Publication detail

Case Study on Multi-domain Decomposition of k-Wave Simulation Framework

VAVERKA, F.

Original Title

Case Study on Multi-domain Decomposition of k-Wave Simulation Framework

Type

article in a collection out of WoS and Scopus

Language

English

Original Abstract

This article describes both advantages and challenges of using GPU equipped clusters to implement efficient distributed algorithms. Our main focus is the decomposition of numerical simulations based on (pseudo-)spectral methods. We use a pseudo-spectral simulation of ultrasound wave propagation in homogenous medium (k-Wave) as a showcase of distributed algorithm. We show that we are able to achieve three fold speedup of the simulation while lowering the cost at the same time by employing a Local Fourier Basis decomposition.

Keywords

Pseudo-spectral Simulation, Local Fourier Basis, GPGPU, CUDA, GPU Cluster

Authors

VAVERKA, F.

Released

15. 6. 2016

Publisher

Faculty of Information Technology BUT

Location

Brno

ISBN

978-80-214-5376-0

Book

Computer achitectures and diagnostics 2016

Pages from

37

Pages to

40

Pages count

4

URL

http://www.fit.vutbr.cz/events/pad2016/download/sbornik_pad_2016.pdf

BibTex

@inproceedings{BUT130994,
  author="Filip {Vaverka}",
  title="Case Study on Multi-domain Decomposition of k-Wave Simulation Framework",
  booktitle="Computer achitectures and diagnostics 2016",
  year="2016",
  pages="37--40",
  publisher="Faculty of Information Technology BUT",
  address="Brno",
  isbn="978-80-214-5376-0",
  url="http://www.fit.vutbr.cz/events/pad2016/download/sbornik_pad_2016.pdf"
}