TimeCube

A manycore embedded processor with interference-agnostic progress tracking

Anshuman Gupta, John Morgan Sampson, Michael Bedford Taylor

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

Recently introduced processors such as Tilera's Tile Gx100 and Intel's 48-core SCC have delivered on the promise of high performance per watt in manycore processors, making these architectures ostensibly as attractive for low-power embedded processors as for cloud services. However, these architectures space-multiplex the microarchitectural resources between many threads to increase utilization, which leads to potentially large and varying levels of interference. This decorrelates CPU-time from actual application progress and decreases the ability of traditional software to accurately track and finely control application progress, hindering the adoption of manycore processors in embedded computing. In this paper we propose Progress Time as the counterpart of CPU-time in space-multiplexed systems and show how it can be used to track application progress. We also introduce TimeCube, a manycore embedded processor that uses dynamic execution isolation and shadow performance modeling to provide an accurate online measurement of each application's Progress Time. Our evaluation shows that a 32-core TimeCube processor can track application progress with less than 1% error even in the presence of a 6× average worst-case slowdown. TimeCube also uses Progress Times to perform online architectural resource management that leads to a 36% improvement in throughput compared to existing microarchitectural resource allocation schemes. Overall, the results argue for adding the requisite microarchitectural structures to support Progress Time in manycore chips for embedded systems.

Original languageEnglish (US)
Title of host publicationProceedings - 2013 International Conference on Embedded Computer Systems
Subtitle of host publicationArchitectures, Modeling and Simulation, IC-SAMOS 2013
PublisherIEEE Computer Society
Pages227-236
Number of pages10
ISBN (Print)9781479901036
DOIs
StatePublished - Jan 1 2013
Event2013 13th International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013 - Samos, Greece
Duration: Jul 15 2013Jul 18 2013

Publication series

NameProceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013

Other

Other2013 13th International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013
CountryGreece
CitySamos
Period7/15/137/18/13

Fingerprint

Embedded Processor
Many-core
Interference
CPU Time
Program processors
On-line Measurement
Performance Modeling
Tile
Resource Management
Embedded systems
Embedded Systems
Thread
Resource Allocation
Resource allocation
Isolation
Chip
Throughput
High Performance
Decrease
Resources

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Electrical and Electronic Engineering
  • Modeling and Simulation

Cite this

Gupta, A., Sampson, J. M., & Taylor, M. B. (2013). TimeCube: A manycore embedded processor with interference-agnostic progress tracking. In Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013 (pp. 227-236). [6621127] (Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013). IEEE Computer Society. https://doi.org/10.1109/SAMOS.2013.6621127
Gupta, Anshuman ; Sampson, John Morgan ; Taylor, Michael Bedford. / TimeCube : A manycore embedded processor with interference-agnostic progress tracking. Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013. IEEE Computer Society, 2013. pp. 227-236 (Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013).
@inproceedings{e3a7c7e62b00475ca56f13db947f5c9b,
title = "TimeCube: A manycore embedded processor with interference-agnostic progress tracking",
abstract = "Recently introduced processors such as Tilera's Tile Gx100 and Intel's 48-core SCC have delivered on the promise of high performance per watt in manycore processors, making these architectures ostensibly as attractive for low-power embedded processors as for cloud services. However, these architectures space-multiplex the microarchitectural resources between many threads to increase utilization, which leads to potentially large and varying levels of interference. This decorrelates CPU-time from actual application progress and decreases the ability of traditional software to accurately track and finely control application progress, hindering the adoption of manycore processors in embedded computing. In this paper we propose Progress Time as the counterpart of CPU-time in space-multiplexed systems and show how it can be used to track application progress. We also introduce TimeCube, a manycore embedded processor that uses dynamic execution isolation and shadow performance modeling to provide an accurate online measurement of each application's Progress Time. Our evaluation shows that a 32-core TimeCube processor can track application progress with less than 1{\%} error even in the presence of a 6× average worst-case slowdown. TimeCube also uses Progress Times to perform online architectural resource management that leads to a 36{\%} improvement in throughput compared to existing microarchitectural resource allocation schemes. Overall, the results argue for adding the requisite microarchitectural structures to support Progress Time in manycore chips for embedded systems.",
author = "Anshuman Gupta and Sampson, {John Morgan} and Taylor, {Michael Bedford}",
year = "2013",
month = "1",
day = "1",
doi = "10.1109/SAMOS.2013.6621127",
language = "English (US)",
isbn = "9781479901036",
series = "Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013",
publisher = "IEEE Computer Society",
pages = "227--236",
booktitle = "Proceedings - 2013 International Conference on Embedded Computer Systems",
address = "United States",

}

Gupta, A, Sampson, JM & Taylor, MB 2013, TimeCube: A manycore embedded processor with interference-agnostic progress tracking. in Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013., 6621127, Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013, IEEE Computer Society, pp. 227-236, 2013 13th International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013, Samos, Greece, 7/15/13. https://doi.org/10.1109/SAMOS.2013.6621127

TimeCube : A manycore embedded processor with interference-agnostic progress tracking. / Gupta, Anshuman; Sampson, John Morgan; Taylor, Michael Bedford.

Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013. IEEE Computer Society, 2013. p. 227-236 6621127 (Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - TimeCube

T2 - A manycore embedded processor with interference-agnostic progress tracking

AU - Gupta, Anshuman

AU - Sampson, John Morgan

AU - Taylor, Michael Bedford

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Recently introduced processors such as Tilera's Tile Gx100 and Intel's 48-core SCC have delivered on the promise of high performance per watt in manycore processors, making these architectures ostensibly as attractive for low-power embedded processors as for cloud services. However, these architectures space-multiplex the microarchitectural resources between many threads to increase utilization, which leads to potentially large and varying levels of interference. This decorrelates CPU-time from actual application progress and decreases the ability of traditional software to accurately track and finely control application progress, hindering the adoption of manycore processors in embedded computing. In this paper we propose Progress Time as the counterpart of CPU-time in space-multiplexed systems and show how it can be used to track application progress. We also introduce TimeCube, a manycore embedded processor that uses dynamic execution isolation and shadow performance modeling to provide an accurate online measurement of each application's Progress Time. Our evaluation shows that a 32-core TimeCube processor can track application progress with less than 1% error even in the presence of a 6× average worst-case slowdown. TimeCube also uses Progress Times to perform online architectural resource management that leads to a 36% improvement in throughput compared to existing microarchitectural resource allocation schemes. Overall, the results argue for adding the requisite microarchitectural structures to support Progress Time in manycore chips for embedded systems.

AB - Recently introduced processors such as Tilera's Tile Gx100 and Intel's 48-core SCC have delivered on the promise of high performance per watt in manycore processors, making these architectures ostensibly as attractive for low-power embedded processors as for cloud services. However, these architectures space-multiplex the microarchitectural resources between many threads to increase utilization, which leads to potentially large and varying levels of interference. This decorrelates CPU-time from actual application progress and decreases the ability of traditional software to accurately track and finely control application progress, hindering the adoption of manycore processors in embedded computing. In this paper we propose Progress Time as the counterpart of CPU-time in space-multiplexed systems and show how it can be used to track application progress. We also introduce TimeCube, a manycore embedded processor that uses dynamic execution isolation and shadow performance modeling to provide an accurate online measurement of each application's Progress Time. Our evaluation shows that a 32-core TimeCube processor can track application progress with less than 1% error even in the presence of a 6× average worst-case slowdown. TimeCube also uses Progress Times to perform online architectural resource management that leads to a 36% improvement in throughput compared to existing microarchitectural resource allocation schemes. Overall, the results argue for adding the requisite microarchitectural structures to support Progress Time in manycore chips for embedded systems.

UR - http://www.scopus.com/inward/record.url?scp=84888881028&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84888881028&partnerID=8YFLogxK

U2 - 10.1109/SAMOS.2013.6621127

DO - 10.1109/SAMOS.2013.6621127

M3 - Conference contribution

SN - 9781479901036

T3 - Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013

SP - 227

EP - 236

BT - Proceedings - 2013 International Conference on Embedded Computer Systems

PB - IEEE Computer Society

ER -

Gupta A, Sampson JM, Taylor MB. TimeCube: A manycore embedded processor with interference-agnostic progress tracking. In Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013. IEEE Computer Society. 2013. p. 227-236. 6621127. (Proceedings - 2013 International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation, IC-SAMOS 2013). https://doi.org/10.1109/SAMOS.2013.6621127