Publication: Automatic Parallelism in Differentiation of Fourier Transforms
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About

Automatic Parallelism in Differentiation of Fourier Transforms

- Part of a collection -
 

Author(s)
H. Martin Bücker , Bruno Lang , A. Rasch , Christian H. Bischof

Published in
Proceedings of the 18th ACM Symposium on Applied Computing, Melbourne, Florida, USA, March 9--12, 2003

Year
2003

Publisher
ACM Press

Abstract
For functions given in the form of a computer program, automatic differentiation is an efficient technique to accurately evaluate the derivatives of that function. Starting from a given computer program, automatic differentiation generates another program for the evaluation of the original function and its derivatives in a fully mechanical way. While the efficiency of this black box approach is already high as compared to numerical differentiation based on divided differences, automatic differentiation can be applied even more efficiently by taking into account high-level knowledge about the given computer program. We show that, in the case where the function involves a Fourier transform, the degree of parallelism in the program generated by automatic differentiation can be increased leading to a rich set of automatic parallelization strategies that are not available when employing a black box automatic parallelization approach. Experiments of the new automatic parallelization approach are reported on a SunFire 6800 server using up to 20 processors.

AD Theory and Techniques
Parallelism

BibTeX
@INPROCEEDINGS{
         Bucker2003APi,
       title = "Automatic Parallelism in Differentiation of {F}ourier Transforms",
       booktitle = "Proceedings of the 18th ACM Symposium on Applied Computing, Melbourne, Florida,
         USA, March~9--12, 2003",
       publisher = "ACM Press",
       pages = "148--152",
       doi = "http://doi.acm.org/10.1145/952532.952565",
       address = "New York",
       abstract = "For functions given in the form of a computer program, automatic differentiation is
         an efficient technique to accurately evaluate the derivatives of that function. Starting from a
         given computer program, automatic differentiation generates another program for the evaluation of
         the original function and its derivatives in a fully mechanical way. While the efficiency of this
         black box approach is already high as compared to numerical differentiation based on divided
         differences, automatic differentiation can be applied even more efficiently by taking into account
         high-level knowledge about the given computer program. We show that, in the case where the function
         involves a Fourier transform, the degree of parallelism in the program generated by automatic
         differentiation can be increased leading to a rich set of automatic parallelization strategies that
         are not available when employing a black box automatic parallelization approach. Experiments of the
         new automatic parallelization approach are reported on a SunFire~6800 server using up to 20
         processors.",
       ad_theotech = "Parallelism",
       year = "2003",
       author = "H. Martin B{\"u}cker and Bruno Lang and A. Rasch and Christian
         H.~Bischof"
}


back
  

Contact:
autodiff.org
Username:
Password:
(lost password)