Using profile information to assist advanced compiler optimization and scheduling

W. Chen; R. Bringmann; S. Mahlke; S. Anik; T. Kiyohara; N. Warter; D. Lavery; W. M. Hwu; R. Hank; J. Gyllenhaal

doi:10.1007/3-540-57502-2_38

Using profile information to assist advanced compiler optimization and scheduling

W. Chen, R. Bringmann, S. Mahlke, S. Anik, T. Kiyohara, N. Warter, D. Lavery, W. M. Hwu, R. Hank, J. Gyllenhaal

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

Abstract

Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.

Original language	English (US)
Title of host publication	Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings
Editors	Utpal Banerjee, David Gelernter, Alex Nicolau, David Padua
Publisher	Springer-Verlag Berlin Heidelberg
Pages	31-48
Number of pages	18
ISBN (Print)	9783540575023
DOIs	https://doi.org/10.1007/3-540-57502-2_38
State	Published - 1993
Event	IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017 - Hamburg, Germany Duration: Sep 3 2017 → Sep 7 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	757 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017
Country	Germany
City	Hamburg
Period	9/3/17 → 9/7/17

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/3-540-57502-2_38

Fingerprint Dive into the research topics of 'Using profile information to assist advanced compiler optimization and scheduling'. Together they form a unique fingerprint.

Cite this

Chen, W., Bringmann, R., Mahlke, S., Anik, S., Kiyohara, T., Warter, N., Lavery, D., Hwu, W. M., Hank, R., & Gyllenhaal, J. (1993). Using profile information to assist advanced compiler optimization and scheduling. In U. Banerjee, D. Gelernter, A. Nicolau, & D. Padua (Eds.), Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings (pp. 31-48). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 757 LNCS). Springer-Verlag Berlin Heidelberg. https://doi.org/10.1007/3-540-57502-2_38

Using profile information to assist advanced compiler optimization and scheduling. / Chen, W.; Bringmann, R.; Mahlke, S.; Anik, S.; Kiyohara, T.; Warter, N.; Lavery, D.; Hwu, W. M.; Hank, R.; Gyllenhaal, J.

Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. ed. / Utpal Banerjee; David Gelernter; Alex Nicolau; David Padua. Springer-Verlag Berlin Heidelberg, 1993. p. 31-48 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 757 LNCS).

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

Chen, W, Bringmann, R, Mahlke, S, Anik, S, Kiyohara, T, Warter, N, Lavery, D, Hwu, WM, Hank, R & Gyllenhaal, J 1993, Using profile information to assist advanced compiler optimization and scheduling. in U Banerjee, D Gelernter, A Nicolau & D Padua (eds), Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 757 LNCS, Springer-Verlag Berlin Heidelberg, pp. 31-48, IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017, Hamburg, Germany, 9/3/17. https://doi.org/10.1007/3-540-57502-2_38

Chen W, Bringmann R, Mahlke S, Anik S, Kiyohara T, Warter N et al. Using profile information to assist advanced compiler optimization and scheduling. In Banerjee U, Gelernter D, Nicolau A, Padua D, editors, Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. Springer-Verlag Berlin Heidelberg. 1993. p. 31-48. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/3-540-57502-2_38

Chen, W. ; Bringmann, R. ; Mahlke, S. ; Anik, S. ; Kiyohara, T. ; Warter, N. ; Lavery, D. ; Hwu, W. M. ; Hank, R. ; Gyllenhaal, J. / Using profile information to assist advanced compiler optimization and scheduling. Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. editor / Utpal Banerjee ; David Gelernter ; Alex Nicolau ; David Padua. Springer-Verlag Berlin Heidelberg, 1993. pp. 31-48 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{57421a0d5f024b8d99bfed14015f284c,

title = "Using profile information to assist advanced compiler optimization and scheduling",

abstract = "Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.",

author = "W. Chen and R. Bringmann and S. Mahlke and S. Anik and T. Kiyohara and N. Warter and D. Lavery and Hwu, {W. M.} and R. Hank and J. Gyllenhaal",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 1993. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.; IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017 ; Conference date: 03-09-2017 Through 07-09-2017",

year = "1993",

doi = "10.1007/3-540-57502-2_38",

language = "English (US)",

isbn = "9783540575023",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer-Verlag Berlin Heidelberg",

pages = "31--48",

editor = "Utpal Banerjee and David Gelernter and Alex Nicolau and David Padua",

booktitle = "Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings",

}

TY - GEN

T1 - Using profile information to assist advanced compiler optimization and scheduling

AU - Chen, W.

AU - Bringmann, R.

AU - Mahlke, S.

AU - Anik, S.

AU - Kiyohara, T.

AU - Warter, N.

AU - Lavery, D.

AU - Hwu, W. M.

AU - Hank, R.

AU - Gyllenhaal, J.

PY - 1993

Y1 - 1993

N2 - Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.

AB - Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.

UR - http://www.scopus.com/inward/record.url?scp=70449836627&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70449836627&partnerID=8YFLogxK

U2 - 10.1007/3-540-57502-2_38

DO - 10.1007/3-540-57502-2_38

M3 - Conference contribution

AN - SCOPUS:70449836627

SN - 9783540575023

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 31

EP - 48

BT - Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings

A2 - Banerjee, Utpal

A2 - Gelernter, David

A2 - Nicolau, Alex

A2 - Padua, David

PB - Springer-Verlag Berlin Heidelberg

T2 - IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017

Y2 - 3 September 2017 through 7 September 2017

ER -