Introduction

  When to Use parfor

  Setting up MATLAB Resources Using matlabpool

  Creating a parfor-Loop

  Differences Between for-Loops and parfor-Loops

  Reduction Assignments

  MATLAB Path

  Error Handling

  Limitations

  Performance Considerations

  Compatibility with Earlier Versions of MATLAB Software

  About Programming Notes

  Classification of Variables

  Improving Performance

  Introduction

  When to Use spmd

  Setting Up MATLAB Resources Using matlabpool

  Defining an spmd Statement

  Introduction

  Creating Composites in spmd Statements

  Variable Persistence and Sequences of spmd

  Creating Composites Outside spmd Statements

  Distributed Versus Codistributed Arrays

  Creating Distributed Arrays

  Creating Codistributed Arrays

  MATLAB Path

  Error Handling

  Limitations

  Using Graphics in pmode

  Connectivity Testing

  Hostname Resolution

  Socket Connections

  Introduction

  Nondistributed Arrays

  Codistributed Arrays

  How MATLAB Software Distributes Arrays

  Creating a Codistributed Array

  Local Arrays

  Obtaining Information About the Array

  Changing the Dimension of Distribution

  Restoring the Full Array

  Indexing into a Codistributed Array

  2-Dimensional Distribution

  Parallelizing a for-Loop

  Codistributed Arrays in a for-drange Loop

  Overview

  Toolbox and Server Components

  Example: Evaluating a Basic Function

  Example: Programming a Basic Job with a Local Scheduler

  Getting Help

  Defining Configurations

  Exporting and Importing Configurations

  Validating Configurations

  Applying Configurations in Client Code

  Saving or Sending Objects

  Current Working Directory of a MATLAB Worker

  Using clear functions

  Running Tasks That Call Simulink Software

  Using the pause Function

  Transmitting Large Amounts of Data

  Interrupting a Job

  Speeding Up a Job

  Introduction

  Collecting Parallel Profile Data

  Viewing Parallel Profile Data

  Object Data Size Limitations

  File Access and Permissions

  No Results or Failed Job

  Connection Problems Between the Client and Job Manager

  Scope of dfeval

  Arguments of dfeval

  Example - Using dfeval

  Creating and Running Jobs with a Local Scheduler

  Local Scheduler Behavior

  Creating and Running Jobs with a Job Manager

  Sharing Code

  Managing Objects in the Job Manager

  Creating and Running Jobs

  Sharing Code

  Managing Objects

  Overview

  MATLAB Client Submit Function

  Example - Writing the Submit Function

  MATLAB Worker Decode Function

  Example - Writing the Decode Function

  Example - Programming and Running a Job in the Client

  Supplied Submit and Decode Functions

  Managing Jobs

  Summary

  Schedulers and Conditions

  Coding the Task Function

  Coding in the Client

  Introduction

  Coding in the Client

  Number of Tasks in a Parallel Job

  Avoiding Deadlock and Other Dependency Errors