Halide Code Generation Framework in Phylanx

R. Tohid; Shahrzad Shirzad; Christopher Taylor; Sayef Azad Sakin; Katherine E. Isaacs; Hartmut Kaiser

doi:10.1007/978-3-031-31209-0_3

Halide Code Generation Framework in Phylanx

R. Tohid, Shahrzad Shirzad, Christopher Taylor, Sayef Azad Sakin, Katherine E. Isaacs, Hartmut Kaiser

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

Abstract

Separating algorithms from their computation schedule has become a de facto solution to tackle the challenges of developing high performance code on modern heterogeneous architectures. Common approaches include Domain-specific languages (DSLs) which provide familiar APIs to domain experts, code generation frameworks that automate the generation of fast and portable code, and runtime systems that manage threads for concurrency and parallelism. In this paper, we present the Halide code generation framework for Phylanx distributed array processing platform. This extension enables compile-time optimization of Phylanx primitives for target architectures. To accomplish this, (1) we implemented new Phylanx primitives using Halide, and (2) partially exported Halide’s thread pool API to carry out parallelism on HPX (Phylanx’s runtime) threads. (3) showcased HPX performance analysis tools made available to Halide applications. The evaluation of the work has been done in two steps. First, we compare the performance of Halide applications running on its native runtime with that of the new HPX backend to verify there is no cost associated with using HPX threads. Next, we compare performances of a number of original implementations of Phylanx primitives against the new ones in Halide to verify performance and portability benefits of Halide in the context of Phylanx.

Original language	English (US)
Title of host publication	Euro-Par 2022
Subtitle of host publication	Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers
Editors	Jeremy Singer, Yehia Elkhatib, Dora Blanco Heras, Patrick Diehl, Nick Brown, Aleksandar Ilic
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	32-45
Number of pages	14
ISBN (Print)	9783031312083
DOIs	https://doi.org/10.1007/978-3-031-31209-0_3
State	Published - 2023
Externally published	Yes
Event	28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022 - Glasgow, United Kingdom Duration: Aug 22 2022 → Aug 26 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13835 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022
Country/Territory	United Kingdom
City	Glasgow
Period	8/22/22 → 8/26/22

Keywords

AMT
DSL
HPX
Halide
Phylanx

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-031-31209-0_3

Cite this

Tohid, R., Shirzad, S., Taylor, C., Sakin, S. A., Isaacs, K. E., & Kaiser, H. (2023). Halide Code Generation Framework in Phylanx. In J. Singer, Y. Elkhatib, D. Blanco Heras, P. Diehl, N. Brown, & A. Ilic (Eds.), Euro-Par 2022: Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers (pp. 32-45). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13835 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-31209-0_3

Halide Code Generation Framework in Phylanx. / Tohid, R.; Shirzad, Shahrzad; Taylor, Christopher et al.
Euro-Par 2022: Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers. ed. / Jeremy Singer; Yehia Elkhatib; Dora Blanco Heras; Patrick Diehl; Nick Brown; Aleksandar Ilic. Springer Science and Business Media Deutschland GmbH, 2023. p. 32-45 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13835 LNCS).

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

Tohid, R, Shirzad, S, Taylor, C, Sakin, SA, Isaacs, KE & Kaiser, H 2023, Halide Code Generation Framework in Phylanx. in J Singer, Y Elkhatib, D Blanco Heras, P Diehl, N Brown & A Ilic (eds), Euro-Par 2022: Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13835 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 32-45, 28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022, Glasgow, United Kingdom, 8/22/22. https://doi.org/10.1007/978-3-031-31209-0_3

Tohid R, Shirzad S, Taylor C, Sakin SA, Isaacs KE, Kaiser H. Halide Code Generation Framework in Phylanx. In Singer J, Elkhatib Y, Blanco Heras D, Diehl P, Brown N, Ilic A, editors, Euro-Par 2022: Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2023. p. 32-45. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-31209-0_3

Tohid, R. ; Shirzad, Shahrzad ; Taylor, Christopher et al. / Halide Code Generation Framework in Phylanx. Euro-Par 2022: Parallel Processing Workshops - Euro-Par 2022 International Workshops, Revised Selected Papers. editor / Jeremy Singer ; Yehia Elkhatib ; Dora Blanco Heras ; Patrick Diehl ; Nick Brown ; Aleksandar Ilic. Springer Science and Business Media Deutschland GmbH, 2023. pp. 32-45 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{9f7490e7e07b4901bc4db3bb0d344e1a,

title = "Halide Code Generation Framework in Phylanx",

abstract = "Separating algorithms from their computation schedule has become a de facto solution to tackle the challenges of developing high performance code on modern heterogeneous architectures. Common approaches include Domain-specific languages (DSLs) which provide familiar APIs to domain experts, code generation frameworks that automate the generation of fast and portable code, and runtime systems that manage threads for concurrency and parallelism. In this paper, we present the Halide code generation framework for Phylanx distributed array processing platform. This extension enables compile-time optimization of Phylanx primitives for target architectures. To accomplish this, (1) we implemented new Phylanx primitives using Halide, and (2) partially exported Halide{\textquoteright}s thread pool API to carry out parallelism on HPX (Phylanx{\textquoteright}s runtime) threads. (3) showcased HPX performance analysis tools made available to Halide applications. The evaluation of the work has been done in two steps. First, we compare the performance of Halide applications running on its native runtime with that of the new HPX backend to verify there is no cost associated with using HPX threads. Next, we compare performances of a number of original implementations of Phylanx primitives against the new ones in Halide to verify performance and portability benefits of Halide in the context of Phylanx.",

keywords = "AMT, DSL, HPX, Halide, Phylanx",

author = "R. Tohid and Shahrzad Shirzad and Christopher Taylor and Sakin, {Sayef Azad} and Isaacs, {Katherine E.} and Hartmut Kaiser",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022 ; Conference date: 22-08-2022 Through 26-08-2022",

year = "2023",

doi = "10.1007/978-3-031-31209-0_3",

language = "English (US)",

isbn = "9783031312083",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "32--45",

editor = "Jeremy Singer and Yehia Elkhatib and {Blanco Heras}, Dora and Patrick Diehl and Nick Brown and Aleksandar Ilic",

booktitle = "Euro-Par 2022",

address = "Germany",

}

TY - GEN

T1 - Halide Code Generation Framework in Phylanx

AU - Tohid, R.

AU - Shirzad, Shahrzad

AU - Taylor, Christopher

AU - Sakin, Sayef Azad

AU - Isaacs, Katherine E.

AU - Kaiser, Hartmut

PY - 2023

Y1 - 2023

N2 - Separating algorithms from their computation schedule has become a de facto solution to tackle the challenges of developing high performance code on modern heterogeneous architectures. Common approaches include Domain-specific languages (DSLs) which provide familiar APIs to domain experts, code generation frameworks that automate the generation of fast and portable code, and runtime systems that manage threads for concurrency and parallelism. In this paper, we present the Halide code generation framework for Phylanx distributed array processing platform. This extension enables compile-time optimization of Phylanx primitives for target architectures. To accomplish this, (1) we implemented new Phylanx primitives using Halide, and (2) partially exported Halide’s thread pool API to carry out parallelism on HPX (Phylanx’s runtime) threads. (3) showcased HPX performance analysis tools made available to Halide applications. The evaluation of the work has been done in two steps. First, we compare the performance of Halide applications running on its native runtime with that of the new HPX backend to verify there is no cost associated with using HPX threads. Next, we compare performances of a number of original implementations of Phylanx primitives against the new ones in Halide to verify performance and portability benefits of Halide in the context of Phylanx.

AB - Separating algorithms from their computation schedule has become a de facto solution to tackle the challenges of developing high performance code on modern heterogeneous architectures. Common approaches include Domain-specific languages (DSLs) which provide familiar APIs to domain experts, code generation frameworks that automate the generation of fast and portable code, and runtime systems that manage threads for concurrency and parallelism. In this paper, we present the Halide code generation framework for Phylanx distributed array processing platform. This extension enables compile-time optimization of Phylanx primitives for target architectures. To accomplish this, (1) we implemented new Phylanx primitives using Halide, and (2) partially exported Halide’s thread pool API to carry out parallelism on HPX (Phylanx’s runtime) threads. (3) showcased HPX performance analysis tools made available to Halide applications. The evaluation of the work has been done in two steps. First, we compare the performance of Halide applications running on its native runtime with that of the new HPX backend to verify there is no cost associated with using HPX threads. Next, we compare performances of a number of original implementations of Phylanx primitives against the new ones in Halide to verify performance and portability benefits of Halide in the context of Phylanx.

KW - AMT

KW - DSL

KW - HPX

KW - Halide

KW - Phylanx

UR - http://www.scopus.com/inward/record.url?scp=85161391629&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85161391629&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-31209-0_3

DO - 10.1007/978-3-031-31209-0_3

M3 - Conference contribution

AN - SCOPUS:85161391629

SN - 9783031312083

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

SP - 32

EP - 45

BT - Euro-Par 2022

A2 - Singer, Jeremy

A2 - Elkhatib, Yehia

A2 - Blanco Heras, Dora

A2 - Diehl, Patrick

A2 - Brown, Nick

A2 - Ilic, Aleksandar

PB - Springer Science and Business Media Deutschland GmbH

T2 - 28th International European Conference on Parallel and Distributed Computing , Euro-Par 2022

Y2 - 22 August 2022 through 26 August 2022

ER -