Tutorials Scheduling for the T3E Jay Blakeborough A presentation of the current state of scheduling for the T3E. A discussion of the solution of current solved and unsolved problems. Questions arising from this tutorial will be treated in a Q&A session later in the meeting. System Tuning for the Origin 2000 Edward Hayes-Hall A focus on the systune parameters. The tutorial will present a list of the most often used parameters (15–20 or so) with descriptions of what they are and how they interact with other systune parameters. This will be a technical presentation for people who know something about systune and system resources. Data Intensive Computing William Kramer Data intensive computing is increasingly important in scientific research, whether for experimental data analysis and visualization, or intensive simulation such as climate modeling. By integrating high performance computation, very large data storage, high bandwidth access, and both local and wide area high-speed networking, data intensive computing stretches the technical resources of most architectures. The synergy of networking and computational initiatives should produce major improvements in data intensive computing technology in the next several years. Preparing for this paradigm shift is the prime motivation for this tutorial. The tutorial provides an overview of these issues followed by case studies that explore in detail different aspects of research and systems solutions for data intensive computing. I/O Tuning Randy Kreiser How to optimize disk I/O for both SCSI and Fibre Channel attached disks and XFS File Systems. To include Optimal File System Block size determination, whether to stripe and what the stripe size should be, systune resources that affect disk I/O performance, performance aspects of internal vs. external Journal files, and how to determine the optimal allocation group count. Some information about specific computers and OS systems will be presented. OpenMP Tom MacDonald OpenMP is a portable shared memory programming model with bindings for Fortran, C and C++. This tutorial will present the major constructs and features in OpenMP that enable programmers to write coarse grain, scalable, shared memory parallel programs. The tutorial is aimed at users interested in writing parallel applications or parallelizing existing applications in Fortran, C and C++. OpenGL Volumizer and Large Data Visualization Chikai Ohazama This tutorial will present the basic structure of OpenGL Volumizer. Details of its use for scientific visualization, and the advantages versus the disadvantages for using volume rendering, will be discussed. The mechanics behind Monster Volumizer and its use will also be addressed. Performance Tools on the Origin Alex Poulos As applications grow larger, compiler optimizations more aggressive, processor designs more sophisticated and system architectures more scalable, performance analysis is getting more and more complex. This tutorial will provide a presentation of SGI’s performance tools that offer practical solutions helping the users understand the behavior of their codes and tune their applications on Origin systems. The tutorial is targeted toward all audiences. A Q&A session will follow. Table of Contents | Author Index | CUG Home Page | Home (Title Page)