Advanced search
1 file | 416.00 KB

How sensitive is processor customization to the workload's input datasets?

Author
Organization
Project
HPC-UGent: the central High Performance Computing infrastructure of Ghent University
Abstract
Hardware customization is an effective approach for meeting application performance requirements while achieving high levels of energy efficiency. Application-specific processors achieve high performance at low energy by tailoring their designs towards a specific workload, i.e., an application or application domain of interest. A fundamental question that has remained unanswered so far though is to what extent processor customization is sensitive to the training workload's input datasets. Current practice is to consider a single or only a few input datasets per workload during the processor design cycle - the reason being that simulation is prohibitively time-consuming which excludes considering a large number of datasets. This paper addresses this fundamental question, for the first time. In order to perform the large number of runs required to address this question in a reasonable amount of time, we first propose a mechanistic analytical model, built from first principles, that is accurate within 3.6% on average across a broad design space. The analytical model is at least 4 orders of magnitude faster than detailed cycle-accurate simulation for design space exploration. Using the model, we are able to study the sensitivity of a workload's input dataset on the optimum customized processor architecture. Considering MiBench benchmarks and 1000 datasets per benchmark, we conclude that processor customization is largely dataset-insensitive. This has an important implication in practice: a single or only a few datasets are sufficient for determining the optimum processor architecture when designing application-specific processors.
Keywords
application specific integrated circuits, computer architecture, energy conservation, instruction sets, logic design, low-power electronics, multiprocessing systems, power aware computing

Downloads

  • sasp11.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 416.00 KB

Citation

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

Chicago
Breughe, Maximilien, Li Zheng, Yang Chen, Stijn Eyerman, Olivier Temam, Chengyong Wu, and Lieven Eeckhout. 2011. “How Sensitive Is Processor Customization to the Workload’s Input Datasets?” In 2011 IEEE 9th International Symposium on Application Specific Processors. Los Alamitos, CA, USA: IEEE Computer Society.
APA
Breughe, M., Zheng, L., Chen, Y., Eyerman, S., Temam, O., Wu, C., & Eeckhout, L. (2011). How sensitive is processor customization to the workload’s input datasets? 2011 IEEE 9th international symposium on application specific processors. Presented at the 2011 IEEE 9th international symposium on Application Specific Processors (SASP 2011), Los Alamitos, CA, USA: IEEE Computer Society.
Vancouver
1.
Breughe M, Zheng L, Chen Y, Eyerman S, Temam O, Wu C, et al. How sensitive is processor customization to the workload’s input datasets? 2011 IEEE 9th international symposium on application specific processors. Los Alamitos, CA, USA: IEEE Computer Society; 2011.
MLA
Breughe, Maximilien, Li Zheng, Yang Chen, et al. “How Sensitive Is Processor Customization to the Workload’s Input Datasets?” 2011 IEEE 9th International Symposium on Application Specific Processors. Los Alamitos, CA, USA: IEEE Computer Society, 2011. Print.
@inproceedings{1977544,
  abstract     = {Hardware customization is an effective approach for meeting application performance requirements while achieving high levels of energy efficiency. Application-specific processors achieve high performance at low energy by tailoring their designs towards a specific workload, i.e., an application or application domain of interest. A fundamental question that has remained unanswered so far though is to what extent processor customization is sensitive to the training workload's input datasets. Current practice is to consider a single or only a few input datasets per workload during the processor design cycle - the reason being that simulation is prohibitively time-consuming which excludes considering a large number of datasets. This paper addresses this fundamental question, for the first time. In order to perform the large number of runs required to address this question in a reasonable amount of time, we first propose a mechanistic analytical model, built from first principles, that is accurate within 3.6\% on average across a broad design space. The analytical model is at least 4 orders of magnitude faster than detailed cycle-accurate simulation for design space exploration. Using the model, we are able to study the sensitivity of a workload's input dataset on the optimum customized processor architecture. Considering MiBench benchmarks and 1000 datasets per benchmark, we conclude that processor customization is largely dataset-insensitive. This has an important implication in practice: a single or only a few datasets are sufficient for determining the optimum processor architecture when designing application-specific processors.},
  author       = {Breughe, Maximilien and Zheng, Li and Chen, Yang and Eyerman, Stijn and Temam, Olivier and Wu, Chengyong and Eeckhout, Lieven},
  booktitle    = {2011 IEEE 9th international symposium on application specific processors},
  isbn         = {9781457712128},
  keyword      = {application specific integrated circuits,computer architecture,energy conservation,instruction sets,logic design,low-power electronics,multiprocessing systems,power aware computing},
  language     = {eng},
  location     = {San Diego, CA, USA},
  pages        = {7},
  publisher    = {IEEE Computer Society},
  title        = {How sensitive is processor customization to the workload's input datasets?},
  url          = {http://dx.doi.org/10.1109/SASP.2011.5941070},
  year         = {2011},
}

Altmetric
View in Altmetric