan of the batteries in embedded systems embodying the DSP. It features a unified and low-overhead pipeline flush design, a common-case oriented and efficient single-instruction repeating design, and retimed and single-cycle address generation. The asynchronous DSP is implemented using 130nm CMOS process and is completely standard-cell based. Pre-layout simulation results demonstrate an equivalent speed of 61.5 MIPS and energy dissipation of 54.5@mW/MIPS @ 1.2V running a Radix 2 FFT benchmark program, and a 30.0% energy reduction compared to the clock-gated synchronous counterpart.