Algorithm-Architecture Affinity - Parallelism Changes the Picture

Rasmus Abildgren; Aleksandras Šaramentovas; Paulius Ruzgys; Peter Koch; Yannick Le Moullec

Algorithm-Architecture Affinity - Parallelism Changes the Picture

Rasmus Abildgren, Aleksandras Šaramentovas, Paulius Ruzgys, Peter Koch, Yannick Le Moullec

Research output: Contribution to conference without publisher/journal › Paper without publisher/journal › Research › peer-review

477 Downloads (Pure)

Abstract

Reducing the time-to-market factor is a challenge
for many embedded systems designers. In that respect, hardwaresoftware
partitioning is a key issue which has been studied
during the last two decades. In this paper we present an
extension to recent works dealing with metrics for guiding the
hardware-software partitioning step. This extension builds upon
and complement our own work with metrics in the Design Trotter
project, and is combined with the affinity metric approach. We
show that the proposed extension improves the original affinity
metric in terms of parallelism detection, and thus can help
system designers to make wiser hardware-software partitioning
decisions, which in turn reduces the time-to-market factor.

Original language	English
Publication date	2007
Number of pages	4
Publication status	Published - 2007
Event	DASIP 2007 - Grenoble, France Duration: 27 Nov 2007 → 29 Nov 2007

Conference

Conference	DASIP 2007
Country/Territory	France
City	Grenoble
Period	27/11/2007 → 29/11/2007

Keywords

HW/SW Partitioning
Algorithm-Architecture Matching

Access to Document

DASIP07Final published version, 165 KB

http://www.ecsi.org/sites/default/files/DASIP2007proceedings.pdf

AUB Link

Search for the material in Aalborg University Library's search engine

Methods for accelerated design for FPGA technology
Abildgren, R., Koch, P., Le Moullec, Y. & Knudsen, R. B.
<ingen navn>
01/10/2005 → 01/05/2009
Project: Research

Cite this

@conference{ed0e05303f4311dc912d000ea68e967b,

title = "Algorithm-Architecture Affinity - Parallelism Changes the Picture",

abstract = "Reducing the time-to-market factor is a challengefor many embedded systems designers. In that respect, hardwaresoftwarepartitioning is a key issue which has been studiedduring the last two decades. In this paper we present anextension to recent works dealing with metrics for guiding thehardware-software partitioning step. This extension builds uponand complement our own work with metrics in the Design Trotterproject, and is combined with the affinity metric approach. Weshow that the proposed extension improves the original affinitymetric in terms of parallelism detection, and thus can helpsystem designers to make wiser hardware-software partitioningdecisions, which in turn reduces the time-to-market factor.",

keywords = "HW/SW partitionering, Algoritme-Arkitektur Sammenligning, HW/SW Partitioning, Algorithm-Architecture Matching",

author = "Rasmus Abildgren and Aleksandras {\v S}aramentovas and Paulius Ruzgys and Peter Koch and {Le Moullec}, Yannick",

year = "2007",

language = "English",

note = "DASIP 2007 ; Conference date: 27-11-2007 Through 29-11-2007",

}

TY - CONF

T1 - Algorithm-Architecture Affinity - Parallelism Changes the Picture

AU - Abildgren, Rasmus

AU - Šaramentovas, Aleksandras

AU - Ruzgys, Paulius

AU - Koch, Peter

AU - Le Moullec, Yannick

PY - 2007

Y1 - 2007

N2 - Reducing the time-to-market factor is a challengefor many embedded systems designers. In that respect, hardwaresoftwarepartitioning is a key issue which has been studiedduring the last two decades. In this paper we present anextension to recent works dealing with metrics for guiding thehardware-software partitioning step. This extension builds uponand complement our own work with metrics in the Design Trotterproject, and is combined with the affinity metric approach. Weshow that the proposed extension improves the original affinitymetric in terms of parallelism detection, and thus can helpsystem designers to make wiser hardware-software partitioningdecisions, which in turn reduces the time-to-market factor.

AB - Reducing the time-to-market factor is a challengefor many embedded systems designers. In that respect, hardwaresoftwarepartitioning is a key issue which has been studiedduring the last two decades. In this paper we present anextension to recent works dealing with metrics for guiding thehardware-software partitioning step. This extension builds uponand complement our own work with metrics in the Design Trotterproject, and is combined with the affinity metric approach. Weshow that the proposed extension improves the original affinitymetric in terms of parallelism detection, and thus can helpsystem designers to make wiser hardware-software partitioningdecisions, which in turn reduces the time-to-market factor.

KW - HW/SW partitionering

KW - Algoritme-Arkitektur Sammenligning

KW - HW/SW Partitioning

KW - Algorithm-Architecture Matching

M3 - Paper without publisher/journal

T2 - DASIP 2007

Y2 - 27 November 2007 through 29 November 2007

ER -

Algorithm-Architecture Affinity - Parallelism Changes the Picture

Abstract

Conference

Keywords

Access to Document

AUB Link

Fingerprint

Projects

Methods for accelerated design for FPGA technology

Cite this