Toward adjoinable MPI
Name
Utke-2009-Toward adjoinable MPI.pdf
Size
220.08 KB
Format
Adobe PDF
Checksum (MD5)
01ff70b03eb86361a42e0281082fa43c
Author(s) • • • • •
Hill, Christopher N.
Utke, Jean
Hascoet, Laurent
Heimbach, Patrick
Hovland, Paul
Naumann, Uwe
Date Issued
July 2009
Journal
IEEE International Symposium on Parallel & Distributed Processing, 2009. IPDPS 2009
Publisher
Institute of Electrical and Electronics Engineers
Citation
Utke, J. et al. “Toward adjoinable MPI.” Parallel & Distributed Processing, 2009. IPDPS 2009. IEEE International Symposium on. 2009. 1-8. ©2009 Institute of Electrical and Electronics Engineers.
Version
Final published version
Abstract
Automatic differentiation is the primary means of obtaining analytic derivatives from a numerical model given as a computer program. Therefore, it is an essential productivity tool in numerous computational science and engineering domains. Computing gradients with the adjoint (also called reverse) mode via source transformation is a particularly beneficial but also challenging use of automatic differentiation. To date only ad hoc solutions for adjoint differentiation of MPI programs have been available, forcing automatic differentiation tool users to reason about parallel communication dataflow and dependencies and manually develop adjoint communication code. Using the communication graph as a model we characterize the principal problems of adjoining the most frequently used communication idioms. We propose solutions to cover these idioms and consider the consequences for the MPI implementation, the MPI user and MPI-aware program analysis. The MIT general circulation model serves as a use case to illustrate the viability of our approach.
Subjects
MPI
automatic differentiation
reverse mode
source transformation
MIT Department
Massachusetts Institute of Technology. Department of Earth, Atmospheric, and Planetary Sciences
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1109/IPDPS.2009.5161165