mapss.dif.csdf.sdf.sched
Class ABMLBDPPOStrategy

java.lang.Object
  mapss.dif.DIFScheduleStrategy
      mapss.dif.csdf.sdf.sched.DPPOStrategy
          mapss.dif.csdf.sdf.sched.ABMLBDPPOStrategy

All Implemented Interfaces:

mocgraph.analysis.analyzer.Analyzer, mocgraph.analysis.analyzer.GraphAnalyzer, mocgraph.sched.ScheduleAnalyzer

public class ABMLBDPPOStrategy
extends DPPOStrategy

ABMLB (Absolute Buffer Memory Lower Bound) DPPO scheduler for SDF graphs. This DPPO scheduler computes an R-schedule with the criterion of ABMLB. The result R-schedule provides a good choice for ProcedureStrategy.

Version:

$Id: ABMLBDPPOStrategy.java 406 2007-05-10 14:27:07Z plishker $

Author:

Mingyung Ko

Field Summary

Fields inherited from class mapss.dif.csdf.sdf.sched.DPPOStrategy

_lexicalOrder, _tableDimension

Fields inherited from class mapss.dif.DIFScheduleStrategy

_clusterManager

Constructor Summary

ABMLBDPPOStrategy(SDFGraph graph, java.util.List lexicalOrder)
Constructor for a given graph and a lexical order.

Method Summary

protected void

_optimumFor(int i, int j) The optimal cost is defined as : prod + cons - gcd(prod, cons) of two SDF clusters.

Methods inherited from class mapss.dif.csdf.sdf.sched.DPPOStrategy

_bufferCost, _computeDPPO, _computeSchedule, _crossingSDFEdges, _DPPOTableElement, _SDFEdges, _SDFNodes, _verifyIndices, _verifyIndices, optimalCost, schedule, setLexicalOrder, toString

Methods inherited from class mapss.dif.DIFScheduleStrategy

getClusterManager, graph, valid

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

ABMLBDPPOStrategy

public ABMLBDPPOStrategy(SDFGraph graph,
                         java.util.List lexicalOrder)

Constructor for a given graph and a lexical order.

Parameters:

graph - The given SDF graph.

lexicalOrder - The lexical order in the form of List.

Method Detail

_optimumFor

protected void _optimumFor(int i,
                           int j)

The optimal cost is defined as : prod + cons - gcd(prod, cons) of two SDF clusters.

Specified by:

_optimumFor in class DPPOStrategy

Parameters:

i - The starting index in the lexical order.

j - The ending index in the lexical order.