GPU08
From NA-Wiki
Contents |
High Performance computing using GPU
A summer course giving in 2008 focusing on high performance computing on streaming architectures.
Course administrators are: Dag Lindbo, Henrik Holst and Tomas Oppelstrup.
Course examinator: Johan Hoffman.
We will present news here related to the course. We will also have a mailing list here (not yet available).
News
The first meeting is on Tuesday 2/6 at 10 AM in room 304 (PDC).
Goals
- Introduction to streaming, massively concurrent architecture (stream processing)
- Formulating suitable data parallel algorithms
- Use of CUDA SDK.
- After the course the students will be able to implement computational kernels on the CUDA architecture.
Plan
There will be four lectures during June.
- Introduction to GPU computing (Dr. Erik Lindahl)
- Introduction to CUDA SDK (Dag), CUBLAS (Tomas) and Fortran interoperability (Henrik)
- Data structures and Advanced examples. Prof. Jesper Oppelstrup will give retrospect on massively concurrent architectures from the past.
- Presentation of student projects
Example student problems
- Monte Carlo for option pricing.
- Iterative method for linear system of equations - Either standalone or as a kernel of a bigger code.
- Explicit finite difference methods - Choose you favourite PDE or ODE and solve it.
- Image processing. Implement some filtering methods of your choice.
- ...Your own ideas!
Links
- CUDA, Supercomputing for the Masses (ddj.com)
- CUDA Programming guide v 1.1 (nvidia.com)
- Free E-books: GPU gems 1 (nvidia.com) and GPU gems 2 (nvidia.com). We will try to provide copies of relevant chapters not available online (GPU Gems vol 2-3).
- Additional links:
- CUDA Zone (nvidia.com)
- General GPGPU research (gpgpu.org) - Not CUDA specific.
Acknowledgments
We thank the University of Houston and Texas Learning & Computation Center (TLC2), KTH Computational Science and Engineering Centre (KCSE) and Prof. Johan Hoffman for suppling the necessary hardware.