Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)

Lingqi Zhang; Mohamed Wahib; Peng Chen; Jintao Meng; Xiao Wang; Toshio Endo; Satoshi Matsuoka

doi:10.1145/3589236.3589242

Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)

Lingqi Zhang, Mohamed Wahib, Peng Chen, Jintao Meng, Xiao Wang, Toshio Endo, Satoshi Matsuoka

Multiscale Biomedical Systems

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

Abstract

General Purpose Graphics Processing Units (GPGPU) are used in most of the top systems in HPC. The total capacity of scratchpad memory has increased by more than 40 times in the last decade. However, existing optimizations for stencil computations using temporal blocking have not aggressively exploited the large capacity of scratchpad memory. This work uses the 2D Jacobian 5-point iterative stencil as a case study to investigate the use of large scratchpad memory. Unlike existing research that tiles the domain in a thread block fashion, we tile the domain so that each tile is large enough to utilize all available scratchpad memory on the GPU. Consequently, we process several time steps inside a single tile before offloading the result back to global memory. Our evaluation shows that our performance is comparable to state-of-the-art implementations, yet our implementation is much simpler and does not require auto-generation of code.

Original language	English
Title of host publication	Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023
Publisher	Association for Computing Machinery
Pages	34-35
Number of pages	2
ISBN (Electronic)	9798400707766
DOIs	https://doi.org/10.1145/3589236.3589242
State	Published - Feb 25 2023
Event	15th Annual Workshop on General Purpose Processing using Graphics Processing Unit, GPGPU 2023 - Montreal, Canada Duration: Feb 25 2023 → …

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	15th Annual Workshop on General Purpose Processing using Graphics Processing Unit, GPGPU 2023
Country/Territory	Canada
City	Montreal
Period	02/25/23 → …

Funding

This work was supported by JSPS KAKENHI under Grant Number JP21K17750. This paper is based on results obtained from a project, JPNP20006, commissioned by the New Energy and Industrial Technology Development Organization (NEDO). This research used resources at the Oak Ridge Leadership Computing Facility, a DOE Office of Science User Facility operated by the Oak Ridge National Laboratory. The authors wish to express their sincere appreciation to Jens Domke, Aleksandr Drozd, Emil Vatai and other RIKEN R-CCS colleagues for their invaluable advice and guidance throughout the course of this research. Finally, the first author would also like to express his gratitude to RIKEN R-CCS for offering the opportunity to undertake this research in an intern program.

Keywords

GPGPU
Iterative Stencil Solvers
Temporal Blocking

Access to Document

10.1145/3589236.3589242

Cite this

Zhang, L., Wahib, M., Chen, P., Meng, J., Wang, X., Endo, T., & Matsuoka, S. (2023). Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt). In Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023 (pp. 34-35). (ACM International Conference Proceeding Series). Association for Computing Machinery. https://doi.org/10.1145/3589236.3589242

@inproceedings{610bcc5ddc2747cc88ba7f80b450ad2d,

title = "Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)",

abstract = "General Purpose Graphics Processing Units (GPGPU) are used in most of the top systems in HPC. The total capacity of scratchpad memory has increased by more than 40 times in the last decade. However, existing optimizations for stencil computations using temporal blocking have not aggressively exploited the large capacity of scratchpad memory. This work uses the 2D Jacobian 5-point iterative stencil as a case study to investigate the use of large scratchpad memory. Unlike existing research that tiles the domain in a thread block fashion, we tile the domain so that each tile is large enough to utilize all available scratchpad memory on the GPU. Consequently, we process several time steps inside a single tile before offloading the result back to global memory. Our evaluation shows that our performance is comparable to state-of-the-art implementations, yet our implementation is much simpler and does not require auto-generation of code.",

keywords = "GPGPU, Iterative Stencil Solvers, Temporal Blocking",

author = "Lingqi Zhang and Mohamed Wahib and Peng Chen and Jintao Meng and Xiao Wang and Toshio Endo and Satoshi Matsuoka",

note = "Publisher Copyright: {\textcopyright} 2023 Copyright held by the owner/author(s).; 15th Annual Workshop on General Purpose Processing using Graphics Processing Unit, GPGPU 2023 ; Conference date: 25-02-2023",

year = "2023",

month = feb,

day = "25",

doi = "10.1145/3589236.3589242",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

pages = "34--35",

booktitle = "Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023",

}

Zhang, L, Wahib, M, Chen, P, Meng, J, Wang, X, Endo, T & Matsuoka, S 2023, Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt). in Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023. ACM International Conference Proceeding Series, Association for Computing Machinery, pp. 34-35, 15th Annual Workshop on General Purpose Processing using Graphics Processing Unit, GPGPU 2023, Montreal, Canada, 02/25/23. https://doi.org/10.1145/3589236.3589242

Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt). / Zhang, Lingqi; Wahib, Mohamed; Chen, Peng et al.
Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023. Association for Computing Machinery, 2023. p. 34-35 (ACM International Conference Proceeding Series).

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

TY - GEN

T1 - Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)

AU - Zhang, Lingqi

AU - Wahib, Mohamed

AU - Chen, Peng

AU - Meng, Jintao

AU - Wang, Xiao

AU - Endo, Toshio

AU - Matsuoka, Satoshi

PY - 2023/2/25

Y1 - 2023/2/25

N2 - General Purpose Graphics Processing Units (GPGPU) are used in most of the top systems in HPC. The total capacity of scratchpad memory has increased by more than 40 times in the last decade. However, existing optimizations for stencil computations using temporal blocking have not aggressively exploited the large capacity of scratchpad memory. This work uses the 2D Jacobian 5-point iterative stencil as a case study to investigate the use of large scratchpad memory. Unlike existing research that tiles the domain in a thread block fashion, we tile the domain so that each tile is large enough to utilize all available scratchpad memory on the GPU. Consequently, we process several time steps inside a single tile before offloading the result back to global memory. Our evaluation shows that our performance is comparable to state-of-the-art implementations, yet our implementation is much simpler and does not require auto-generation of code.

AB - General Purpose Graphics Processing Units (GPGPU) are used in most of the top systems in HPC. The total capacity of scratchpad memory has increased by more than 40 times in the last decade. However, existing optimizations for stencil computations using temporal blocking have not aggressively exploited the large capacity of scratchpad memory. This work uses the 2D Jacobian 5-point iterative stencil as a case study to investigate the use of large scratchpad memory. Unlike existing research that tiles the domain in a thread block fashion, we tile the domain so that each tile is large enough to utilize all available scratchpad memory on the GPU. Consequently, we process several time steps inside a single tile before offloading the result back to global memory. Our evaluation shows that our performance is comparable to state-of-the-art implementations, yet our implementation is much simpler and does not require auto-generation of code.

KW - GPGPU

KW - Iterative Stencil Solvers

KW - Temporal Blocking

UR - https://www.scopus.com/pages/publications/85163890843

U2 - 10.1145/3589236.3589242

DO - 10.1145/3589236.3589242

M3 - Conference contribution

AN - SCOPUS:85163890843

T3 - ACM International Conference Proceeding Series

SP - 34

EP - 35

BT - Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023

PB - Association for Computing Machinery

T2 - 15th Annual Workshop on General Purpose Processing using Graphics Processing Unit, GPGPU 2023

Y2 - 25 February 2023

ER -

Zhang L, Wahib M, Chen P, Meng J, Wang X, Endo T et al. Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt). In Proceedings of the 15th Workshop on General Purpose Processing Using GPU, GPGPU 2023. Association for Computing Machinery. 2023. p. 34-35. (ACM International Conference Proceeding Series). doi: 10.1145/3589236.3589242

Exploiting Scratchpad Memory for Deep Temporal Blocking A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this