*banner
 

A generalized framework for modeling and scheduling heterogeneous dataflow applications
Nimish Sane

Citation
Nimish Sane. "A generalized framework for modeling and scheduling heterogeneous dataflow applications". Talk or presentation, 16, April, 2009; Poster presented at the 8th Biennial Ptolemy Miniconference.

Abstract
MocGraph is a Java-based package of generic graph data structures and algorithms with emphasis on supporting graph-theoretic analysis for models of computation (mocs). MocGraph has evolved from the the graph package in Ptolemy II. In our recent work on the MocGraph package, we have added support for trees and generalized schedule trees (GSTs), in particular. A GST is a data structure for representing schedules of dataflow graphs. It is a general schedule representation in that it can be used to represent dataflow schedules for a wide variety of dataflow models of computation. As part of the DIF (dataflow interchange format) project at the University of Maryland, we have also developed the core functional dataflow (CFDF) model of computation. CFDF facilitates natural deion of actors and provides a general framework for specifying static and dynamic dataflow applications. We have demonstrated that a variety of commonly used, existing dataflow models can be represented naturally in CFDF. The CFDF model, along with the features of GST representations in MocGraph, has enabled us to support simulations in the DIF package that can be used for verification and analysis of static and dynamic dataflow models. This generalized framework coupled with efficient scheduling strategies have facilitated rapid prototyping of heterogeneous dataflow applications. Our recently developed scheduling techniques for heterogeneous dataflow graphs exploit statically known mode transition patterns in dynamic dataflow actors. In this poster, we present features of GSTs, the CFDF model, The DIF Language (TDL), and our recent research in developing generalized scheduling techniques for CFDF representations in TDL.

Electronic downloads

Citation formats  
  • HTML
    Nimish Sane. <a
    href="http://chess.eecs.berkeley.edu/pubs/567.html"
    ><i>A generalized framework for modeling and
    scheduling heterogeneous dataflow
    applications</i></a>, Talk or presentation,  16,
    April, 2009; Poster presented at the 8th Biennial Ptolemy
    Miniconference.
  • Plain text
    Nimish Sane. "A generalized framework for modeling and
    scheduling heterogeneous dataflow applications". Talk
    or presentation,  16, April, 2009; Poster presented at the
    8th Biennial Ptolemy Miniconference.
  • BibTeX
    @presentation{Sane09_GeneralizedFrameworkForModelingSchedulingHeterogeneous,
        author = {Nimish Sane},
        title = {A generalized framework for modeling and
                  scheduling heterogeneous dataflow applications},
        day = {16},
        month = {April},
        year = {2009},
        note = {Poster presented at the 8th Biennial Ptolemy
                  Miniconference.},
        abstract = {MocGraph is a Java-based package of generic graph
                  data structures and algorithms with emphasis on
                  supporting graph-theoretic analysis for models of
                  computation (mocs). MocGraph has evolved from the
                  the graph package in Ptolemy II. In our recent
                  work on the MocGraph package, we have added
                  support for trees and generalized schedule trees
                  (GSTs), in particular. A GST is a data structure
                  for representing schedules of dataflow graphs. It
                  is a general schedule representation in that it
                  can be used to represent dataflow schedules for a
                  wide variety of dataflow models of computation. As
                  part of the DIF (dataflow interchange format)
                  project at the University of Maryland, we have
                  also developed the core functional dataflow (CFDF)
                  model of computation. CFDF facilitates natural
                  deion of actors and provides a general framework
                  for specifying static and dynamic dataflow
                  applications. We have demonstrated that a variety
                  of commonly used, existing dataflow models can be
                  represented naturally in CFDF. The CFDF model,
                  along with the features of GST representations in
                  MocGraph, has enabled us to support simulations in
                  the DIF package that can be used for verification
                  and analysis of static and dynamic dataflow
                  models. This generalized framework coupled with
                  efficient scheduling strategies have facilitated
                  rapid prototyping of heterogeneous dataflow
                  applications. Our recently developed scheduling
                  techniques for heterogeneous dataflow graphs
                  exploit statically known mode transition patterns
                  in dynamic dataflow actors. In this poster, we
                  present features of GSTs, the CFDF model, The DIF
                  Language (TDL), and our recent research in
                  developing generalized scheduling techniques for
                  CFDF representations in TDL.},
        URL = {http://chess.eecs.berkeley.edu/pubs/567.html}
    }
    

Posted by Christopher Brooks on 17 Apr 2009.
Groups: ptolemy
For additional information, see the Publications FAQ or contact webmaster at chess eecs berkeley edu.

Notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright.

©2002-2018 Chess