Summary
We give an overview of some contemporary processor developments and draw conclusions for developments of numerical algorithms. Multicore CPUs are developed by nearly all hardware manufactures. The bandwidth peak performance relation is decreasing. Both have consequences for the efficiency of todays computer codes. We propose a Finite Difference discretization technique which is decreasing the data intensity.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
http://www.clearspeed.com/products/si.php
http://www-306.ibm.com/chips/techlib/techlib.nsf/products/Cell
http://www.ibm.com/developerworks/power/cell/
http://researchweb.watson.ibm.com/journal/rd/494/kahle.pdf
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2006 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Küster, U. (2006). Challenges of future hardware development and consequences for numerical algorithms. In: Shokin, Y., Resch, M., Shokina, N., Danaev, N., Orunkhanov, M. (eds) Advances in High Performance Computing and Computational Sciences. Notes on Numerical Fluid Mechanics and Multidisciplinary Design (NNFM), vol 93. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-33844-4_9
Download citation
DOI: https://doi.org/10.1007/978-3-540-33844-4_9
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-33864-2
Online ISBN: 978-3-540-33844-4
eBook Packages: EngineeringEngineering (R0)