Naz, A., & Toler, M. M. (2019, June), Investigating Communication Patterns for Distributed Fast Fourier Transforms  Paper presented at 2019 ASEE Annual Conference & Exposition , Tampa, Florida. 10.18260/1-2--33024

\bibitem{asee_peer_33024}
  Naz, A., \& Toler, M. M. (2019, June), \emph{Investigating Communication Patterns for Distributed Fast Fourier Transforms}
  Paper presented at 2019 ASEE Annual Conference \& Exposition , Tampa, Florida.
  10.18260/1-2--33024

Afrin Naz and Mardigon Max Toler.  "Investigating Communication Patterns for Distributed Fast Fourier Transforms".  2019 ASEE Annual Conference & Exposition , Tampa, Florida, 2019, June.  ASEE Conferences, 2019.  https://peer.asee.org/33024 Internet.  23 Apr, 2024

\bibitem{asee_peer_33024}
  Afrin Naz and Mardigon Max Toler.
  "Investigating Communication Patterns for Distributed Fast Fourier Transforms".
  \emph{2019 ASEE Annual Conference \& Exposition , Tampa, Florida, 2019, June}.
  ASEE Conferences, 2019.
  https://peer.asee.org/33024 Internet.  23 Apr, 2024

@INPROCEEDINGS{asee_peer_33024
author = "Afrin Naz and Mardigon Max Toler"
title = "Investigating Communication Patterns for Distributed Fast Fourier Transforms"
booktitle = "2019 ASEE Annual Conference \& Exposition "
year = "2019"
month = "June"
address = "Tampa, Florida"
publisher = "ASEE Conferences"
note = {https://peer.asee.org/33024}
number = {10.18260/1-2--33024}
}

TY  - CPAPER
AB  - In this paper, we analyze the effectiveness of different communication patterns for performing a 1-dimensional Fast Fourier Transform on distributed memory computing clusters. The experimental communication pattern is intended to minimize time wasted by participant nodes being inactive during computation. 
Fast Fourier Transforms are a very computationally efficient signal processing technique for calculating the Discrete Fourier Transform of a complex-valued sequence of discrete samples. This Discrete Fourier Transform can be interpreted as a new signal that represents the original signal’s frequency content. Thus, they find the frequency components that make up a signal.
Distributed memory computers are clusters of computers that may all work on a task simultaneously by passing relevant information about the problem’s current state to one another. They are suitable for highly parallelizable routines such as the Fast Fourier Transform, in which the different parts of the input data can be operationally isolated at predictable moments in time, allowing many different computers to independently work on these parts in parallel, ideally reducing the total running time of the algorithm. However, there are many ways that this algorithm can be parallelized. Of particular importance is the way information is initially distributed across the machines and the pattern that the machines use to communicated with each other and collect a final result on a host machine.
We describe how these communication patterns differ and discuss details of their implementation. In addition, data will be presented that illustrates how these implementations differ on two different clusters. 

AU  - Afrin Naz
AU  - Mardigon Max Toler
CY  - Tampa, Florida
DA  - 2019/06/15
PB  - ASEE Conferences
TI  - Investigating Communication Patterns for Distributed Fast Fourier Transforms
UR  - https://peer.asee.org/33024
DO  - 10.18260/1-2--33024
ER  -