Saved in:
Bibliographic Details
Main Authors: Zhai, Yifeng, Li, Bing, Yan, Bonan, Wang, Jing
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2401.17582
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911768310185984
author Zhai, Yifeng
Li, Bing
Yan, Bonan
Wang, Jing
author_facet Zhai, Yifeng
Li, Bing
Yan, Bonan
Wang, Jing
contents RRAM crossbars have been studied to construct in-memory accelerators for neural network applications due to their in-situ computing capability. However, prior RRAM-based accelerators show efficiency degradation when executing the popular attention models. We observed that the frequent softmax operations arise as the efficiency bottleneck and also are insensitive to computing precision. Thus, we propose STAR, which boosts the computing efficiency with an efficient RRAM-based softmax engine and a fine-grained global pipeline for the attention models. Specifically, STAR exploits the versatility and flexibility of RRAM crossbars to trade off the model accuracy and hardware efficiency. The experimental results evaluated on several datasets show STAR achieves up to 30.63x and 1.31x computing efficiency improvements over the GPU and the state-of-the-art RRAM-based attention accelerators, respectively.
format Preprint
id arxiv_https___arxiv_org_abs_2401_17582
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle STAR: An Efficient Softmax Engine for Attention Model with RRAM Crossbar
Zhai, Yifeng
Li, Bing
Yan, Bonan
Wang, Jing
Hardware Architecture
RRAM crossbars have been studied to construct in-memory accelerators for neural network applications due to their in-situ computing capability. However, prior RRAM-based accelerators show efficiency degradation when executing the popular attention models. We observed that the frequent softmax operations arise as the efficiency bottleneck and also are insensitive to computing precision. Thus, we propose STAR, which boosts the computing efficiency with an efficient RRAM-based softmax engine and a fine-grained global pipeline for the attention models. Specifically, STAR exploits the versatility and flexibility of RRAM crossbars to trade off the model accuracy and hardware efficiency. The experimental results evaluated on several datasets show STAR achieves up to 30.63x and 1.31x computing efficiency improvements over the GPU and the state-of-the-art RRAM-based attention accelerators, respectively.
title STAR: An Efficient Softmax Engine for Attention Model with RRAM Crossbar
topic Hardware Architecture
url https://arxiv.org/abs/2401.17582