Logo des Repositoriums
 

Performance Prediction for Multicore Environments — An Experiment Report

dc.contributor.authorFrank, Markus
dc.contributor.authorHilbrich, Marcus
dc.date.accessioned2023-03-02T13:44:26Z
dc.date.available2023-03-02T13:44:26Z
dc.date.issued2016
dc.description.abstractMulticore systems are a permanent part of our daily life. Regardless whether we consider nowadays desktop PC’s, notebooks, or smart phones: all devices are running on multicore CPUs. To use such hardware in an efficient way, we need parallel enabled software. But the development of such software is more complex and more error-prone than developing sequential software. To handle the rising complexity, it is necessary to develop software in an engineering way. In such a process, software architects have to plan and analyze software designs on model level. Software architects can use approaches like Palladio to simulate and analyze early phase software designs. However, it is uncertain how Palladio can handle multicore systems. In this paper we evaluate the current state of Palladio regarding multicore awareness based on an experiment. We implemented an easy to parallelize use case, modeled, and simulated it using Palladio. We predicted the performance of an 16 core system with an accuracy of 79 %, but noticed a decreasing accuracy for a rising number of cores. Based on the experiment, we discuss the need to model attributes like memory, memory bandwidth, and caches, which are currently included.en
dc.identifier.pissn0720-8928
dc.identifier.urihttps://dl.gi.de/handle/20.500.12116/40634
dc.language.isoen
dc.publisherGeselllschaft für Informatik e.V.
dc.relation.ispartofSoftwaretechnik-Trends Band 36, Heft 4
dc.titlePerformance Prediction for Multicore Environments — An Experiment Reporten
dc.typeText/Journal Article
gi.citation.publisherPlaceBonn
gi.conference.sessiontitleSonderteil: Proceedings of the Symposium on Software Performance (SSP 2016), 8. - 9. November 2016, Kiel

Dateien

Originalbündel
1 - 1 von 1
Lade...
Vorschaubild
Name:
12-Performance_Prediction_for_Multicore_Environments_-_An_Experiment_Report.pdf
Größe:
248.13 KB
Format:
Adobe Portable Document Format