Videos

Distributed Computing (MPI) by C. Garrett
Parallel Computing Research Internship (PCSRI) 
LA-UR-16-26329

Abstract – The dominant programming language for distributed computing, MPI, has become synonymous with high performance computing. The basics of MPI and how it enables parallelism across distributed memory architectures is presented.

Parallel Algorithms and Patterns by R. Robey
Parallel Computing Research Internship (PCSRI)
LA-UR-16-26383

Abstract – At the heart of large high performance computing applications are the parallel algorithms and the parallel coding patterns that deliver scalable performance. What makes a good parallel algorithm? This talk presents recent research in new parallel algorithms and a few of the important scalable parallel patterns.

Intro to Benchmarks and Proxies 
by R. Robey, H. Nam, G. Rockefeller, C. Garrett, B. Krueger, J. Schoonover
Parallel Computing Research Internship (PCSRI)
LA-UR-16-26440

Abstract – This presentation gives an overview of the benchmarks commonly used for measuring the performance of High Performance Computing (HPC) systems. The new DOE proxy applications and benchmarks used in procurements are also described. In addition, two new open-source applications, SELF, and Tycho 2 are detailed as well as the LANL Higrad suite of codes which is to be used as a relatively large application code example. All of these are potential codes for use in the parallel computing workshop.  

Linear Solvers Overview by C. Garrett
Parallel Computing Research Internship (PCSRI)
LA-UR-16-27180

Abstract – Solving problems involving linear systems is ubiquitous in scientific computing. The basics of numerical linear algebra are presented for both dense and sparse linear systems. Libraries that implement linear solvers are also presented.

Software Development Practices by B. Krueger
Parallel Computing Research Internship (PCSRI)
LA-UR-16-26375

Abstract – Good software development practices are essential on large scientific code projects. The value of documentation, version control, tests, verification and validation are all presented. The cost of hurried development through the accumulation of “technical debt” is weighed. This leads to a discussion of what practices should be used in student’s summer projects.

HIGRAD Science by Rachel Robey
Parallel Computing Research Internship (PCSRI)
LA-UR-16-26621

Abstract – The Higrad suite of codes couples a massively parallel atmospheric model with a variety of phenomena to create useful simulation tools. This includes Firetec, a wildfire simulation code, Windblade, a wind turbine modeling application, as well as a vegetation and climate change model, hurricane model and other disaster modeling. Examples of these simulations are shown and how they have helped understand key aspects of the underlying phenomena.