Skip to main content
Top

2014 | OriginalPaper | Chapter

Extending Explicit Management of SMMHS to OpenMP for Heterogeneous Architectures

Authors : Xiaoxian Liu, Rongcai Zhao, Yuan Yao, Pinfeng Huang

Published in: Unifying Electrical Engineering and Electronics Engineering

Publisher: Springer New York

Activate our intelligent search to find suitable subject content or patents.

search-config
loading …

Abstract

In heterogeneous architectures with software-managed memory hierarchies system (SMMHS), software is deployed in charge of data placement of device memories and data movement between main memory and device ones. Considering the simplicity, portability, and performance characteristics of OpenMP, we argue that it is the most appropriate model for high level heterogeneous system programming. In this chapter, we address the problem of managing data storage and data transfer within SMMHS by proposing a potential extension to OpenMP. Meanwhile, we develop a method to compute the array region by applying the polyhedral model. Finally, we implement the automatic generation of the extended OpenMP clauses based on Open64, providing support for the migration of sequential programs to heterogeneous platforms. The experimental results show that OpenMP programs automatically generated with extended clauses get significant performance improvements on heterogeneous architectures. The extended OpenMP clauses could improve data placement of device memories and data movement between main memory and device ones evidently.

Dont have a licence yet? Then find out more about our products and how to get one now:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

  • über 102.000 Bücher
  • über 537 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Maschinenbau + Werkstoffe
  • Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

  • über 67.000 Bücher
  • über 390 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Maschinenbau + Werkstoffe




 

Jetzt Wissensvorsprung sichern!

Literature
1.
go back to reference NVIDIA corporation, NVIDIA CUDA Compute Unified Device Architecture Version 2.0 (2008) NVIDIA corporation, NVIDIA CUDA Compute Unified Device Architecture Version 2.0 (2008)
2.
go back to reference OpenMP Architecture Review Board, OpenMP Application Program Interface Version 3.1 (2011) OpenMP Architecture Review Board, OpenMP Application Program Interface Version 3.1 (2011)
3.
go back to reference High Performance Computer Tools Group, Overview of the Open64 Compiler Infrastructure (2002) High Performance Computer Tools Group, Overview of the Open64 Compiler Infrastructure (2002)
4.
go back to reference Bastoul C, Cohen A (2004) Putting polyhedral loop transformations to work. Languages and Compilers for Parallel Computing 2958:209–225 Bastoul C, Cohen A (2004) Putting polyhedral loop transformations to work. Languages and Compilers for Parallel Computing 2958:209–225
5.
go back to reference Zhao Y, Kennedy K (2007) Dependence-based code generation for a CELL processor. Languages and Compilers for Parallel Computing 4382:64–79 Zhao Y, Kennedy K (2007) Dependence-based code generation for a CELL processor. Languages and Compilers for Parallel Computing 4382:64–79
6.
go back to reference Chen T, Sura Z, O’Brien K et al (2007) Optimizing the use of static buffers for DMA on a CELL chip. Languages and Compilers for Parallel Computing 4382:314–329 Chen T, Sura Z, O’Brien K et al (2007) Optimizing the use of static buffers for DMA on a CELL chip. Languages and Compilers for Parallel Computing 4382:314–329
7.
go back to reference Baskaran MM, Bondhugula U, Krishnamoorthy S et al (2008) Automatic data movement and computation mapping for multi-level parallel architectures with explicitly managed memories. In: Proc. PPoPP ‘08. pp 1–10 Baskaran MM, Bondhugula U, Krishnamoorthy S et al (2008) Automatic data movement and computation mapping for multi-level parallel architectures with explicitly managed memories. In: Proc. PPoPP ‘08. pp 1–10
8.
go back to reference Bondhugula U, Hartono A, Ramanujam J et al (2008) A practical automatic polyhedral parallelizer and locality optimizer. In: Proc. PLDI ‘08. pp 101–113 Bondhugula U, Hartono A, Ramanujam J et al (2008) A practical automatic polyhedral parallelizer and locality optimizer. In: Proc. PLDI ‘08. pp 101–113
9.
go back to reference Jablin TB, Prabhu P, Jablin JA et al (2011) Automatic CPU-GPU communication management and optimization. In: Proc. PLDI ‘11. pp 142–151 Jablin TB, Prabhu P, Jablin JA et al (2011) Automatic CPU-GPU communication management and optimization. In: Proc. PLDI ‘11. pp 142–151
Metadata
Title
Extending Explicit Management of SMMHS to OpenMP for Heterogeneous Architectures
Authors
Xiaoxian Liu
Rongcai Zhao
Yuan Yao
Pinfeng Huang
Copyright Year
2014
Publisher
Springer New York
DOI
https://doi.org/10.1007/978-1-4614-4981-2_242