### Abstract

Original language | English |
---|---|

Title of host publication | Proceedings of the 4th North American Transputer User Group Meeting |

Place of Publication | Amsterdam |

Publisher | IOS Press |

Pages | - |

Number of pages | 0 |

ISBN (Print) | 9051990405 |

Publication status | Published - 1 Sep 1990 |

### Publication series

Name | |
---|---|

Publisher | IOS Press |

### Fingerprint

### Keywords

- IR-96060
- METIS-113531

### Cite this

*Proceedings of the 4th North American Transputer User Group Meeting*(pp. -). Amsterdam: IOS Press.

}

*Proceedings of the 4th North American Transputer User Group Meeting.*IOS Press, Amsterdam, pp. -.

**A generalized FFT algorithm on transputers.** / Roebbers, Herman; Roebbers, Herman; Welch, P.H.; Welch, Peter; Wijbrans, K.C.J.; Wijbrans, Klaas.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - A generalized FFT algorithm on transputers

AU - Roebbers, Herman

AU - Roebbers, Herman

AU - Welch, P.H.

AU - Welch, Peter

AU - Wijbrans, K.C.J.

AU - Wijbrans, Klaas

PY - 1990/9/1

Y1 - 1990/9/1

N2 - A generalized algorithm has been derived for the execution of the Cooley-Tukey FFT algorithm on a distributed memory machine. This algorithm is based on an approach that combines a large number of butterfly operations into one large process per processor. The performance can be predicted from theory. The actual algorithm has been implemented on a transputer array, and the performance of the implementation has been measured for various sizes of the complex input vector. It is shown that the algorithm scales linearly with the number of transputers and the problem size.

AB - A generalized algorithm has been derived for the execution of the Cooley-Tukey FFT algorithm on a distributed memory machine. This algorithm is based on an approach that combines a large number of butterfly operations into one large process per processor. The performance can be predicted from theory. The actual algorithm has been implemented on a transputer array, and the performance of the implementation has been measured for various sizes of the complex input vector. It is shown that the algorithm scales linearly with the number of transputers and the problem size.

KW - IR-96060

KW - METIS-113531

M3 - Conference contribution

SN - 9051990405

SP - -

BT - Proceedings of the 4th North American Transputer User Group Meeting

PB - IOS Press

CY - Amsterdam

ER -