DOE Patents title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Effecting a broadcast with an allreduce operation on a parallel computer

Abstract

A parallel computer comprises a plurality of compute nodes organized into at least one operational group for collective parallel operations. Each compute node is assigned a unique rank and is coupled for data communications through a global combining network. One compute node is assigned to be a logical root. A send buffer and a receive buffer is configured. Each element of a contribution of the logical root in the send buffer is contributed. One or more zeros corresponding to a size of the element are injected. An allreduce operation with a bitwise OR using the element and the injected zeros is performed. And the result for the allreduce operation is determined and stored in each receive buffer.

Inventors:
 [1];  [2];  [2];  [2]
  1. Ardsley, NY
  2. Rochester, MN
Issue Date:
Research Org.:
International Business Machines Corp., Armonk, NY (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1017453
Patent Number(s):
7827385
Application Number:
11/832,918
Assignee:
International Business Machines Corporation (Armonk, NY)
Patent Classifications (CPCs):
G - PHYSICS G06 - COMPUTING G06F - ELECTRIC DIGITAL DATA PROCESSING
DOE Contract Number:  
B519700
Resource Type:
Patent
Country of Publication:
United States
Language:
English

Citation Formats

Almasi, Gheorghe, Archer, Charles J, Ratterman, Joseph D, and Smith, Brian E. Effecting a broadcast with an allreduce operation on a parallel computer. United States: N. p., 2010. Web.
Almasi, Gheorghe, Archer, Charles J, Ratterman, Joseph D, & Smith, Brian E. Effecting a broadcast with an allreduce operation on a parallel computer. United States.
Almasi, Gheorghe, Archer, Charles J, Ratterman, Joseph D, and Smith, Brian E. Tue . "Effecting a broadcast with an allreduce operation on a parallel computer". United States. https://www.osti.gov/servlets/purl/1017453.
@article{osti_1017453,
title = {Effecting a broadcast with an allreduce operation on a parallel computer},
author = {Almasi, Gheorghe and Archer, Charles J and Ratterman, Joseph D and Smith, Brian E},
abstractNote = {A parallel computer comprises a plurality of compute nodes organized into at least one operational group for collective parallel operations. Each compute node is assigned a unique rank and is coupled for data communications through a global combining network. One compute node is assigned to be a logical root. A send buffer and a receive buffer is configured. Each element of a contribution of the logical root in the send buffer is contributed. One or more zeros corresponding to a size of the element are injected. An allreduce operation with a bitwise OR using the element and the injected zeros is performed. And the result for the allreduce operation is determined and stored in each receive buffer.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2010},
month = {11}
}

Works referenced in this record:

Computing parallel prefix and reduction using coterie structures
conference, January 1992

  • Herbordt, M. C.; Weems, C. C.
  • [1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation, [Proceedings 1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation
  • https://doi.org/10.1109/FMPC.1992.234895

Optimization of MPI collectives on clusters of large-scale SMP's
conference, January 1999


Universality of mixed action extrapolation formulae
journal, April 2009


Computing the Hough transform on a scan line array processor (image processing)
journal, March 1989


Interleaved all-to-all reliable broadcast on meshes and hypercubes
journal, May 1994


Efficient algorithms for all-to-all communications in multiport message-passing systems
journal, January 1997