PET-to-MLIR: A polyhedral front-end for MLIR

Konrad Komisarczyk; Lorenzo Chelini; Kanishkan Vadivel; Roel Jordans; Henk Corporaal

doi:10.1109/DSD51259.2020.00091

PET-to-MLIR: A polyhedral front-end for MLIR

Konrad Komisarczyk, Lorenzo Chelini, Kanishkan Vadivel, Roel Jordans, Henk Corporaal

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

4 Citations (Scopus)

718 Downloads (Pure)

Abstract

We present PET-to-MLIR, a new tool to enter the MLIR compiler framework from C source. The tool is based on the popular PET and ISL libraries for extracting and manipulating quasi-affine sets and relations, and Loop Tactics, a declarative optimizer. The use of PET brings advanced diagnosis and full support for C by relying on the Clang parser. ISL allows easy manipulation of the polyhedral representation and efficient code generation. Loop Tactics, on the other hand, enable us to detect computational motifs transparently and lift the entry point in MLIR, thus enabling domain-specific optimizations in general-purpose code.

We demonstrate our tool using the Polybench/C benchmark suite and show that it can lower most of the benchmarks to the MLIR’s affine dialect successfully. We believe that our tool can benefit research in the compiler community by providing an automatic way to translate C code to the MLIR affine dialect.

Original language	English
Title of host publication	2020 23rd Euromicro Conference on Digital System Design (DSD)
Editors	Andrej Trost, Andrej Zemva, Amund Skavhaug
Publisher	Institute of Electrical and Electronics Engineers
Pages	551-556
Number of pages	6
ISBN (Electronic)	9781728195353
DOIs	https://doi.org/10.1109/DSD51259.2020.00091
Publication status	Published - 8 Oct 2020
Event	23rd Euromicro Conference on Digital System Design, DSD 2020 - Kranj, Slovenia, Kranj, Slovenia Duration: 26 Aug 2020 → 28 Aug 2020

Conference

Conference	23rd Euromicro Conference on Digital System Design, DSD 2020
Country/Territory	Slovenia
City	Kranj
Period	26/08/20 → 28/08/20

Keywords

Loop Tactics
MLIR
polyhedral model

Access to Document

10.1109/DSD51259.2020.00091

Paper DSD Pet_to_MLIRLicence: CC BY-NC-ND

Cite this

@inproceedings{329922007156419298a4af949c216f28,

title = "PET-to-MLIR: A polyhedral front-end for MLIR",

abstract = "We present PET-to-MLIR, a new tool to enter the MLIR compiler framework from C source. The tool is based on the popular PET and ISL libraries for extracting and manipulating quasi-affine sets and relations, and Loop Tactics, a declarative optimizer. The use of PET brings advanced diagnosis and full support for C by relying on the Clang parser. ISL allows easy manipulation of the polyhedral representation and efficient code generation. Loop Tactics, on the other hand, enable us to detect computational motifs transparently and lift the entry point in MLIR, thus enabling domain-specific optimizations in general-purpose code.We demonstrate our tool using the Polybench/C benchmark suite and show that it can lower most of the benchmarks to the MLIR{\textquoteright}s affine dialect successfully. We believe that our tool can benefit research in the compiler community by providing an automatic way to translate C code to the MLIR affine dialect.",

keywords = "Loop Tactics, MLIR, polyhedral model",

author = "Konrad Komisarczyk and Lorenzo Chelini and Kanishkan Vadivel and Roel Jordans and Henk Corporaal",

year = "2020",

month = oct,

day = "8",

doi = "10.1109/DSD51259.2020.00091",

language = "English",

pages = "551--556",

editor = "Andrej Trost and Andrej Zemva and Amund Skavhaug",

booktitle = "2020 23rd Euromicro Conference on Digital System Design (DSD)",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

note = "23rd Euromicro Conference on Digital System Design, DSD 2020 ; Conference date: 26-08-2020 Through 28-08-2020",

}

Komisarczyk, K, Chelini, L, Vadivel, K , Jordans, R & Corporaal, H 2020, PET-to-MLIR: A polyhedral front-end for MLIR. in A Trost, A Zemva & A Skavhaug (eds), 2020 23rd Euromicro Conference on Digital System Design (DSD)., 9217876, Institute of Electrical and Electronics Engineers, pp. 551-556, 23rd Euromicro Conference on Digital System Design, DSD 2020, Kranj, Slovenia, 26/08/20. https://doi.org/10.1109/DSD51259.2020.00091

PET-to-MLIR: A polyhedral front-end for MLIR. / Komisarczyk, Konrad; Chelini, Lorenzo; Vadivel, Kanishkan et al.
2020 23rd Euromicro Conference on Digital System Design (DSD). ed. / Andrej Trost; Andrej Zemva; Amund Skavhaug. Institute of Electrical and Electronics Engineers, 2020. p. 551-556 9217876.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - PET-to-MLIR

T2 - 23rd Euromicro Conference on Digital System Design, DSD 2020

AU - Komisarczyk, Konrad

AU - Chelini, Lorenzo

AU - Vadivel, Kanishkan

AU - Jordans, Roel

AU - Corporaal, Henk

PY - 2020/10/8

Y1 - 2020/10/8

N2 - We present PET-to-MLIR, a new tool to enter the MLIR compiler framework from C source. The tool is based on the popular PET and ISL libraries for extracting and manipulating quasi-affine sets and relations, and Loop Tactics, a declarative optimizer. The use of PET brings advanced diagnosis and full support for C by relying on the Clang parser. ISL allows easy manipulation of the polyhedral representation and efficient code generation. Loop Tactics, on the other hand, enable us to detect computational motifs transparently and lift the entry point in MLIR, thus enabling domain-specific optimizations in general-purpose code.We demonstrate our tool using the Polybench/C benchmark suite and show that it can lower most of the benchmarks to the MLIR’s affine dialect successfully. We believe that our tool can benefit research in the compiler community by providing an automatic way to translate C code to the MLIR affine dialect.

AB - We present PET-to-MLIR, a new tool to enter the MLIR compiler framework from C source. The tool is based on the popular PET and ISL libraries for extracting and manipulating quasi-affine sets and relations, and Loop Tactics, a declarative optimizer. The use of PET brings advanced diagnosis and full support for C by relying on the Clang parser. ISL allows easy manipulation of the polyhedral representation and efficient code generation. Loop Tactics, on the other hand, enable us to detect computational motifs transparently and lift the entry point in MLIR, thus enabling domain-specific optimizations in general-purpose code.We demonstrate our tool using the Polybench/C benchmark suite and show that it can lower most of the benchmarks to the MLIR’s affine dialect successfully. We believe that our tool can benefit research in the compiler community by providing an automatic way to translate C code to the MLIR affine dialect.

KW - Loop Tactics

KW - MLIR

KW - polyhedral model

UR - http://www.scopus.com/inward/record.url?scp=85096354372&partnerID=8YFLogxK

U2 - 10.1109/DSD51259.2020.00091

DO - 10.1109/DSD51259.2020.00091

M3 - Conference contribution

SP - 551

EP - 556

BT - 2020 23rd Euromicro Conference on Digital System Design (DSD)

A2 - Trost, Andrej

A2 - Zemva, Andrej

A2 - Skavhaug, Amund

PB - Institute of Electrical and Electronics Engineers

Y2 - 26 August 2020 through 28 August 2020

ER -

PET-to-MLIR: A polyhedral front-end for MLIR

Abstract

Conference

Keywords

Access to Document

Other files and links

MNEMOSENE - Computation-in-memory architecture based on resistive devices

NeMeCo

Cite this

PET-to-MLIR: A polyhedral front-end for MLIR

Abstract

Conference

Keywords

Access to Document

Other files and links

Projects

MNEMOSENE - Computation-in-memory architecture based on resistive devices

NeMeCo

Cite this