Towards energy-proportional datacenter memory with mobile DRAM

Author:

Malladi Krishna T.1,Lee Benjamin C.2,Nothaft Frank A.1,Kozyrakis Christos1,Periyathambi Karthika1,Horowitz Mark1

Affiliation:

1. Stanford University

2. Duke University

Abstract

To increase datacenter energy efficiency, we need memory systems that keep pace with processor efficiency gains. Currently, servers use DDR3 memory, which is designed for high bandwidth but not for energy proportionality. A system using 20% of the peak DDR3 bandwidth consumes 2.3x the energy per bit compared to the energy consumed by a system with fully utilized memory bandwidth. Nevertheless, many datacenter applications stress memory capacity and latency but not memory bandwidth. In response, we architect server memory systems using mobile DRAM devices, trading peak bandwidth for lower energy consumption per bit and more efficient idle modes. We demonstrate 3-5x lower memory power, better proportionality, and negligible performance penalties for datacenter workloads.

Publisher

Association for Computing Machinery (ACM)

Cited by 20 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. Unity ECC: Unified Memory Protection Against Bit and Chip Errors;Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis;2023-11-11

2. Simulation analysis on energy consumption and economy of CPU cooling system based on loop heat pipe for data center;Thermal Science and Engineering Progress;2023-10

3. SW-PCM: Graceful Degradation Support in PCM Main Memories by Using Swaption Mechanism;Proceedings of the Future Technologies Conference (FTC) 2023, Volume 3;2023

4. Modeling the correlation between the workload and the power consumed by a server using stochastic and non‐parametric approaches;Software: Practice and Experience;2022-06-27

5. Innovations in the Memory System;Synthesis Lectures on Computer Architecture;2019-09-10

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