Approaches for Task Affinity in OpenMP

被引：7

作者：

Terboven, Christian ^{[1
]}

Hahnfeld, Jonas ^{[1
]}

Teruel, Xavier ^{[2
]}

Mateo, Sergi ^{[2
]}

Duran, Alejandro ^{[3
]}

Klemm, Michael ^{[3
]}

Olivier, Stephen L. ^{[4
]}

de Supinski, Bronis R. ^{[5
]}

机构：

[1] Rhein Westfal TH Aachen, IT Ctr, Chair High Performance Comp, Aachen, Germany

[2] Barcelona Supercomp Ctr, Barcelona, Spain

[3] Intel, Santa Clara, CA USA

[4] Sandia Natl Labs, Ctr Res Comp, Albuquerque, NM 87185 USA

[5] LLNL, Livermore, CA USA

来源：

OPENMP: MEMORY, DEVICES, AND TASKS | 2016年 / 9903卷

关键词：

D O I：

10.1007/978-3-319-45550-1_8

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

OpenMP tasking supports parallelization of irregular algorithms. Recent OpenMP specifications extended tasking to increase functionality and to support optimizations, for instance with the taskloop construct. However, task scheduling remains opaque, which leads to inconsistent performance on NUMA architectures. We assess design issues for task affinity and explore several approaches to enable it. We evaluate these proposals with implementations in the Nanos++ and LLVM OpenMP runtimes that improve performance up to 40% and significantly reduce execution time variation.

引用

页码：102 / 115

页数：14

共 50 条

[31] Patterns for OpenMP Task Data Dependency Overhead Measurements
Schuchart, Joseph
Nachtmann, Mathias
Gracia, Jose
SCALING OPENMP FOR EXASCALE PERFORMANCE AND PORTABILITY (IWOMP 2017), 2017, 10468 : 156 - 168
[32] Automatic scoping of task clauses for the OpenMP tasking model
Chun-Kun Wang
Peng-Sheng Chen
The Journal of Supercomputing, 2015, 71 : 808 - 823
[33] Enabling OpenMP Task Parallelism on Multi-FPGAs
Nepomuceno, Ramon
Sterle, Renan
Valarini, Guilherme
Pereira, Marcio
Yviquel, Herve
Araujo, Guido
2021 IEEE 29TH ANNUAL INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM 2021), 2021, : 260 - 260
[34] Formalizing Bottlenecks in Task-Based OpenMP Applications
Benedict, Shajulin
Gerndt, Michael
Gudu, Diana-Mihaela
PARALLEL COMPUTING: ACCELERATING COMPUTATIONAL SCIENCE AND ENGINEERING (CSE), 2014, 25 : 103 - 112
[35] OpenMP Extension for Explicit Task Allocation on NUMA Architecture
Lee, Jinpil
Tsugane, Keisuke
Murai, Hitoshi
Sato, Mitsuhisa
OpenMP: Memory, Devices, and Tasks, 2016, 9903 : 89 - 101
[36] Performance Monitoring and Analysis of Task-Based OpenMP
Ding, Yi
Hu, Kai
Wu, Kai
Zhao, Zhenlong
PLOS ONE, 2013, 8 (10):
[37] Automatic scoping of task clauses for the OpenMP tasking model
Wang, Chun-Kun
Chen, Peng-Sheng
JOURNAL OF SUPERCOMPUTING, 2015, 71 (03): : 808 - 823
[38] Adaptive Task Granularity Strategy for OpenMP3.0 Task Model on Cell Architecture
Cao, Qian
Hu, Changjun
Li, Shigang
He, Haohu
HIGH PERFORMANCE NETWORKING, COMPUTING, AND COMMUNICATION SYSTEMS, 2011, 163 : 393 - 400
[39] GLTO: On the Adequacy of Lightweight Thread Approaches for OpenMP Implementations
Castello, Adrian
Mayo, Rafael
Quintana-Orti, Enrique S.
Seo, Sangmin
Balaji, Pavan
Pena, Antonio J.
2017 46TH INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING (ICPP), 2017, : 60 - 69
[40] Multiple Target Task Sharing Support for the OpenMP Accelerator Model
Ozen, Guray
Mateo, Sergi
Ayguade, Eduard
Labarta, Jesus
Beyer, James
OPENMP: MEMORY, DEVICES, AND TASKS, 2016, 9903 : 268 - 280

← 1 2 3 4 5 →