»	Details below

 
 
 

Who Can Benefit

Students who can benefit from this course are:

• Experienced C, Fortran, or MPI programmers developing applications under the Sun HPC ClusterTools 4 computing environment.
• Application programmers responsible for porting existing applications to the Sun HPC ClusterTools 4 computing environment.

 

Prerequisites

To succeed fully in this course, students should be able to:

• Use basic Solaris Operating Environment commands (create, copy, and rename files and compile, run, and monitor jobs)
• Create and edit text files using a UNIX text editor
• Write any of F77, F90, or C programs proficiently
• Understand or write an MPI program (or understand the basics of message-passing distributed-memory programming)
• Understand the basic concepts of UNIX processes and POSIX-threaded processes

 

Skills Gained

Upon completion of this course, students should be able to:

• Compile and execute a program using Sun Performance Workshop with basic serial optimization
• Use the Sun Performance Workshop to identify performance bottlenecks
• Know why, when and how to parallelize and run programs in Sun's SMP or clustered environment
• Successfully use the Sun HPC ClusterTools 4 environment to run MPI and other parallel programs in Sun's clustered SMP environment

 

Related Courses

Before:

• High Performance Computing Administration (ES-320)

 

Course Content

Module 1 - Program Optimization and Sun HPC Product Overview

• HPC purpose and features
• HPC software and hardware components
• Application development and optimization methodology

Module 2 - Serial Optimization

• Optimization of single-process, single-thread applications
• Characterization of application run-time behavior using tools and utilities
• Compiler options to enhance application performance
• Porting issues between 64-bit and 32-bit environments
• Using the Sun Performance Library

Module 3 - Program Parallelization

• Parallel programming overview
• Workshop compiler automatic parallelization
• Explicit parallelization with compiler directives
• Explicit parallelization using processes and threads
• Tools and techniques to guide parallelization
• Recoding for improved parallelism

Module 4 - Optimizing MPI Applications

• Overview of MPI and message passing programming
• Using the Cluster Runtime Environment (CRE) to launch and monitor MPI programs
• Tools and techniques to analyze MPI application run-time behavior
• Optimization techniques for MPI applications
• Using the Sun(TM) Scalable Scientific Subroutine Library (Sun S3L)

 
Copyright 2004-2009 Sun Microsystems, Inc.