research-article

Free access

ASIC clouds: specializing the datacenter for planet-scale applications

Authors:

Michael Bedford Taylor,

Moein Khazraee,

Scott Davidson,

Dustin RichmondAuthors Info & Claims

Communications of the ACM, Volume 63, Issue 7

Pages 103 - 109

https://doi.org/10.1145/3399734

Published: 18 June 2020 Publication History

All formats PDF

Abstract

Planet-scale applications are driving the exponential growth of the Cloud, and datacenter specialization is the key enabler of this trend. GPU- and FPGA-based clouds have already been deployed to accelerate compute-intensive workloads. ASIC-based clouds are a natural evolution as cloud services expand across the planet. ASIC Clouds are purpose-built datacenters comprised of large arrays of ASIC accelerators that optimize the total cost of ownership (TCO) of large, high-volume scale-out computations. On the surface, ASIC Clouds may seem improbable due to high NREs and ASIC inflexibility, but large-scale ASIC Clouds have already been deployed for the Bitcoin cryptocurrency system. This paper distills lessons from these Bitcoin ASIC Clouds and applies them to other large scale workloads such as YouTube-style video-transcoding and Deep Learning, showing superior TCO versus CPU and GPU. It derives Pareto-optimal ASIC Cloud servers based on accelerator properties, by jointly optimizing ASIC architecture, DRAM, motherboard, power delivery, cooling, and operating voltage. Finally, the authors examine the impact of ASIC NRE and when it makes sense to build an ASIC Cloud.

References

[1]

ASIC clouds: Specializing the datacenter. UCSD CSE Tech Report CS2016-1016. May 8, 2016. https://csetechrep.ucsd.edu/Dienst/UI/2.0/Describe/ncstrl.ucsd_cse/CS2016-1016.

[2]

Esmaeilzadeh, H., Blem, E., Amant, R.S., Sankaralingam, K., Burger, D. Power limitations and dark silicon are challenging the future of multicore. In TOCS, 2012.

[3]

Goulding, N., et al. GreenDroid: A mobile application processor for a future of dark silicon. In HOTCHIPS, 2010.

[4]

Goulding-Hotta, N., Sampson, J., Venkatesh, G., Garcia, S., Auricchio, J., Huang, P.-C., Arora, M., Nath, S., Bhatt, V., Babb, J., Swanson, S., Taylor, M.B. The greendroid mobile application processor: An architecture for silicon's dark future. IEEE Micro 2, 31 (2011), 86--95.

[5]

Jouppi, N.P., et al. In-datacenter performance analysis of a tensor processing unit. In International Symposium on Computer Architecture (ISCA), 2017.

Digital Library

[6]

Khazraee, M., et al. Moonwalk: NRE optimization in ASIC clouds. In International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2017.

Digital Library

[7]

Khazraee, M., Vega, L., Magaki, I., Taylor, M. Specializing a planet's computation: ASIC clouds. IEEE Micro, May 2017.

Digital Library

[8]

Magaki, I., et al. ASIC clouds: Specializing the datacenter. In International Symposium on Computer Architecture (ISCA), 2016.

[9]

Putnam et al. A reconfigurable fabric for accelerating large-scale datacenter services. In International Symposium on Computer Architecture (ISCA), 2014.

Digital Library

[10]

Sampson, J., Venkatesh, G., Goulding-Hotta, N., Garcia, S., Swanson, S., Taylor, M.B. Efficient complex operators for irregular codes. In HPCA, 2011.

[11]

Shafique, M., Garg, S., Henkel, J., Marculescu, D. The EDA challenges in the dark silicon era: Temperature, reliability, and variability perspectives. In DAC, 2014.

Digital Library

[12]

Weston, S. FPGA accelerators at JP Morgan chase, 2011. Stanford Computer Systems Colloquium, https://www.youtube.com/watch?v=9NqX1ETADn0.

[13]

Taylor, M. A landscape of the new dark silicon design regime. IEEE Micro, Sept-Oct. 2013.

[14]

Taylor, M.B. Is dark silicon useful? Harnessing the four horesemen of the coming dark silicon apocalypse. In DAC, 2012.

[15]

Venkatesh, G., Sampson, J., Goulding, N., Garcia, S., Bryksin, V., Lugo-Martinez, J., Swanson, S., Taylor, M.B. Conservation cores: Reducing the energy of mature computations. In Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2010.

[16]

Venkatesh and others. Qscores: Configurable co-processors to trade dark silicon for energy efficiency in a scalable manner. In MICRO, 2011.

[17]

Xie, S., Davidson, S., Magaki, I., Khazraee, M., Vega, L., Zhang, L., Taylor, M.B. Extreme datacenter specialization for planet-scale computing: Asic clouds. SIGOPS Oper. Syst. Rev. 1, 52 (2018), 96--108.

Cited By

Wyeth RRella LAtkins E(2023)The material geographies of Bitfury in Georgia: Integrating cryptoasset firms into global financial networksEnvironment and Planning A: Economy and Space10.1177/0308518X23121178956:3(816-832)Online publication date: 30-Nov-2023
https://doi.org/10.1177/0308518X231211789
Mokhtari ARawls DHuynh TGreen JSalehi M(2023)E2C: A Visual Simulator to Reinforce Education of Heterogeneous Computing Systems2023 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)10.1109/IPDPSW59300.2023.00052(270-277)Online publication date: May-2023
https://doi.org/10.1109/IPDPSW59300.2023.00052
Yadav ARizkalla MYtterdal TLee JBalasubramanian LGopinath A(2022)Evaluation of FinFET in Ultra Low Power ALU2022 IEEE International Conference on Design & Test of Integrated Micro & Nano-Systems (DTS)10.1109/DTS55284.2022.9809864(01-04)Online publication date: 6-Jun-2022
https://doi.org/10.1109/DTS55284.2022.9809864
Show More Cited By

Index Terms

ASIC clouds: specializing the datacenter for planet-scale applications
1. General and reference
  1. Cross-computing tools and techniques
    1. Performance
2. Networks
  1. Network performance evaluation

Recommendations

ASIC clouds: specializing the datacenter
ISCA'16

GPU and FPGA-based clouds have already demonstrated the promise of accelerating computing-intensive workloads with greatly improved power and performance.

In this paper, we examine the design of ASIC Clouds, which are purpose-built datacenters comprised ...
ASIC clouds: specializing the datacenter
ISCA '16: Proceedings of the 43rd International Symposium on Computer Architecture

GPU and FPGA-based clouds have already demonstrated the promise of accelerating computing-intensive workloads with greatly improved power and performance.

In this paper, we examine the design of ASIC Clouds, which are purpose-built datacenters comprised ...
Specializing a Planet's Computation: ASIC Clouds
GPU- and FPGA-based clouds have been deployed to accelerate computationally intensive workloads. ASIC-based clouds are a natural evolution as cloud services expand across the planet. ASIC Clouds are purpose-built datacenters comprising large arrays of ...

Comments

Information & Contributors

Information

Published In

cover image Communications of the ACM

Communications of the ACM Volume 63, Issue 7

July 2020

102 pages

ISSN:0001-0782

EISSN:1557-7317

DOI:10.1145/3407166

Editor:
Andrew A. Chien
Association for Computing Machinery, New York, NY

Issue’s Table of Contents

Copyright © 2020 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: 18 June 2020

Published in CACM Volume 63, Issue 7

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

5
Total Citations
View Citations
8,686
Total Downloads

Downloads (Last 12 months)373
Downloads (Last 6 weeks)40

Reflects downloads up to 03 Jan 2025

Other Metrics

View Author Metrics

Citations

Cited By

Wyeth RRella LAtkins E(2023)The material geographies of Bitfury in Georgia: Integrating cryptoasset firms into global financial networksEnvironment and Planning A: Economy and Space10.1177/0308518X23121178956:3(816-832)Online publication date: 30-Nov-2023
https://doi.org/10.1177/0308518X231211789
Mokhtari ARawls DHuynh TGreen JSalehi M(2023)E2C: A Visual Simulator to Reinforce Education of Heterogeneous Computing Systems2023 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)10.1109/IPDPSW59300.2023.00052(270-277)Online publication date: May-2023
https://doi.org/10.1109/IPDPSW59300.2023.00052
Yadav ARizkalla MYtterdal TLee JBalasubramanian LGopinath A(2022)Evaluation of FinFET in Ultra Low Power ALU2022 IEEE International Conference on Design & Test of Integrated Micro & Nano-Systems (DTS)10.1109/DTS55284.2022.9809864(01-04)Online publication date: 6-Jun-2022
https://doi.org/10.1109/DTS55284.2022.9809864
Kotlar MPunt MMilutinović V(2022)Energy efficient implementation of tensor operations using dataflow paradigm for machine learning10.1016/bs.adcom.2021.11.011(151-199)Online publication date: 2022
https://doi.org/10.1016/bs.adcom.2021.11.011
Taylor MXie Y(2020)Your agile open source HW stinks (because it is not a system)Proceedings of the 39th International Conference on Computer-Aided Design10.1145/3400302.3415791(1-6)Online publication date: 2-Nov-2020
https://dl.acm.org/doi/10.1145/3400302.3415791

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)

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