Real-Time and Energy-Efficient Face Detection on CPU-GPU Heterogeneous Embedded Platforms

Chanyoung OH; Saehanseul YI; Youngmin YI

doi:10.1587/transinf.2018PAP0004

Real-Time and Energy-Efficient Face Detection on CPU-GPU Heterogeneous Embedded Platforms

Chanyoung OH, Saehanseul YI, Youngmin YI

Full Text Views

0

Cite this

Summary :

As energy efficiency has become a major design constraint or objective, heterogeneous manycore architectures have emerged as mainstream target platforms not only in server systems but also in embedded systems. Manycore accelerators such as GPUs are getting also popular in embedded domains, as well as the heterogeneous CPU cores. However, as the number of cores in an embedded GPU is far less than that of a server GPU, it is important to utilize both heterogeneous multi-core CPUs and GPUs to achieve the desired throughput with the minimal energy consumption. In this paper, we present a case study of mapping LBP-based face detection onto a recent CPU-GPU heterogeneous embedded platform, which exploits both task parallelism and data parallelism to achieve maximal energy efficiency with a real-time constraint. We first present the parallelization technique of each task for the GPU execution, then we propose performance and energy models for both task-parallel and data-parallel executions on heterogeneous processors, which are used in design space exploration for the optimal mapping. The design space is huge since not only processor heterogeneity such as CPU-GPU and big.LITTLE, but also various data partitioning ratios for the data-parallel execution on these heterogeneous processors are considered. In our case study of LBP face detection on Exynos 5422, the estimation error of the proposed performance and energy models were on average -2.19% and -3.67% respectively. By systematically finding the optimal mappings with the proposed models, we could achieve 28.6% less energy consumption compared to the manual mapping, while still meeting the real-time constraint.

Publication: IEICE TRANSACTIONS on Information Vol.E101-D No.12 pp.2878-2888

Publication Date: 2018/12/01

Publicized: 2018/09/18

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2018PAP0004

Type of Manuscript: Special Section PAPER (Special Section on Parallel and Distributed Computing and Networking)

Category: Real-time Systems

Authors

Chanyoung OH
  University of Seoul
Saehanseul YI
  University of Seoul
Youngmin YI
  University of Seoul

Keyword

CPU-GPU heterogeneous execution, performance and energy estimation, task mapping, face detection

Cite this

Copy

Chanyoung OH, Saehanseul YI, Youngmin YI, "Real-Time and Energy-Efficient Face Detection on CPU-GPU Heterogeneous Embedded Platforms" in IEICE TRANSACTIONS on Information, vol. E101-D, no. 12, pp. 2878-2888, December 2018, doi: 10.1587/transinf.2018PAP0004.
Abstract: As energy efficiency has become a major design constraint or objective, heterogeneous manycore architectures have emerged as mainstream target platforms not only in server systems but also in embedded systems. Manycore accelerators such as GPUs are getting also popular in embedded domains, as well as the heterogeneous CPU cores. However, as the number of cores in an embedded GPU is far less than that of a server GPU, it is important to utilize both heterogeneous multi-core CPUs and GPUs to achieve the desired throughput with the minimal energy consumption. In this paper, we present a case study of mapping LBP-based face detection onto a recent CPU-GPU heterogeneous embedded platform, which exploits both task parallelism and data parallelism to achieve maximal energy efficiency with a real-time constraint. We first present the parallelization technique of each task for the GPU execution, then we propose performance and energy models for both task-parallel and data-parallel executions on heterogeneous processors, which are used in design space exploration for the optimal mapping. The design space is huge since not only processor heterogeneity such as CPU-GPU and big.LITTLE, but also various data partitioning ratios for the data-parallel execution on these heterogeneous processors are considered. In our case study of LBP face detection on Exynos 5422, the estimation error of the proposed performance and energy models were on average -2.19% and -3.67% respectively. By systematically finding the optimal mappings with the proposed models, we could achieve 28.6% less energy consumption compared to the manual mapping, while still meeting the real-time constraint.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018PAP0004/_p

Copy

@ARTICLE{e101-d_12_2878,
author={Chanyoung OH, Saehanseul YI, Youngmin YI, },
journal={IEICE TRANSACTIONS on Information},
title={Real-Time and Energy-Efficient Face Detection on CPU-GPU Heterogeneous Embedded Platforms},
year={2018},
volume={E101-D},
number={12},
pages={2878-2888},
abstract={As energy efficiency has become a major design constraint or objective, heterogeneous manycore architectures have emerged as mainstream target platforms not only in server systems but also in embedded systems. Manycore accelerators such as GPUs are getting also popular in embedded domains, as well as the heterogeneous CPU cores. However, as the number of cores in an embedded GPU is far less than that of a server GPU, it is important to utilize both heterogeneous multi-core CPUs and GPUs to achieve the desired throughput with the minimal energy consumption. In this paper, we present a case study of mapping LBP-based face detection onto a recent CPU-GPU heterogeneous embedded platform, which exploits both task parallelism and data parallelism to achieve maximal energy efficiency with a real-time constraint. We first present the parallelization technique of each task for the GPU execution, then we propose performance and energy models for both task-parallel and data-parallel executions on heterogeneous processors, which are used in design space exploration for the optimal mapping. The design space is huge since not only processor heterogeneity such as CPU-GPU and big.LITTLE, but also various data partitioning ratios for the data-parallel execution on these heterogeneous processors are considered. In our case study of LBP face detection on Exynos 5422, the estimation error of the proposed performance and energy models were on average -2.19% and -3.67% respectively. By systematically finding the optimal mappings with the proposed models, we could achieve 28.6% less energy consumption compared to the manual mapping, while still meeting the real-time constraint.},
keywords={},
doi={10.1587/transinf.2018PAP0004},
ISSN={1745-1361},
month={December},}

Copy

TY - JOUR
TI - Real-Time and Energy-Efficient Face Detection on CPU-GPU Heterogeneous Embedded Platforms
T2 - IEICE TRANSACTIONS on Information
SP - 2878
EP - 2888
AU - Chanyoung OH
AU - Saehanseul YI
AU - Youngmin YI
PY - 2018
DO - 10.1587/transinf.2018PAP0004
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2018
AB - As energy efficiency has become a major design constraint or objective, heterogeneous manycore architectures have emerged as mainstream target platforms not only in server systems but also in embedded systems. Manycore accelerators such as GPUs are getting also popular in embedded domains, as well as the heterogeneous CPU cores. However, as the number of cores in an embedded GPU is far less than that of a server GPU, it is important to utilize both heterogeneous multi-core CPUs and GPUs to achieve the desired throughput with the minimal energy consumption. In this paper, we present a case study of mapping LBP-based face detection onto a recent CPU-GPU heterogeneous embedded platform, which exploits both task parallelism and data parallelism to achieve maximal energy efficiency with a real-time constraint. We first present the parallelization technique of each task for the GPU execution, then we propose performance and energy models for both task-parallel and data-parallel executions on heterogeneous processors, which are used in design space exploration for the optimal mapping. The design space is huge since not only processor heterogeneity such as CPU-GPU and big.LITTLE, but also various data partitioning ratios for the data-parallel execution on these heterogeneous processors are considered. In our case study of LBP face detection on Exynos 5422, the estimation error of the proposed performance and energy models were on average -2.19% and -3.67% respectively. By systematically finding the optimal mappings with the proposed models, we could achieve 28.6% less energy consumption compared to the manual mapping, while still meeting the real-time constraint.
ER -