Streaming-enabled parallel dataflow architecture for multicore systems

Huy T. Vo, Daniel K. Osmari, Brian Summa, João L D Comba, Valerio Pascucci, Cláudio T. Silva

Research output: Contribution to journalArticle

Abstract

We propose a new framework design for exploiting multi-core architectures in the context of visualization dataflow systems. Recent hardware advancements have greatly increased the levels of parallelism available with all indications showing this trend will continue in the future. Existing visualization dataflow systems have attempted to take advantage of these new resources, though they still have a number of limitations when deployed on shared memory multi-core architectures. Ideally, visualization systems should be built on top of a parallel dataflow scheme that can optimally utilize CPUs and assign resources adaptively to pipeline elements. We propose the design of a flexible dataflow architecture aimed at addressing many of the shortcomings of existing systems including a unified execution model for both demand-driven and event-driven models; a resource scheduler that can automatically make decisions on how to allocate computing resources; and support for more general streaming data structures which include unstructured elements. We have implemented our system on top of VTK with backward compatibility. In this paper, we provide evidence of performance improvements on a number of applications.

Original languageEnglish (US)
Pages (from-to)1073-1082
Number of pages10
JournalComputer Graphics Forum
Volume29
Issue number3
DOIs
StatePublished - Jun 2010

Fingerprint

Visualization
Program processors
Data structures
Pipelines
Hardware
Data storage equipment

Keywords

  • C.1.3 [Processor Architectures]: Other Architecture Styles - Data-flow architectures

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Streaming-enabled parallel dataflow architecture for multicore systems. / Vo, Huy T.; Osmari, Daniel K.; Summa, Brian; Comba, João L D; Pascucci, Valerio; Silva, Cláudio T.

In: Computer Graphics Forum, Vol. 29, No. 3, 06.2010, p. 1073-1082.

Research output: Contribution to journalArticle

Vo, Huy T. ; Osmari, Daniel K. ; Summa, Brian ; Comba, João L D ; Pascucci, Valerio ; Silva, Cláudio T. / Streaming-enabled parallel dataflow architecture for multicore systems. In: Computer Graphics Forum. 2010 ; Vol. 29, No. 3. pp. 1073-1082.
@article{ae0e86c24bbb4a26b9a44f0220f83835,
title = "Streaming-enabled parallel dataflow architecture for multicore systems",
abstract = "We propose a new framework design for exploiting multi-core architectures in the context of visualization dataflow systems. Recent hardware advancements have greatly increased the levels of parallelism available with all indications showing this trend will continue in the future. Existing visualization dataflow systems have attempted to take advantage of these new resources, though they still have a number of limitations when deployed on shared memory multi-core architectures. Ideally, visualization systems should be built on top of a parallel dataflow scheme that can optimally utilize CPUs and assign resources adaptively to pipeline elements. We propose the design of a flexible dataflow architecture aimed at addressing many of the shortcomings of existing systems including a unified execution model for both demand-driven and event-driven models; a resource scheduler that can automatically make decisions on how to allocate computing resources; and support for more general streaming data structures which include unstructured elements. We have implemented our system on top of VTK with backward compatibility. In this paper, we provide evidence of performance improvements on a number of applications.",
keywords = "C.1.3 [Processor Architectures]: Other Architecture Styles - Data-flow architectures",
author = "Vo, {Huy T.} and Osmari, {Daniel K.} and Brian Summa and Comba, {Jo{\~a}o L D} and Valerio Pascucci and Silva, {Cl{\'a}udio T.}",
year = "2010",
month = "6",
doi = "10.1111/j.1467-8659.2009.01704.x",
language = "English (US)",
volume = "29",
pages = "1073--1082",
journal = "Computer Graphics Forum",
issn = "0167-7055",
publisher = "Wiley-Blackwell",
number = "3",

}

TY - JOUR

T1 - Streaming-enabled parallel dataflow architecture for multicore systems

AU - Vo, Huy T.

AU - Osmari, Daniel K.

AU - Summa, Brian

AU - Comba, João L D

AU - Pascucci, Valerio

AU - Silva, Cláudio T.

PY - 2010/6

Y1 - 2010/6

N2 - We propose a new framework design for exploiting multi-core architectures in the context of visualization dataflow systems. Recent hardware advancements have greatly increased the levels of parallelism available with all indications showing this trend will continue in the future. Existing visualization dataflow systems have attempted to take advantage of these new resources, though they still have a number of limitations when deployed on shared memory multi-core architectures. Ideally, visualization systems should be built on top of a parallel dataflow scheme that can optimally utilize CPUs and assign resources adaptively to pipeline elements. We propose the design of a flexible dataflow architecture aimed at addressing many of the shortcomings of existing systems including a unified execution model for both demand-driven and event-driven models; a resource scheduler that can automatically make decisions on how to allocate computing resources; and support for more general streaming data structures which include unstructured elements. We have implemented our system on top of VTK with backward compatibility. In this paper, we provide evidence of performance improvements on a number of applications.

AB - We propose a new framework design for exploiting multi-core architectures in the context of visualization dataflow systems. Recent hardware advancements have greatly increased the levels of parallelism available with all indications showing this trend will continue in the future. Existing visualization dataflow systems have attempted to take advantage of these new resources, though they still have a number of limitations when deployed on shared memory multi-core architectures. Ideally, visualization systems should be built on top of a parallel dataflow scheme that can optimally utilize CPUs and assign resources adaptively to pipeline elements. We propose the design of a flexible dataflow architecture aimed at addressing many of the shortcomings of existing systems including a unified execution model for both demand-driven and event-driven models; a resource scheduler that can automatically make decisions on how to allocate computing resources; and support for more general streaming data structures which include unstructured elements. We have implemented our system on top of VTK with backward compatibility. In this paper, we provide evidence of performance improvements on a number of applications.

KW - C.1.3 [Processor Architectures]: Other Architecture Styles - Data-flow architectures

UR - http://www.scopus.com/inward/record.url?scp=77955738660&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955738660&partnerID=8YFLogxK

U2 - 10.1111/j.1467-8659.2009.01704.x

DO - 10.1111/j.1467-8659.2009.01704.x

M3 - Article

AN - SCOPUS:77955738660

VL - 29

SP - 1073

EP - 1082

JO - Computer Graphics Forum

JF - Computer Graphics Forum

SN - 0167-7055

IS - 3

ER -