Look-Up Table based Neural Network Hardware

TL;DR

This paper presents a CMOS LUT-based neural accelerator that significantly reduces power, area, and latency compared to traditional digital and analog approaches, while maintaining accuracy across various models.

Contribution

It introduces a scalable, low-power LUT-based neural accelerator with a divide-and-conquer approach for high-precision MAC operations, outperforming conventional digital and analog methods.

Findings

Achieves up to 29.54× lower area and 3.34× lower energy than traditional LUT-based techniques.

Reduces area by up to 1.23× and energy by 1.80× compared to conventional digital MACs without accuracy loss.

Demonstrates significant area and energy savings across multiple models with minimal accuracy impact.

Abstract

Traditional digital implementations of neural accelerators are limited by high power and area overheads, while analog and non-CMOS implementations suffer from noise, device mismatch, and reliability issues. This paper introduces a CMOS Look-Up Table (LUT)-based Neural Accelerator (LUT-NA) framework that reduces the power, latency, and area consumption of traditional digital accelerators through pre-computed, faster look-ups while avoiding noise and mismatch of analog circuits. To solve the scalability issues of conventional LUT-based computation, we split the high-precision multiply and accumulate (MAC) operations into lower-precision MACs using a divide-and-conquer-based approach. We show that LUT-NA achieves up to $29.54\times$ lower area with $3.34\times$ lower energy per inference task than traditional LUT-based techniques and up to $1.23\times$ lower area with $1.80\times$ lower energy per inference task than conventional digital MAC-based techniques (Wallace Tree/Array Multipliers) without retraining and without affecting accuracy, even on lottery ticket pruned (LTP) models that already reduce the number of required MAC operations by up to 98%. Finally, we introduce mixed precision analysis in LUT-NA framework for various LTP models (VGG11, VGG19, Resnet18, Resnet34, GoogleNet) that achieved up to $32.22\times$-$50.95\times$ lower area across models with $3.68\times$-$6.25\times$ lower energy per inference than traditional LUT-based techniques, and up to $1.35\times$-$2.14\times$ lower area requirement with $1.99\times$-$3.38\times$ lower energy per inference across models as compared to conventional digital MAC-based techniques with $\sim$1% accuracy loss.