Class MultiThreadTCSolver

java.lang.Object

solvers.queueingnet.treeAlgorithms.TreeTraverserMulti

solvers.queueingnet.treeAlgorithms.treeConvolution.MultiThreadTCSolver

All Implemented Interfaces:

IThreadedTCSolver

public class MultiThreadTCSolver
extends TreeTraverserMulti
implements IThreadedTCSolver

Tree Convolution solver that uses multiple threads.

Author:

Ben Homer, 2014.

Field Summary

Fields inherited from class TreeTraverserMulti

taskExecutor

Fields inherited from interface IThreadedTCSolver

normalisationConstantTimer, queueLengthTimer, throughputTimer

Constructor Summary

Constructors

Constructor	Description
MultiThreadTCSolver(QNModel qnm, ConvolutionLayer cl, TreePlanter tp, int numThreads)	Constructor.

Method Summary

Modifier and Type	Method	Description
BigRational[][]	calculateMeanQueueLengths(PopulationVector p)	Calculates per class and per station mean queue lengths using methods outlined in LamLien83 paper.
BigRational[]	calculateThroughputs(PopulationVector p)	Calculates per class throughputs using methods outlined in LamLien83 paper.
BigRational	computeG(Node node, PopulationVector pv)	Computes G using postorder tree traversal.
Thread	getComputeLeafNodeThread(Node node, PopulationVector pv)	Computes results at leaf node by creating an appropriate thread to run the computation.
Thread	getSubnetNodeThread(Node node, PopulationVector pv)	Computes results at subnetwork node by creating an appropriate thread to run the computation.

Methods inherited from class TreeTraverserMulti

traverse

Methods inherited from class Object

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

Constructor Details

MultiThreadTCSolver

public MultiThreadTCSolver(QNModel qnm,
 ConvolutionLayer cl,
 TreePlanter tp,
 int numThreads)

Constructor.

Parameters:

qnm - The queueing network model to solve.

cl - The convolution layer to delegate to.

tp - The tree planter.

Method Details

computeG

public BigRational computeG(Node node,
 PopulationVector pv)

Computes G using postorder tree traversal.

Specified by:

computeG in interface IThreadedTCSolver

Parameters:

node - The current node to compute the normalisation constant G for.

pv - The population vector to be used during the computation.

Returns:

The normalisation constant G at the given node, calculated using the specified population vector.

calculateThroughputs

public BigRational[] calculateThroughputs(PopulationVector p)

Calculates per class throughputs using methods outlined in LamLien83 paper.

Specified by:

calculateThroughputs in interface IThreadedTCSolver

Parameters:

p - The population vector to use during the calculation.

Returns:

The per class throughputs as an array of BigRationals.

calculateMeanQueueLengths

public BigRational[][] calculateMeanQueueLengths(PopulationVector p)

Calculates per class and per station mean queue lengths using methods outlined in LamLien83 paper.

Specified by:

calculateMeanQueueLengths in interface IThreadedTCSolver

Parameters:

p - The population vector to use during the calculation.

Returns:

The per class per station mean queue lengths as a 2D array of BigRationals.

getComputeLeafNodeThread

public Thread getComputeLeafNodeThread(Node node,
 PopulationVector pv)

Computes results at leaf node by creating an appropriate thread to run the computation.

Specified by:

getComputeLeafNodeThread in class TreeTraverserMulti

Parameters:

node - The leaf node.

pv - The population vector to use for the computation.

Returns:

The thread which will run the computation.

getSubnetNodeThread

public Thread getSubnetNodeThread(Node node,
 PopulationVector pv)

Computes results at subnetwork node by creating an appropriate thread to run the computation.

Specified by:

getSubnetNodeThread in class TreeTraverserMulti

Parameters:

node - The subnetwork node.

pv - The population vector to use for the computation.

Returns:

The thread which will run the computation.