CBC Talk on A Case Study of GPUs in Scientific Computing: Low-Order FEM - September 3, 2010
Matt Knepley from University of Chicago will share a case study of GPUs in Scientific Computing, Friday September 3, at 11 o'clock in Bakrommet.
Total number of participants: 16
Total number of guests outside of CBC: 2
Number of different
nationalities represented: 5
Total number of speakers: 1
Total
number of talks: 1
Abstract:
We present a model for scientific applications based upon a Python framework, in which linear algebra and solver work is handled by libraries, and integration, or physics, is handled by massively parallel accelerators, in this case a GPU, through the PyCUDA framework. We illustrate this paradigm by discussing the step-by-step development of a high performance, portable engine for low-order FEM integration.
About the speaker:
Matt Knepley is a well known guest at CBC, and currently works at the University of Chicago.
His current research interests includes:
- Software for computational science
- Modeling of scientific computation
- Fast summation methods
- Radial basis function methods
- FEM and BEM, accelerated by fast solver
- Automated code generation for FEM and BEM
- Vortex particle methods
- Earthquake modeling
- Fracture mechanics
- Algorithms for massively multicore hardware, such as NVIDIA GPUs
For more information about Matt, please visit his website at people.cs.uchicago.edu/~knepley/
| What |
|
|---|---|
| When |
Sep 03, 2010 from 11:00 to 12:05 |
| Where | Bakrommet @ Simula |
| Contact Name | Hans Petter Langtangen |
| Attendees |
Anders Helgeland Anders Logg Andre Massing Benjamin Kehlet Hans Petter Langtangen Johannes Ring Karen Støverud Kent-André Mardal Kristian Valen-Sendstad Lyudmyla Vynntska Margarete Jadamec Marie rognes Mikael Mortensen Ola Skavhaug Stuart Clark Svein Linge |
| Add event to calendar |
vCal
iCal
|
