Saved in:
Bibliographic Details
Main Author: Raja, Tejas
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.22595
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914997623324672
author Raja, Tejas
author_facet Raja, Tejas
contents The paper discusses how Systolic Arrays can improve matrix multiplication for deep neural networks (DNNs). With AI models like OpenAI's GPT now containing trillions of parameters, the need for efficient matrix multiplication is more critical than ever. In this paper, the three main systolic array data flows: Weight Stationary (WS), Input Stationary (IS), and Output Stationary (OS) are discussed. Each data flow's energy consumption and efficiency across various matrix sizes are calculated using the SCALE-Sim simulator. The results show that selecting the right data flow for specific matrix configurations can drastically reduce energy consumption. The conclusions provide helpful insights into optimizing hardware for AI and machine learning applications, offering potential improvements in designing energy-efficient DNN accelerators.
format Preprint
id arxiv_https___arxiv_org_abs_2410_22595
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Systolic Array Data Flows for Efficient Matrix Multiplication in Deep Neural Networks
Raja, Tejas
Hardware Architecture
The paper discusses how Systolic Arrays can improve matrix multiplication for deep neural networks (DNNs). With AI models like OpenAI's GPT now containing trillions of parameters, the need for efficient matrix multiplication is more critical than ever. In this paper, the three main systolic array data flows: Weight Stationary (WS), Input Stationary (IS), and Output Stationary (OS) are discussed. Each data flow's energy consumption and efficiency across various matrix sizes are calculated using the SCALE-Sim simulator. The results show that selecting the right data flow for specific matrix configurations can drastically reduce energy consumption. The conclusions provide helpful insights into optimizing hardware for AI and machine learning applications, offering potential improvements in designing energy-efficient DNN accelerators.
title Systolic Array Data Flows for Efficient Matrix Multiplication in Deep Neural Networks
topic Hardware Architecture
url https://arxiv.org/abs/2410.22595