ReDy: A Novel ReRAM-centric Dynamic Quantization Approach for Energy-efficient CNN Inference

TL;DR

ReDy introduces a ReRAM-centric dynamic quantization method for CNN inference that adaptively reduces energy consumption by optimizing activation precision, significantly improving energy efficiency with minimal accuracy loss.

Contribution

The paper presents a novel hardware accelerator that dynamically adjusts activation quantization levels based on statistical analysis, enhancing energy efficiency in ReRAM-based DNN accelerators.

Findings

ReDy achieves 13% average energy savings over existing accelerators.

It reduces the number of analog-to-digital conversions needed.

ReDy maintains negligible accuracy loss with minimal area overhead.

Abstract

The primary operation in DNNs is the dot product of quantized input activations and weights. Prior works have proposed the design of memory-centric architectures based on the Processing-In-Memory (PIM) paradigm. Resistive RAM (ReRAM) technology is especially appealing for PIM-based DNN accelerators due to its high density to store weights, low leakage energy, low read latency, and high performance capabilities to perform the DNN dot-products massively in parallel within the ReRAM crossbars. However, the main bottleneck of these architectures is the energy-hungry analog-to-digital conversions (ADCs) required to perform analog computations in-ReRAM, which penalizes the efficiency and performance benefits of PIM. To improve energy-efficiency of in-ReRAM analog dot-product computations we present ReDy, a hardware accelerator that implements a ReRAM-centric Dynamic quantization scheme to take advantage of the bit serial streaming and processing of activations. The energy consumption of ReRAM-based DNN accelerators is directly proportional to the numerical precision of the input activations of each DNN layer. In particular, ReDy exploits that activations of CONV layers from Convolutional Neural Networks (CNNs), a subset of DNNs, are commonly grouped according to the size of their filters and the size of the ReRAM crossbars. Then, ReDy quantizes on-the-fly each group of activations with a different numerical precision based on a novel heuristic that takes into account the statistical distribution of each group. Overall, ReDy greatly reduces the activity of the ReRAM crossbars and the number of A/D conversions compared to an static 8-bit uniform quantization. We evaluate ReDy on a popular set of modern CNNs. On average, ReDy provides 13\% energy savings over an ISAAC-like accelerator with negligible accuracy loss and area overhead.