research-article

Free access

A reconfigurable fabric for accelerating large-scale datacenter services

Communications of the ACM, Volume 59, Issue 11

Pages 114 - 122

https://doi.org/10.1145/2996868

Published: 28 October 2016 Publication History

All formats PDF

Abstract

Datacenter workloads demand high computational capabilities, flexibility, power efficiency, and low cost. It is challenging to improve all of these factors simultaneously. To advance datacenter capabilities beyond what commodity server designs can provide, we designed and built a composable, reconfigurable hardware fabric based on field programmable gate arrays (FPGA). Each server in the fabric contains one FPGA, and all FPGAs within a 48-server rack are interconnected over a low-latency, high-bandwidth network.

We describe a medium-scale deployment of this fabric on a bed of 1632 servers, and measure its effectiveness in accelerating the ranking component of the Bing web search engine. We describe the requirements and architecture of the system, detail the critical engineering challenges and solutions needed to make the system robust in the presence of failures, and measure the performance, power, and resilience of the system. Under high load, the large-scale reconfigurable fabric improves the ranking throughput of each server by 95% at a desirable latency distribution or reduces tail latency by 29% at a fixed throughput. In other words, the reconfigurable fabric enables the same throughput using only half the number of servers.

References

[1]

Altera. Nios II Processor Reference Handbook, 13.1.0 edition, 2014.

[2]

Altera. Stratix V Device Handbook, 14.01.10 edition, 2014.

[3]

Baxter, R., Booth, S., Bull, M., Cawood, G., Perry, J., Parsons, M., Simpson, A., Trew, A., Mccormick, A., Smart, G., Smart, R., Cantle, A., Chamberlain, R., Genest, G. Maxwell -- A 64 FPGA Supercomputer. Eng. Lett. 16 (2008), 426--433, 2008.

[4]

BEECube. BEE4 Hardware Platform, 1.0 edition, 2011.

[5]

Blott, M., Vissers, K. Dataflow architectures for 10Gbps line-rate key-value stores. In HotChips 2013 (August 2013).

[6]

Convey. The Convey HC-2 Computer, conv-12-030.2 edition, 2012.

[7]

Cray. Cray XD1 Datasheet, 1.3 edition, 2005.

[8]

Dennard, R., Rideout, V., Bassous, E., LeBlanc, A. Design of ion-implanted MOSFET's with very small physical dimensions. IEEE J. Solid-State Circ. 9, 5 (Oct. 1974), 256--268.

[9]

Estlick, M., Leeser, M., Theiler, J., Szymanski, J.J. Algorithmic transformations in the implementation of K-means clustering on reconfigurable hardware. In Proceedings of the 2001 ACM/SIGDA Ninth International Symposium on Field Programmable Gate Arrays, FPGA'01 (New York, NY, USA, 2001). ACM.

Digital Library

[10]

Gens, F. Worldwide and Regional Public IT Cloud Services 2014--2018 Forecast (Oct. 2014).

[11]

George, A., Lam, H., Stitt, G. Novo-G: At the forefront of scalable reconfigurable supercomputing. Comput. Sci. Eng. 13, 1 (2011), 82--86.

Digital Library

[12]

Hussain, H.M., Benkrid, K., Erdogan, A.T., Seker, H. Highly parameterized K-means clustering on FPGAs: Comparative results with GPPs and GPUs. In Proceedings of the 2011 International Conference on Reconfigurable Computing and FPGAs, RECONFIG'11 (Washington, DC, USA, 2011). IEEE Computer Society.

Digital Library

[13]

IBM. IBM PureData System for Analytics N2001, WAD12353-USEN-01 edition, 2013.

[14]

Lavasani, M., Angepat, H., Chiou, D. An FPGA-based in-line accelerator for memcached. Comput. Arch. Lett. PP, 99 (2013), 1--1.

Digital Library

[15]

Martin, A., Jamsek, D., Agarawal, K. FPGA-based application acceleration: Case study with GZIP compression/decompression streaming engine. In ICCAD Special Session 7C (November 2013).

[16]

Microsoft. How Microsoft Designs Its Cloud-Scale Servers, 2014.

[17]

Pell, O., Mencer, O. Surviving the end of frequency scaling with reconfigurable dataflow computing. SIGARCH Comput. Archit. News 39, 4 (Dec. 2011).

Digital Library

[18]

Showerman, M., Enos, J., Pant, A., Kindratenko, V., Steffen, C., Pennington, R., Hwu, W. QP: A Heterogeneous Multi-accelerator Cluster. 2009.

[19]

SRC. MAPstation Systems, 70000 AH edition, 2014.

[20]

Yan, J., Zhao, Z.-X., Xu, N.-Y., Jin, X., Zhang, L.-T., Hsu, F.-H. Efficient query processing for web search engine with FPGAs. In Proceedings of the 2012 IEEE 20th International Symposium on Field-Programmable Custom Computing Machines, FCCM'12 (Washington, DC, USA, 2012). IEEE Computer Society.

Digital Library

Cited By

Zheng HZhao MZhang HXiong YCai XJia Z(2024)Towards Efficient Reconfiguration through Lightweight Input Inversion for MLC NVFPGAs2024 Design, Automation & Test in Europe Conference & Exhibition (DATE)10.23919/DATE58400.2024.10546880(1-6)Online publication date: 25-Mar-2024
https://doi.org/10.23919/DATE58400.2024.10546880
Zeno LChen ASilberstein MSekar VYu MSeneviratne AVeitch D(2024)In-Network Address Caching for Virtual NetworksProceedings of the ACM SIGCOMM 2024 Conference10.1145/3651890.3672213(735-749)Online publication date: 4-Aug-2024
https://dl.acm.org/doi/10.1145/3651890.3672213
Tang SYu YWang HWang GChen WXu ZGuo SGao W(2024)A Survey on Scheduling Techniques in Computing and Network ConvergenceIEEE Communications Surveys & Tutorials10.1109/COMST.2023.332902726:1(160-195)Online publication date: 1-Jan-2024
https://dl.acm.org/doi/10.1109/COMST.2023.3329027
Show More Cited By

Index Terms

A reconfigurable fabric for accelerating large-scale datacenter services
1. Hardware
  1. Integrated circuits
    1. Interconnect
  2. Very large scale integration design
    1. Application-specific VLSI designs
2. Software and its engineering

Recommendations

A reconfigurable fabric for accelerating large-scale datacenter services
ISCA '14

Datacenter workloads demand high computational capabilities, flexibility, power efficiency, and low cost. It is challenging to improve all of these factors simultaneously. To advance datacenter capabilities beyond what commodity server designs can ...
A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services
To advance datacenter capabilities beyond what commodity server designs can provide, the authors designed and built a composable, reconfigurable fabric to accelerate large-scale software services. Each instantiation of the fabric consists of a 6 x 8 2D ...
A reconfigurable fabric for accelerating large-scale datacenter services
ISCA '14: Proceeding of the 41st annual international symposium on Computer architecuture

Datacenter workloads demand high computational capabilities, flexibility, power efficiency, and low cost. It is challenging to improve all of these factors simultaneously. To advance datacenter capabilities beyond what commodity server designs can ...

Comments

Please enable JavaScript to view thecomments powered by Disqus.

Information & Contributors

Information

Published In

cover image Communications of the ACM

Communications of the ACM Volume 59, Issue 11

November 2016

118 pages

ISSN:0001-0782

EISSN:1557-7317

DOI:10.1145/3013530

Editor:
Moshe Y. Vardi
Association for Computing Machinery, New York, NY

Issue’s Table of Contents

Copyright © 2016 ACM.

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected].

Publisher

Association for Computing Machinery

New York, NY, United States

Publication History

Published: 28 October 2016

Published in CACM Volume 59, Issue 11

Permissions

Request permissions for this article.

Request Permissions

Check for updates

Qualifiers

Research-article
Research
Refereed

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

32
Total Citations
View Citations
5,697
Total Downloads

Downloads (Last 12 months)320
Downloads (Last 6 weeks)26

Reflects downloads up to 24 Sep 2024

Other Metrics

View Author Metrics

Citations

Cited By

Zheng HZhao MZhang HXiong YCai XJia Z(2024)Towards Efficient Reconfiguration through Lightweight Input Inversion for MLC NVFPGAs2024 Design, Automation & Test in Europe Conference & Exhibition (DATE)10.23919/DATE58400.2024.10546880(1-6)Online publication date: 25-Mar-2024
https://doi.org/10.23919/DATE58400.2024.10546880
Zeno LChen ASilberstein MSekar VYu MSeneviratne AVeitch D(2024)In-Network Address Caching for Virtual NetworksProceedings of the ACM SIGCOMM 2024 Conference10.1145/3651890.3672213(735-749)Online publication date: 4-Aug-2024
https://dl.acm.org/doi/10.1145/3651890.3672213
Tang SYu YWang HWang GChen WXu ZGuo SGao W(2024)A Survey on Scheduling Techniques in Computing and Network ConvergenceIEEE Communications Surveys & Tutorials10.1109/COMST.2023.332902726:1(160-195)Online publication date: 1-Jan-2024
https://dl.acm.org/doi/10.1109/COMST.2023.3329027
Wang JZhang QRong HXu GKim MChandra SBlincoe KTonella P(2023)Leveraging Hardware Probes and Optimizations for Accelerating Fuzz Testing of Heterogeneous ApplicationsProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering10.1145/3611643.3616318(1101-1113)Online publication date: 30-Nov-2023
https://dl.acm.org/doi/10.1145/3611643.3616318
Landgraf JGiordano MYoon ERossbach CAamodt TJerger NSwift M(2023)Reconfigurable Virtual Memory for FPGA-Driven I/OProceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 310.1145/3582016.3582048(556-571)Online publication date: 25-Mar-2023
https://dl.acm.org/doi/10.1145/3582016.3582048
Lyerly RBilbao CMin CRossbach CRavindran B(2022)An OpenMP Runtime for Transparent Work Sharing across Cache-Incoherent Heterogeneous NodesACM Transactions on Computer Systems10.1145/350522439:1-4(1-30)Online publication date: 5-Jul-2022
https://dl.acm.org/doi/10.1145/3505224
Eran HFudim MMalka GShalom GCohen NHermony ALevi DLiss LSilberstein MFalsafi BFerdman MLu SWenisch T(2022)FlexDriver: a network driver for your acceleratorProceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems10.1145/3503222.3507776(1115-1129)Online publication date: 28-Feb-2022
https://dl.acm.org/doi/10.1145/3503222.3507776
Jain AMaskell DFahmy S(2022)Coarse Grained FPGA Overlay for Rapid Just-In-Time Accelerator CompilationIEEE Transactions on Parallel and Distributed Systems10.1109/TPDS.2021.311685933:6(1478-1490)Online publication date: 1-Jun-2022
https://doi.org/10.1109/TPDS.2021.3116859
Zheng HZhang HXu SXu FZhao M(2022)Adaptive Mode Transformation for Wear Leveling in Nonvolatile FPGAsIEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems10.1109/TCAD.2022.319768541:11(3591-3601)Online publication date: 1-Nov-2022
https://dl.acm.org/doi/10.1109/TCAD.2022.3197685
Vu HTran DBui NLe TNguyen H(2022)Mapping Boolean Functions onto Lookup-Tables on FPGAs2022 RIVF International Conference on Computing and Communication Technologies (RIVF)10.1109/RIVF55975.2022.10013797(508-512)Online publication date: 20-Dec-2022
https://doi.org/10.1109/RIVF55975.2022.10013797
Show More Cited By

View Options

View options

PDF

View or Download as a PDF file.

eReader

View online with eReader.

Digital Edition

View this article in digital edition.

Digital Edition

Magazine Site

View this article on the magazine site (external)

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Full Access

Get this Article

Media

Figures

Other

Tables

View Issue’s Table of Contents