NL-DPE: An Analog In-memory Non-Linear Dot Product Engine for Efficient CNN and LLM Inference

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhao, Lei, Buonanno, Luca, Gajjar, Archit, Moon, John, Natarajan, Aishwarya, Serebryakov, Sergey, Roth, Ron M., Sheng, Xia, Zhang, Youtao, Faraboschi, Paolo, Ignowski, Jim, Pedretti, Giacomo
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908660592017408
author Zhao, Lei
Buonanno, Luca
Gajjar, Archit
Moon, John
Natarajan, Aishwarya
Serebryakov, Sergey
Roth, Ron M.
Sheng, Xia
Zhang, Youtao
Faraboschi, Paolo
Ignowski, Jim
Pedretti, Giacomo
author_facet Zhao, Lei
Buonanno, Luca
Gajjar, Archit
Moon, John
Natarajan, Aishwarya
Serebryakov, Sergey
Roth, Ron M.
Sheng, Xia
Zhang, Youtao
Faraboschi, Paolo
Ignowski, Jim
Pedretti, Giacomo
contents Resistive Random Access Memory (RRAM) based in-memory computing (IMC) accelerators offer significant performance and energy advantages for deep neural networks (DNNs), but face three major limitations: (1) they support only \textit{static} dot-product operations and cannot accelerate arbitrary non-linear functions or data-dependent multiplications essential to modern LLMs; (2) they demand large, power-hungry analog-to-digital converter (ADC) circuits; and (3) mapping model weights to device conductance introduces errors from cell nonidealities. These challenges hinder scalable and accurate IMC acceleration as models grow. We propose NL-DPE, a Non-Linear Dot Product Engine that overcomes these barriers. NL-DPE augments crosspoint arrays with RRAM-based Analog Content Addressable Memory (ACAM) to execute arbitrary non-linear functions and data-dependent matrix multiplications in the analog domain by transforming them into decision trees, fully eliminating ADCs. To address device noise, NL-DPE uses software-based Noise Aware Fine-tuning (NAF), requiring no in-device calibration. Experiments show that NL-DPE delivers 28X energy efficiency and 249X speedup over a GPU baseline, and 22X energy efficiency and 245X speedup over existing IMC accelerators, while maintaining high accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2511_13950
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle NL-DPE: An Analog In-memory Non-Linear Dot Product Engine for Efficient CNN and LLM Inference
Zhao, Lei
Buonanno, Luca
Gajjar, Archit
Moon, John
Natarajan, Aishwarya
Serebryakov, Sergey
Roth, Ron M.
Sheng, Xia
Zhang, Youtao
Faraboschi, Paolo
Ignowski, Jim
Pedretti, Giacomo
Hardware Architecture
Resistive Random Access Memory (RRAM) based in-memory computing (IMC) accelerators offer significant performance and energy advantages for deep neural networks (DNNs), but face three major limitations: (1) they support only \textit{static} dot-product operations and cannot accelerate arbitrary non-linear functions or data-dependent multiplications essential to modern LLMs; (2) they demand large, power-hungry analog-to-digital converter (ADC) circuits; and (3) mapping model weights to device conductance introduces errors from cell nonidealities. These challenges hinder scalable and accurate IMC acceleration as models grow. We propose NL-DPE, a Non-Linear Dot Product Engine that overcomes these barriers. NL-DPE augments crosspoint arrays with RRAM-based Analog Content Addressable Memory (ACAM) to execute arbitrary non-linear functions and data-dependent matrix multiplications in the analog domain by transforming them into decision trees, fully eliminating ADCs. To address device noise, NL-DPE uses software-based Noise Aware Fine-tuning (NAF), requiring no in-device calibration. Experiments show that NL-DPE delivers 28X energy efficiency and 249X speedup over a GPU baseline, and 22X energy efficiency and 245X speedup over existing IMC accelerators, while maintaining high accuracy.
title NL-DPE: An Analog In-memory Non-Linear Dot Product Engine for Efficient CNN and LLM Inference
topic Hardware Architecture
url https://arxiv.org/abs/2511.13950