YOCO: A Hybrid In-Memory Computing Architecture with 8-bit Sub-PetaOps/W In-Situ Multiply Arithmetic for Large-Scale AI

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Xuan, Zihao, Yang, Yuxuan, Xuan, Wei, Su, Zijia, Chen, Song, Kang, Yi
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910999136698368
author Xuan, Zihao
Yang, Yuxuan
Xuan, Wei
Su, Zijia
Chen, Song
Kang, Yi
author_facet Xuan, Zihao
Yang, Yuxuan
Xuan, Wei
Su, Zijia
Chen, Song
Kang, Yi
contents In this paper, we further explore the potential of analog in-memory computing (AiMC) and introduce an innovative artificial intelligence (AI) accelerator architecture named YOCO, featuring three key proposals: (1) YOCO proposes a novel 8-bit in-situ multiply arithmetic (IMA) achieving 123.8 TOPS/W energy-efficiency and 34.9 TOPS throughput through efficient charge-domain computation and timedomain accumulation mechanism. (2) YOCO employs a hybrid ReRAM-SRAM memory structure to balance computational efficiency and storage density. (3) YOCO tailors an IMC-friendly attention computing flow with an efficient pipeline to accelerate the inference of transformer-based AI models. Compared to three SOTA baselines, YOCO on average improves energy efficiency by up to 3.9x-19.9x and throughput by up to 6.8x-33.6x across 10 CNN/transformer models.
format Preprint
id arxiv_https___arxiv_org_abs_2312_11836
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle YOCO: A Hybrid In-Memory Computing Architecture with 8-bit Sub-PetaOps/W In-Situ Multiply Arithmetic for Large-Scale AI
Xuan, Zihao
Yang, Yuxuan
Xuan, Wei
Su, Zijia
Chen, Song
Kang, Yi
Hardware Architecture
Artificial Intelligence
In this paper, we further explore the potential of analog in-memory computing (AiMC) and introduce an innovative artificial intelligence (AI) accelerator architecture named YOCO, featuring three key proposals: (1) YOCO proposes a novel 8-bit in-situ multiply arithmetic (IMA) achieving 123.8 TOPS/W energy-efficiency and 34.9 TOPS throughput through efficient charge-domain computation and timedomain accumulation mechanism. (2) YOCO employs a hybrid ReRAM-SRAM memory structure to balance computational efficiency and storage density. (3) YOCO tailors an IMC-friendly attention computing flow with an efficient pipeline to accelerate the inference of transformer-based AI models. Compared to three SOTA baselines, YOCO on average improves energy efficiency by up to 3.9x-19.9x and throughput by up to 6.8x-33.6x across 10 CNN/transformer models.
title YOCO: A Hybrid In-Memory Computing Architecture with 8-bit Sub-PetaOps/W In-Situ Multiply Arithmetic for Large-Scale AI
topic Hardware Architecture
Artificial Intelligence
url https://arxiv.org/abs/2312.11836