RUCA: RUntime Configurable Approximate Circuits with Self-Correcting Capability

TL;DR

RUCA introduces a runtime configurable approximate circuit framework with self-correcting capability, enabling flexible accuracy-power trade-offs and significant power savings while maintaining specified error bounds.

Contribution

The paper presents a novel approach to design runtime quality-configurable circuits with self-correcting features, improving scalability and flexibility over existing methods.

Findings

RUCA achieves up to 36.57% power savings within 1% error.

RUCA achieves up to 51.32% power savings within 2% error.

The approach outperforms existing quality-configurable methods in benchmarks.

Abstract

Approximate computing is an emerging computing paradigm that offers improved power consumption by relaxing the requirement for full accuracy. Since real-world applications may have different requirements for design accuracy, one trend of approximate computing is to design runtime quality-configurable circuits, which are able to operate under different accuracy modes with different power consumption. In this paper, we present a novel framework RUCA which aims to approximate an arbitrary input circuit in a runtime configurable fashion. By factorizing and decomposing the truth table, our approach aims to approximate and separate the input circuit into multiple configuration blocks which support different accuracy levels, including a corrector circuit to restore full accuracy. By activating different blocks, the approximate circuit is able to operate at different accuracy-power configurations. To improve the scalability of our algorithm, we also provide a design space exploration scheme with circuit partitioning to navigate the search space of possible approximations of subcircuits during design time. We thoroughly evaluate our methodology on a set of benchmarks and compare against another quality-configurable approach, showcasing the benefits and flexibility of RUCA. For 3-level designs, RUCA saves power consumption by 36.57% within 1% error and by 51.32% within 2% error on average.