Memory-aware application mapping on coarse-grained reconfigurable arrays

Yongjoo Kim, Jongeun Lee, Aviral Shrivastava, Jonghee Yoon, Yunheung Paek

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

13 Citations (Scopus)

Abstract

Coarse-Grained Reconfigurable Arrays (CGRAs) are a very promising platform, providing both, up to 10-100 MOps/mW of power efficiency and are software programmable. However, this cardinal promise of CGRAs critically hinges on the effectiveness of application mapping onto CGRA platforms. While previous solutions have greatly improved the computation speed, they have largely ignored the impact of the local memory architecture on the achievable power and performance. This paper motivates the need for memory-aware application mapping for CGRAs, and proposes an effective solution for application mapping that considers the effects of various memory architecture parameters including the number of banks, local memory size, and the communication bandwidth between the local memory and the external main memory. Our proposed solution achieves 62% reduction in the energy-delay product, which factors into about 47% and 28% reduction in the energy consumption and runtime, respectively, as compared to memory-unaware mapping for realistic local memory architectures. We also show that our scheme scales across a range of applications, and memory parameters.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages171-185
Number of pages15
Volume5952 LNCS
DOIs
StatePublished - 2010
Event5th International Conference on High Performance Embedded Architectures and Compilers, HiPEAC 2010 - Pisa, Italy
Duration: Jan 25 2010Jan 27 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5952 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other5th International Conference on High Performance Embedded Architectures and Compilers, HiPEAC 2010
CountryItaly
CityPisa
Period1/25/101/27/10

Fingerprint

Data storage equipment
Memory architecture
Hinges
Surjection
Energy utilization
Bandwidth
Energy Consumption
Communication
Software
Energy
Range of data
Architecture

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kim, Y., Lee, J., Shrivastava, A., Yoon, J., & Paek, Y. (2010). Memory-aware application mapping on coarse-grained reconfigurable arrays. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 5952 LNCS, pp. 171-185). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5952 LNCS). https://doi.org/10.1007/978-3-642-11515-8_14

Memory-aware application mapping on coarse-grained reconfigurable arrays. / Kim, Yongjoo; Lee, Jongeun; Shrivastava, Aviral; Yoon, Jonghee; Paek, Yunheung.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5952 LNCS 2010. p. 171-185 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5952 LNCS).

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

Kim, Y, Lee, J, Shrivastava, A, Yoon, J & Paek, Y 2010, Memory-aware application mapping on coarse-grained reconfigurable arrays. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 5952 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5952 LNCS, pp. 171-185, 5th International Conference on High Performance Embedded Architectures and Compilers, HiPEAC 2010, Pisa, Italy, 1/25/10. https://doi.org/10.1007/978-3-642-11515-8_14
Kim Y, Lee J, Shrivastava A, Yoon J, Paek Y. Memory-aware application mapping on coarse-grained reconfigurable arrays. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5952 LNCS. 2010. p. 171-185. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-11515-8_14
Kim, Yongjoo ; Lee, Jongeun ; Shrivastava, Aviral ; Yoon, Jonghee ; Paek, Yunheung. / Memory-aware application mapping on coarse-grained reconfigurable arrays. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5952 LNCS 2010. pp. 171-185 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f0451b9e591247b5ad6cb9ccfd748269,
title = "Memory-aware application mapping on coarse-grained reconfigurable arrays",
abstract = "Coarse-Grained Reconfigurable Arrays (CGRAs) are a very promising platform, providing both, up to 10-100 MOps/mW of power efficiency and are software programmable. However, this cardinal promise of CGRAs critically hinges on the effectiveness of application mapping onto CGRA platforms. While previous solutions have greatly improved the computation speed, they have largely ignored the impact of the local memory architecture on the achievable power and performance. This paper motivates the need for memory-aware application mapping for CGRAs, and proposes an effective solution for application mapping that considers the effects of various memory architecture parameters including the number of banks, local memory size, and the communication bandwidth between the local memory and the external main memory. Our proposed solution achieves 62{\%} reduction in the energy-delay product, which factors into about 47{\%} and 28{\%} reduction in the energy consumption and runtime, respectively, as compared to memory-unaware mapping for realistic local memory architectures. We also show that our scheme scales across a range of applications, and memory parameters.",
author = "Yongjoo Kim and Jongeun Lee and Aviral Shrivastava and Jonghee Yoon and Yunheung Paek",
year = "2010",
doi = "10.1007/978-3-642-11515-8_14",
language = "English (US)",
isbn = "3642115144",
volume = "5952 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "171--185",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Memory-aware application mapping on coarse-grained reconfigurable arrays

AU - Kim, Yongjoo

AU - Lee, Jongeun

AU - Shrivastava, Aviral

AU - Yoon, Jonghee

AU - Paek, Yunheung

PY - 2010

Y1 - 2010

N2 - Coarse-Grained Reconfigurable Arrays (CGRAs) are a very promising platform, providing both, up to 10-100 MOps/mW of power efficiency and are software programmable. However, this cardinal promise of CGRAs critically hinges on the effectiveness of application mapping onto CGRA platforms. While previous solutions have greatly improved the computation speed, they have largely ignored the impact of the local memory architecture on the achievable power and performance. This paper motivates the need for memory-aware application mapping for CGRAs, and proposes an effective solution for application mapping that considers the effects of various memory architecture parameters including the number of banks, local memory size, and the communication bandwidth between the local memory and the external main memory. Our proposed solution achieves 62% reduction in the energy-delay product, which factors into about 47% and 28% reduction in the energy consumption and runtime, respectively, as compared to memory-unaware mapping for realistic local memory architectures. We also show that our scheme scales across a range of applications, and memory parameters.

AB - Coarse-Grained Reconfigurable Arrays (CGRAs) are a very promising platform, providing both, up to 10-100 MOps/mW of power efficiency and are software programmable. However, this cardinal promise of CGRAs critically hinges on the effectiveness of application mapping onto CGRA platforms. While previous solutions have greatly improved the computation speed, they have largely ignored the impact of the local memory architecture on the achievable power and performance. This paper motivates the need for memory-aware application mapping for CGRAs, and proposes an effective solution for application mapping that considers the effects of various memory architecture parameters including the number of banks, local memory size, and the communication bandwidth between the local memory and the external main memory. Our proposed solution achieves 62% reduction in the energy-delay product, which factors into about 47% and 28% reduction in the energy consumption and runtime, respectively, as compared to memory-unaware mapping for realistic local memory architectures. We also show that our scheme scales across a range of applications, and memory parameters.

UR - http://www.scopus.com/inward/record.url?scp=77949614384&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77949614384&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-11515-8_14

DO - 10.1007/978-3-642-11515-8_14

M3 - Conference contribution

AN - SCOPUS:77949614384

SN - 3642115144

SN - 9783642115141

VL - 5952 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 171

EP - 185

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -