Design Space Optimization of Shared Memory Architecture in Accelerator-rich Systems

Author:

Sinha Mitali1,Harsha Gade Sri1,Bhattacharyya Pramit1,Deb Sujay1

Affiliation:

1. Indraprastha Institute of Information Technology Delhi, Okhla Phase III, New Delhi, Delhi, India

Abstract

Shared memory architectures, as opposed to private-only memories, provide a viable alternative to meet the ever-increasing memory requirements of multi-accelerator systems to achieve high performance under stringent area and energy constraints. However, an impulsive memory sharing degrades performance due to network contention and latency to access shared memory. We propose the Accelerator Shared Memory (ASM) framework to provide an optimal private/shared memory configuration and shared data allocation under a system’s resource and network constraints. Evaluations show ASM provides up to 34.35% and 31.34% improvement in performance and energy, respectively, over baseline systems.

Publisher

Association for Computing Machinery (ACM)

Subject

Electrical and Electronic Engineering,Computer Graphics and Computer-Aided Design,Computer Science Applications

Reference47 articles.

1. [n.d.]. Synopsis Platform Architect. Retrieved from http://www.eigen.in/pdf/Platform_Architect.pdf. [n.d.]. Synopsis Platform Architect. Retrieved from http://www.eigen.in/pdf/Platform_Architect.pdf.

2. [n.d.]. Synopsis Platform Architect Ultra. Retrieved from https://www.synopsys.com/verification/virtual-prototyping/platform-architect.html. [n.d.]. Synopsis Platform Architect Ultra. Retrieved from https://www.synopsys.com/verification/virtual-prototyping/platform-architect.html.

3. mRTS: Run-time system for reconfigurable processors with multi-grained instruction-set extensions

4. Always-on Vision Processing Unit for Mobile Applications

5. Networks on chips: a new SoC paradigm

Cited by 5 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Simulation-driven Latency Estimations for Multi-core Machine Learning Accelerators;2023 IEEE 5th International Conference on Artificial Intelligence Circuits and Systems (AICAS);2023-06-11

2. DoS Attack Models and Mitigation Frameworks for NoC-Based SoCs;Frontiers of Quality Electronic Design (QED);2022-09-06

3. Experimental research of a shared memory subsystem with limited queue length for specialized reconfigurable multiprocessor systems;Independent Journal of Management & Production;2022-06-01

4. Optimization of Multi-Core Accelerator Performance Based on Accurate Performance Estimation;IEEE Access;2022

5. Sniffer: A Machine Learning Approach for DoS Attack Localization in NoC-Based SoCs;IEEE Journal on Emerging and Selected Topics in Circuits and Systems;2021-06

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3