On using incremental profiling for the performance analysis of shared memory parallel applications

Karl Fuerlinger; Michael Gerndt; Jack Dongarra

doi:10.1007/978-3-540-74466-5_8

On using incremental profiling for the performance analysis of shared memory parallel applications

Karl Fuerlinger, Michael Gerndt, Jack Dongarra

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

Abstract

Profiling is often the method of choice for performance analysis of parallel applications due to its low overhead and easily comprehensible results. However, a disadvantage of profiling is the loss of temporal information that makes it impossible to causally relate performance phenomena to events that happened prior or later during execution. We investigate techniques to add temporal dimension to profiling data by incrementally capturing profiles during the runtime of the application and discuss the insights that can be gained from this type of performance data. The context in which we explore these ideas is an existing profiling tool for OpenMP applications. © Springer-Verlag Berlin Heidelberg 2007.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)\|Lect. Notes Comput. Sci.
Publisher	Springer Nature
Pages	62-71
Number of pages	9
Volume	4641
ISBN (Print)	9783540744658
DOIs	https://doi.org/10.1007/978-3-540-74466-5_8
Publication status	Published - 2007
Event	13th International Euro-Par Conference on Parallel Processing, Euro-Par 2007 - Rennes Duration: 1 Jul 2007 → … http://dblp.uni-trier.de/db/conf/europar/europar2007.html#Pjesivac-GrbovicBFAD07http://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07

Publication series

Name	Lecture Notes in Computer Science

Conference

Conference	13th International Euro-Par Conference on Parallel Processing, Euro-Par 2007
City	Rennes
Period	1/07/07 → …
Internet address	http://dblp.uni-trier.de/db/conf/europar/europar2007.html#Pjesivac-GrbovicBFAD07http://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07

Access to Document

10.1007/978-3-540-74466-5_8

Cite this

Fuerlinger, K., Gerndt, M., & Dongarra, J. (2007). On using incremental profiling for the performance analysis of shared memory parallel applications. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci. (Vol. 4641, pp. 62-71). (Lecture Notes in Computer Science). Springer Nature. https://doi.org/10.1007/978-3-540-74466-5_8

Fuerlinger, Karl ; Gerndt, Michael ; Dongarra, Jack. / On using incremental profiling for the performance analysis of shared memory parallel applications. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. Vol. 4641 Springer Nature, 2007. pp. 62-71 (Lecture Notes in Computer Science).

@inproceedings{ce098bf1f43946d3b2954f581eb0ac25,

title = "On using incremental profiling for the performance analysis of shared memory parallel applications",

abstract = "Profiling is often the method of choice for performance analysis of parallel applications due to its low overhead and easily comprehensible results. However, a disadvantage of profiling is the loss of temporal information that makes it impossible to causally relate performance phenomena to events that happened prior or later during execution. We investigate techniques to add temporal dimension to profiling data by incrementally capturing profiles during the runtime of the application and discuss the insights that can be gained from this type of performance data. The context in which we explore these ideas is an existing profiling tool for OpenMP applications. {\textcopyright} Springer-Verlag Berlin Heidelberg 2007.",

author = "Karl Fuerlinger and Michael Gerndt and Jack Dongarra",

year = "2007",

doi = "10.1007/978-3-540-74466-5_8",

language = "English",

isbn = "9783540744658",

volume = "4641",

series = "Lecture Notes in Computer Science",

publisher = "Springer Nature",

pages = "62--71",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.",

address = "United States",

note = "13th International Euro-Par Conference on Parallel Processing, Euro-Par 2007 ; Conference date: 01-07-2007",

url = "http://dblp.uni-trier.de/db/conf/europar/europar2007.html#Pjesivac-GrbovicBFAD07http://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/europar/Pjesivac-GrbovicBFAD07",

}

Fuerlinger, K, Gerndt, M & Dongarra, J 2007, On using incremental profiling for the performance analysis of shared memory parallel applications. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. vol. 4641, Lecture Notes in Computer Science, Springer Nature, pp. 62-71, 13th International Euro-Par Conference on Parallel Processing, Euro-Par 2007, Rennes, 1/07/07. https://doi.org/10.1007/978-3-540-74466-5_8

On using incremental profiling for the performance analysis of shared memory parallel applications. / Fuerlinger, Karl; Gerndt, Michael; Dongarra, Jack.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. Vol. 4641 Springer Nature, 2007. p. 62-71 (Lecture Notes in Computer Science).

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

TY - GEN

T1 - On using incremental profiling for the performance analysis of shared memory parallel applications

AU - Fuerlinger, Karl

AU - Gerndt, Michael

AU - Dongarra, Jack

PY - 2007

Y1 - 2007

N2 - Profiling is often the method of choice for performance analysis of parallel applications due to its low overhead and easily comprehensible results. However, a disadvantage of profiling is the loss of temporal information that makes it impossible to causally relate performance phenomena to events that happened prior or later during execution. We investigate techniques to add temporal dimension to profiling data by incrementally capturing profiles during the runtime of the application and discuss the insights that can be gained from this type of performance data. The context in which we explore these ideas is an existing profiling tool for OpenMP applications. © Springer-Verlag Berlin Heidelberg 2007.

AB - Profiling is often the method of choice for performance analysis of parallel applications due to its low overhead and easily comprehensible results. However, a disadvantage of profiling is the loss of temporal information that makes it impossible to causally relate performance phenomena to events that happened prior or later during execution. We investigate techniques to add temporal dimension to profiling data by incrementally capturing profiles during the runtime of the application and discuss the insights that can be gained from this type of performance data. The context in which we explore these ideas is an existing profiling tool for OpenMP applications. © Springer-Verlag Berlin Heidelberg 2007.

U2 - 10.1007/978-3-540-74466-5_8

DO - 10.1007/978-3-540-74466-5_8

M3 - Conference contribution

SN - 9783540744658

VL - 4641

T3 - Lecture Notes in Computer Science

SP - 62

EP - 71

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.

PB - Springer Nature

T2 - 13th International Euro-Par Conference on Parallel Processing, Euro-Par 2007

Y2 - 1 July 2007

ER -

Fuerlinger K, Gerndt M, Dongarra J. On using incremental profiling for the performance analysis of shared memory parallel applications. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. Vol. 4641. Springer Nature. 2007. p. 62-71. (Lecture Notes in Computer Science). doi: 10.1007/978-3-540-74466-5_8

On using incremental profiling for the performance analysis of shared memory parallel applications

Abstract

Publication series

Conference

Access to Document

Fingerprint

Cite this