On Dynamic Precision Scaling

TL;DR

This paper introduces Dynamic Precision Scaling (DPS), an adaptive approach that adjusts computational precision during execution to optimize power efficiency while maintaining acceptable accuracy levels.

Contribution

It proposes a novel DPS method that dynamically tunes precision based on phase sensitivity, improving power efficiency without significant accuracy loss.

Findings

DPS reduces power consumption during noise-tolerant phases.

DPS maintains accuracy within bounded limits.

Experimental results show improved energy efficiency.

Abstract

Based on the observation that application phases exhibit varying degrees of sensitivity to noise (i.e., accuracy loss) in computation during execution, this paper explores how Dynamic Precision Scaling (DPS) can maximize power efficiency by tailoring the precision of computation adaptively to temporal changes in algorithmic noise tolerance. DPS can decrease the arithmetic precision of noise-tolerant phases to result in power savings at the same operating speed (or faster execution within the same power budget), while keeping the overall loss in accuracy due to precision reduction bounded.