Pointers inside lambda closure objects in OpenMP target offload regions

David Truby; Carlo Bertolli; Steven A. Wright; Gheorghe Teodor Bercea; Kevin O'Brien; Stephen A. Jarvis

doi:10.1109/LLVM-HPC.2018.8639410

Pointers inside lambda closure objects in OpenMP target offload regions

David Truby, Carlo Bertolli, Steven A. Wright, Gheorghe Teodor Bercea, Kevin O'Brien, Stephen A. Jarvis

Engineering and Physical Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

With the diversification of HPC architectures beyond traditional CPU-based clusters, a number of new frameworks for performance portability across architectures have arisen. One way of implementing such frameworks is to use C++ templates and lambda expressions to design loop-like functions. However, lower level programming APIs that these implementations must use are often designed with C in mind and do not specify how they interact with C++ features such as lambda expressions. This paper discusses a change to the behavior of the OpenMP specification with respect to lambda expressions such that when functions generated by lambda expressions are called inside GPU regions, any pointers used in the lambda expression correctly refer to device pointers. This change has been implemented in a branch of the Clang C++ compiler and demonstrated with two representative codes. This change has also been accepted into the draft OpenMP^®specification for inclusion in OpenMP 5. Our results show that the implicit mapping of lambda expressions always exhibits identical performance to an explicit mapping but without breaking the abstraction provided by the high level frameworks.

Original language	English
Title of host publication	Proceedings of LLVM-HPC 2018
Subtitle of host publication	5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	10-17
Number of pages	8
ISBN (Electronic)	9781728101880
DOIs	https://doi.org/10.1109/LLVM-HPC.2018.8639410
Publication status	Published - 11 Feb 2019
Event	5th IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC, LLVM-HPC 2018 - Dallas, United States Duration: 12 Nov 2018 → …

Publication series

Name	Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis

Conference

Conference	5th IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC, LLVM-HPC 2018
Country/Territory	United States
City	Dallas
Period	12/11/18 → …

Bibliographical note

Funding Information:
This work was completed as part of IBM’s 2017 Summer Research Internship Program. This work was also partially supported by the Centre for Computational Plasma Physics, a collaboration between the University of Warwick and the UK Atomic Weapons Establishment. Prof. Stephen Jarvis is an AWE William Penney Fellow.

Publisher Copyright:
© 2018 IEEE.

ASJC Scopus subject areas

Hardware and Architecture
Software

Access to Document

10.1109/LLVM-HPC.2018.8639410

Cite this

Truby, D., Bertolli, C., Wright, S. A., Bercea, G. T., O'Brien, K., & Jarvis, S. A. (2019). Pointers inside lambda closure objects in OpenMP target offload regions. In Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis (pp. 10-17). Article 8639410 (Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/LLVM-HPC.2018.8639410

Truby, David ; Bertolli, Carlo ; Wright, Steven A. et al. / Pointers inside lambda closure objects in OpenMP target offload regions. Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE), 2019. pp. 10-17 (Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis).

@inproceedings{50f6794f724241b0a9e241ae88b35a23,

title = "Pointers inside lambda closure objects in OpenMP target offload regions",

abstract = "With the diversification of HPC architectures beyond traditional CPU-based clusters, a number of new frameworks for performance portability across architectures have arisen. One way of implementing such frameworks is to use C++ templates and lambda expressions to design loop-like functions. However, lower level programming APIs that these implementations must use are often designed with C in mind and do not specify how they interact with C++ features such as lambda expressions. This paper discusses a change to the behavior of the OpenMP specification with respect to lambda expressions such that when functions generated by lambda expressions are called inside GPU regions, any pointers used in the lambda expression correctly refer to device pointers. This change has been implemented in a branch of the Clang C++ compiler and demonstrated with two representative codes. This change has also been accepted into the draft OpenMP{\textregistered}specification for inclusion in OpenMP 5. Our results show that the implicit mapping of lambda expressions always exhibits identical performance to an explicit mapping but without breaking the abstraction provided by the high level frameworks.",

author = "David Truby and Carlo Bertolli and Wright, {Steven A.} and Bercea, {Gheorghe Teodor} and Kevin O'Brien and Jarvis, {Stephen A.}",

note = "Funding Information: This work was completed as part of IBM{\textquoteright}s 2017 Summer Research Internship Program. This work was also partially supported by the Centre for Computational Plasma Physics, a collaboration between the University of Warwick and the UK Atomic Weapons Establishment. Prof. Stephen Jarvis is an AWE William Penney Fellow. Publisher Copyright: {\textcopyright} 2018 IEEE.; 5th IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC, LLVM-HPC 2018 ; Conference date: 12-11-2018",

year = "2019",

month = feb,

day = "11",

doi = "10.1109/LLVM-HPC.2018.8639410",

language = "English",

series = "Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

pages = "10--17",

booktitle = "Proceedings of LLVM-HPC 2018",

}

Truby, D, Bertolli, C, Wright, SA, Bercea, GT, O'Brien, K & Jarvis, SA 2019, Pointers inside lambda closure objects in OpenMP target offload regions. in Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis., 8639410, Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis, Institute of Electrical and Electronics Engineers (IEEE), pp. 10-17, 5th IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC, LLVM-HPC 2018, Dallas, United States, 12/11/18. https://doi.org/10.1109/LLVM-HPC.2018.8639410

Pointers inside lambda closure objects in OpenMP target offload regions. / Truby, David; Bertolli, Carlo; Wright, Steven A. et al.
Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE), 2019. p. 10-17 8639410 (Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Pointers inside lambda closure objects in OpenMP target offload regions

AU - Truby, David

AU - Bertolli, Carlo

AU - Wright, Steven A.

AU - Bercea, Gheorghe Teodor

AU - O'Brien, Kevin

AU - Jarvis, Stephen A.

N1 - Funding Information: This work was completed as part of IBM’s 2017 Summer Research Internship Program. This work was also partially supported by the Centre for Computational Plasma Physics, a collaboration between the University of Warwick and the UK Atomic Weapons Establishment. Prof. Stephen Jarvis is an AWE William Penney Fellow. Publisher Copyright: © 2018 IEEE.

PY - 2019/2/11

Y1 - 2019/2/11

N2 - With the diversification of HPC architectures beyond traditional CPU-based clusters, a number of new frameworks for performance portability across architectures have arisen. One way of implementing such frameworks is to use C++ templates and lambda expressions to design loop-like functions. However, lower level programming APIs that these implementations must use are often designed with C in mind and do not specify how they interact with C++ features such as lambda expressions. This paper discusses a change to the behavior of the OpenMP specification with respect to lambda expressions such that when functions generated by lambda expressions are called inside GPU regions, any pointers used in the lambda expression correctly refer to device pointers. This change has been implemented in a branch of the Clang C++ compiler and demonstrated with two representative codes. This change has also been accepted into the draft OpenMP®specification for inclusion in OpenMP 5. Our results show that the implicit mapping of lambda expressions always exhibits identical performance to an explicit mapping but without breaking the abstraction provided by the high level frameworks.

AB - With the diversification of HPC architectures beyond traditional CPU-based clusters, a number of new frameworks for performance portability across architectures have arisen. One way of implementing such frameworks is to use C++ templates and lambda expressions to design loop-like functions. However, lower level programming APIs that these implementations must use are often designed with C in mind and do not specify how they interact with C++ features such as lambda expressions. This paper discusses a change to the behavior of the OpenMP specification with respect to lambda expressions such that when functions generated by lambda expressions are called inside GPU regions, any pointers used in the lambda expression correctly refer to device pointers. This change has been implemented in a branch of the Clang C++ compiler and demonstrated with two representative codes. This change has also been accepted into the draft OpenMP®specification for inclusion in OpenMP 5. Our results show that the implicit mapping of lambda expressions always exhibits identical performance to an explicit mapping but without breaking the abstraction provided by the high level frameworks.

UR - http://www.scopus.com/inward/record.url?scp=85063103386&partnerID=8YFLogxK

U2 - 10.1109/LLVM-HPC.2018.8639410

DO - 10.1109/LLVM-HPC.2018.8639410

M3 - Conference contribution

AN - SCOPUS:85063103386

T3 - Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis

SP - 10

EP - 17

BT - Proceedings of LLVM-HPC 2018

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - 5th IEEE/ACM Workshop on the LLVM Compiler Infrastructure in HPC, LLVM-HPC 2018

Y2 - 12 November 2018

ER -

Truby D, Bertolli C, Wright SA, Bercea GT, O'Brien K, Jarvis SA. Pointers inside lambda closure objects in OpenMP target offload regions. In Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers (IEEE). 2019. p. 10-17. 8639410. (Proceedings of LLVM-HPC 2018: 5th Workshop on the LLVM Compiler Infrastructure in HPC, Held in conjunction with SC 2018: The International Conference for High Performance Computing, Networking, Storage and Analysis). doi: 10.1109/LLVM-HPC.2018.8639410

Pointers inside lambda closure objects in OpenMP target offload regions

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this