Invocation-driven Neural Approximate Computing with a Multiclass-Classifier and Multiple Approximators

TL;DR

This paper introduces MCMA, a novel neural approximate computing architecture with multiple approximators and a multiclass classifier, significantly increasing invocation rates and energy efficiency by optimizing the mapping of inputs to approximators.

Contribution

The paper proposes a new MCMA architecture with shared hardware for multiple approximators and a multiclass classifier, enabling higher invocation rates and energy savings.

Findings

Increased invocation rate of approximators.

Enhanced energy efficiency in neural approximate computing.

Effective co-training methods for MCMA architecture.

Abstract

Neural approximate computing gains enormous energy-efficiency at the cost of tolerable quality-loss. A neural approximator can map the input data to output while a classifier determines whether the input data are safe to approximate with quality guarantee. However, existing works cannot maximize the invocation of the approximator, resulting in limited speedup and energy saving. By exploring the mapping space of those target functions, in this paper, we observe a nonuniform distribution of the approximation error incurred by the same approximator. We thus propose a novel approximate computing architecture with a Multiclass-Classifier and Multiple Approximators (MCMA). These approximators have identical network topologies and thus can share the same hardware resource in a neural processing unit(NPU) clip. In the runtime, MCMA can swap in the invoked approximator by merely shipping the synapse weights from the on-chip memory to the buffers near MAC within a cycle. We also propose efficient co-training methods for such MCMA architecture. Experimental results show a more substantial invocation of MCMA as well as the gain of energy-efficiency.