Parallel Programming with MPI
Presented by Masao Fujinaga - University of Alberta
Wednesday, January 9, 2008
This presentation provides an introduction to the basics of programming with Message Passing Interface, using examples in Fortran and in C.
Parallel Programming with OpenMP
Presented by Edmund Sumbar - University of Alberta
Wednesday, January 23, 2008
This presentation is targeted at researchers who would like an introduction to parallel programming on shared-memory machines.
Introduction to Grid Technologies
Presented by Cameron Kiddle - University of Calgary
Wednesday, February 6, 2008
The aim of this seminar is to examine existing grid technologies and to learn how researchers might benefit from the use of these technologies.
Advanced Topics: MPI Programming Models
Presented by Roman Baranowski - UBC
Wednesday, February 20, 2008
The main goal of parallel programming is to utilize all processors and minimize the elapsed time of the user program. Message Passing Interface allows users to realize various parallel models such as SPMD (Single program multiple data) and MPMD (Multiple programs multiple data). The focus of this presentation will be on various programming techniques using MPI in designing a parallel program or parallelizing existing serial code. The talk is addressed to all researchers interested in parallel computations in distributed memory environments.
To view this presentation's slides click here.
Interoperable Mesh Tools for PDE Simulations
Presented by Carl Ollivier-Gooch - UBC
Wednesday, March 5, 2008
All PDE-based numerical simulation software must be able to access and manipulate mesh data, and an increasing number of simulations require the ability to adapt the mesh based on solution data. However, such sophisticated mesh manipulation operations are quite difficult to implement well. The obvious solution to this --- mesh adaptation libraries --- does not work well in practice because these libraries require specific data structures.
This talk will describe a mesh component architecture that overcomes this problem. The mesh component has a defined data model --- describing what data must be represented, but not how --- and a defined interface for accessing and manipulating that data. Applications and services such as adaptation interact with mesh data through the interface. A key advantage of the component architecture is that services can be used with any mesh database that supports the interface, including data native to a particular application with the interface functionality added on top of it.
The talk will include a description of the mesh component data model and interface, as well as several usage examples, from simple to relatively sophisticated.
To view this presentation's slides click here .