Publication detail
Design Space Exploration of Parallel Embedded Applications Based on Performance-Oriented Specifications
DVOŘÁK, V. STAROBA, J.
Czech title
Průzkum prostoru návrhu paralelních vestavěných aplikací založený na výkonnostně orientovaných specifikacích
English title
Design Space Exploration of Parallel Embedded Applications Based on Performance-Oriented Specifications
Type
article in a collection out of WoS and Scopus
Language
en
Original abstract
The choice of HW/SW multiprocessor architecture for a specific embedded application is not an easy task due to many possible processing elements, type of their interconnection, various partitioning of the code and data and mapping them on processing elements. The paper addresses this issue and uses a single CSP-based simulation tool Transim for multiple architectures. Exploration of a vast design space for a generally heterogeneous network of processing elements is quick, because only first-order effects can be considered. Variations in processor count, clock rate, link speed, bus bandwidth, cache line size, as well as in partitioning and mapping the resulting sw components to processors can be easily accounted for. The technique is demonstrated on solution of large dense systems of linear equations in real time.
Czech abstract
Výběr HW/SW multiprocesorové architektury pro konkrétní vestavěnou aplikaci není snadná úloha kvůli mnoha možným procesním prvkům, typu jejich propojení, rozmanitému členění kódu a dat a jejich mapování na procesní prvky. Článek se zabývá tímto problémem a používá jediný simulační nástroj Transim založený na CSP pro řadu architektur. Průzkum rozsáhlého prostoru návrhu pro obecně heterogenní síť procesních prvků je rychlý, protože se uvažují pouze efekty prvního řádu. Změny v počtu procesorů, hodinovém kmitočtu, rychlosti llinek, propustnosti sběrnice, velikosti bloku skryté paměti i změny v rozčlenění a mapování výsledných sw komponent na procesory je snadno zohledněno. Technika je demonstrována na řešení velkého hustého systému lineárních rovnic v reálném čase.
English abstract
The choice of HW/SW multiprocessor architecture for a specific embedded application is not an easy task due to many possible processing elements, type of their interconnection, various partitioning of the code and data and mapping them on processing elements. The paper addresses this issue and uses a single CSP-based simulation tool Transim for multiple architectures. Exploration of a vast design space for a generally heterogeneous network of processing elements is quick, because only first-order effects can be considered. Variations in processor count, clock rate, link speed, bus bandwidth, cache line size, as well as in partitioning and mapping the resulting sw components to processors can be easily accounted for. The technique is demonstrated on solution of large dense systems of linear equations in real time.
Keywords in English
application-specific architectures, multiprocessor simulation, hardware-software codesign
RIV year
2002
Released
23.04.2002
Publisher
University of Stirling
Location
Stirling
ISBN
1-85769-169-5
Book
Proceedings of the Joint Workshop on Formal Specifications of Computer-Based Systems
Pages from–to
71–75
Pages count
5
BIBTEX
@inproceedings{BUT9824,
author="Václav {Dvořák} and Jiří {Staroba},
title="Design Space Exploration of Parallel Embedded Applications Based on Performance-Oriented Specifications",
booktitle="Proceedings of the Joint Workshop on Formal Specifications of Computer-Based Systems",
year="2002",
month="April",
pages="71--75",
publisher="University of Stirling",
address="Stirling",
isbn="1-85769-169-5"
}