A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services

被引：30

作者：

Putnam, Andrew ^{[1
]}

Caulfield, Adrian M. ^{[1
]}

Chung, Eric S. ^{[1
]}

Chiou, Derek ^{[1
,2
]}

Constantinides, Kypros ^{[3
]}

Demme, John ^{[4
]}

Esmaeilzadeh, Hadi ^{[5
]}

Fowers, Jeremy ^{[1
]}

Gopal, Gopi Prashanth ^{[1
]}

Gray, Jan ^{[1
]}

Haselman, Michael ^{[1
]}

Hauck, Scott ^{[1
,6
]}

Heil, Stephen ^{[1
]}

Hormati, Amir ^{[7
]}

Kim, Joo-Young ^{[1
]}

Lanka, Sitaram ^{[1
]}

Larus, James ^{[8
]}

Peterson, Eric ^{[1
]}

Pope, Simon ^{[1
]}

Smith, Aaron ^{[1
]}

Thong, Jason ^{[1
]}

Xiao, Phillip Yi ^{[1
]}

Burger, Doug ^{[1
]}

机构：

[1] Microsoft, Redmond, WA 98052 USA

[2] Univ Texas Austin, Austin, TX 78712 USA

[3] Amazon Web Serv, Boston, MA USA

[4] Columbia Univ, New York, NY USA

[5] Georgia Inst Technol, Atlanta, GA 30332 USA

[6] Univ Washington, Seattle, WA 98195 USA

[7] Google Inc, Mountain View, CA USA

[8] Ecole Polytech Fed Lausanne, Lausanne, Switzerland

来源：

COMMUNICATIONS OF THE ACM | 2016年 / 59卷 / 11期

关键词：

D O I：

10.1145/2996868

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

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.

引用

页码：114 / 122

页数：9

共 50 条

[11] Parallel Simulation Models for the Evaluation of Future Large-Scale Datacenter Networks
Lugones, Diego
Katrinis, Kostas
Collier, Martin
Theodoropoulos, Georgios
2012 IEEE/ACM 16TH INTERNATIONAL SYMPOSIUM ON DISTRIBUTED SIMULATION AND REAL TIME APPLICATIONS (DS-RT), 2012, : 85 - 92
[12] Joint Progressive Network and Datacenter Recovery After Large-Scale Disasters
Ferdousi, Sifat
Tornatore, Massimo
Dikbiyik, Ferhat
Martel, Charles U.
Xu, Sugang
Hirota, Yusuke
Awaji, Yoshinari
Mukherjee, Biswanath
IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT, 2020, 17 (03): : 1501 - 1514
[13] Decoupling Datacenter Storage Studies from Access to Large-Scale Applications
Delimitrou, Christina
Sankar, Sriram
Vaid, Kushagra
Kozyrakis, Christos
IEEE COMPUTER ARCHITECTURE LETTERS, 2012, 11 (02) : 53 - 56
[14] Accelerating Large-Scale Inference with Anisotropic Vector Quantization
Guo, Ruiqi
Sun, Philip
Lindgren, Erik
Geng, Quan
Simcha, David
Chern, Felix
Kumar, Sanjiv
INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 119, 2020, 119
[15] Dimensioning large-scale membership services
de Vericourt, Francis
Jennings, Otis B.
OPERATIONS RESEARCH, 2008, 56 (01) : 173 - 187
[16] Accelerating large-scale HPC Applications using FPGAs
Dimond, Rob
Racaniere, Sebastien
Pell, Oliver
2011 20TH IEEE SYMPOSIUM ON COMPUTER ARITHMETIC (ARITH-20), 2011, : 191 - 192
[17] ggViz: Accelerating Large-Scale Esports Game Analysis
Xenopoulos P.
Rulff J.
Silva C.
Proceedings of the ACM on Human-Computer Interaction, 2022, 6
[18] Accelerating Large-Scale Statistical Computation With the GOEM Algorithm
Nie, Xiao
Huling, Jared
Qian, Peter Z. G.
TECHNOMETRICS, 2017, 59 (04) : 416 - 425
[19] Accelerating large-scale graph analytics with FPGA and HMC
Khoram, Soroosh
Zhang, Jialiang
Strange, Maxwell
Li, Jing
2017 IEEE 25TH ANNUAL INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM 2017), 2017, : 82 - 82
[20] Accelerating Range Queries for Large-scale Unstructured Meshes
Nguyen, Cuong
Rhodes, Philip J.
2016 IEEE INTERNATIONAL CONFERENCE ON BIG DATA (BIG DATA), 2016, : 502 - 511

← 1 2 3 4 5 →