arrow
Return

Kokkos: Enabling manycore performance portability through polymorphic memory access patterns

delete2014-12-01
delete686
delete
OA
AI
C
Christian Robert Trott
D
Daniel Sunderland
DOI:10.1016/j.jpdc.2014.07.003delete
deleteOriginal
deleteShare
deleteSave
View PDF
Abstract

Abstract

En 中文
The manycore revolution can be characterized by increasing thread counts, decreasing memory per thread, and diversity of continually evolving manycore architectures. High performance computing (HPC) applications and libraries must exploit increasingly finer levels of parallelism within their codes to sustain scalability on these devices. A major obstacle to performance portability is the diverse and conflicting set of constraints on memory access patterns across devices. Contemporary portable programming models address manycore parallelism (e.g., OpenMP, OpenACC, OpenCL) but fail to address memory access patterns. The Kokkos C++ library enables applications and domain libraries to achieve performance portability on diverse manycore architectures by unifying abstractions for both fine-grain data parallelism and memory access patterns. In this paper we describe Kokkos' abstractions, summarize its application programmer interface (API), present performance results for unit-test kernels and mini-applications, and outline an incremental strategy for migrating legacy C++ codes to Kokkos. The Kokkos library is under active research and development to incorporate capabilities from new generations of manycore architectures, and to address a growing list of applications and domain libraries. (C) 2014 Elsevier Inc. All rights reserved.
Keywords:
Parallel computing
Thread parallelism
Manycore
GPU
Performance portability
Multidimensional array
Mini-application
AI Summary

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.

Journal

Journal of Parallel and Distributed Computing cover
Journal of Parallel and Distributed Computing
IF:
4
Papers:
3.8K
Citations:
4.8K

Organization

U
united states department of energy (doe)
Scholars:
11.3W
Papers: 9.6W
Citations: 246
Cited Papers

Cited Papers

Modeling and Mapping of Nanotopography Interactions with CMP
err2011-02-01
err0
PREAI
errBrian Lee; Duane S. Boning; Winthrop Baylies; Noel Poduje; John Valley
errShare
errSave
HIERARCHICAL TASK-BASED PROGRAMMING WITH STARSS
err2009-06-02
err119
PREAI
errPlanas, Judit; Badia, Rosa M.; Ayguade, Eduard; Labarta, Jesus
errShare
errSave
Evaluation of NCS-1, DARPP-32, and neurotrophins in hippocampus and prefrontal cortex in rats submitted to sepsis
err2014-07-16
err0
PREAI
errClarissa M. Comim; Napoleão C. Silva; Francielle Mina; Diogo Dominguini; Giselli Scaini; Meline O.S. Morais; Daniela V. Rosa; Luiz Alexandre V. Magno; Emilio L. Streck; Marco A. Romano-Silva; João Quevedo; Felipe Dal-Pizzol
errShare
errSave
CRISPR tracking reveals global spreading of antimicrobial resistance genes by Staphylococcus of canine origin
err2019-05-01
err0
PREAI
errCiro César Rossi; Ana Luisa Andrade-Oliveira; Marcia Giambiagi-deMarval
errShare
errSave
no more