Advanced search
1 file | 415.82 KB

Strategies for dynamic memory allocation in hybrid architectures

Peter Bertels (UGent) , Wim Heirman (UGent) and Dirk Stroobandt (UGent)
Author
Organization
Abstract
Hybrid architectures combining the strengths of general-purpose processors with application-specific hardware accelerators can lead to a significant performance improvement. Our hybrid architecture uses a Java Virtual Machine as an abstraction layer to hide the complexity of the hardware/software interface between processor and accelerator from the programmer. The data communication between the accelerator and the processor often incurs a significant cost, which sometimes annihilates the original speedup obtained by the accelerator. This article shows how we minimise this communication cost by dynamically chosing an optimal data layout in the Java heap memory which is distributed over both the accelerator and the processor memory. The proposed self-learning memory allocation strategy finds the optimal location for each Java object's data by means of runtime profiling. The communication cost is effectively reduced by up to 86% for the benchmarks in the DaCapo suite (51% on average).
Keywords
java, hardware acceleration, memory management

Downloads

  • bertels2009sfdmaiha.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 415.82 KB

Citation

Please use this url to cite or link to this publication:

Chicago
Bertels, Peter, Wim Heirman, and Dirk Stroobandt. 2009. “Strategies for Dynamic Memory Allocation in Hybrid Architectures.” In CF’09: CONFERENCE ON COMPUTING FRONTIERS & WORKSHOPS, 217–220. New York, NY, USA: Association for Computing Machinery (ACM).
APA
Bertels, P., Heirman, W., & Stroobandt, D. (2009). Strategies for dynamic memory allocation in hybrid architectures. CF’09: CONFERENCE ON COMPUTING FRONTIERS & WORKSHOPS (pp. 217–220). Presented at the 6th ACM International Conference on Computing Frontiers and Workshops, New York, NY, USA: Association for Computing Machinery (ACM).
Vancouver
1.
Bertels P, Heirman W, Stroobandt D. Strategies for dynamic memory allocation in hybrid architectures. CF’09: CONFERENCE ON COMPUTING FRONTIERS & WORKSHOPS. New York, NY, USA: Association for Computing Machinery (ACM); 2009. p. 217–20.
MLA
Bertels, Peter, Wim Heirman, and Dirk Stroobandt. “Strategies for Dynamic Memory Allocation in Hybrid Architectures.” CF’09: CONFERENCE ON COMPUTING FRONTIERS & WORKSHOPS. New York, NY, USA: Association for Computing Machinery (ACM), 2009. 217–220. Print.
@inproceedings{819988,
  abstract     = {Hybrid architectures combining the strengths of general-purpose processors with application-specific hardware accelerators can lead to a significant performance improvement. Our hybrid architecture uses a Java Virtual Machine as an abstraction layer to hide the complexity of the hardware/software interface between processor and accelerator from the programmer. The data communication between the accelerator and the processor often incurs a significant cost, which sometimes annihilates the original speedup obtained by the accelerator. This article shows how we minimise this communication cost by dynamically chosing an optimal data layout in the Java heap memory which is distributed over both the accelerator and the processor memory. The proposed self-learning memory allocation strategy finds the optimal location for each Java object's data by means of runtime profiling. The communication cost is effectively reduced by up to 86\% for the benchmarks in the DaCapo suite (51\% on average).},
  author       = {Bertels, Peter and Heirman, Wim and Stroobandt, Dirk},
  booktitle    = {CF'09: CONFERENCE ON COMPUTING FRONTIERS \& WORKSHOPS},
  issn         = {978-1-60558-413-3},
  language     = {eng},
  location     = {Ischia, Italia},
  pages        = {217--220},
  publisher    = {Association for Computing Machinery (ACM)},
  title        = {Strategies for dynamic memory allocation in hybrid architectures},
  url          = {http://dx.doi.org/10.1145/1531743.1531778},
  year         = {2009},
}

Altmetric
View in Altmetric
Web of Science
Times cited: