Slides (Keynote)
|
Slides (Keynote)
|
| Abstract: The Ising model of ferromagnetism has an special role in statistical physics: it is remarkably simple and contains an interesting phase transition. Like real ferromagnets, the Ising model loses its ferromagnetism above the critical (Curie) temperature. Although the equilibrium properties of the Ising model are essentially understood, much remains unknown about the model's surprisingly rich dynamical behavior. In this talk I will consider the ferromagnetic ordering process following a quench from above to below the critical temperature. Two dimensional simulations have demonstrated that the Ising system evolves to a long lived metastable state, consisting of striped magnetic domains, with a probability of roughly 1/3. I will employ a surprising connection with critical percolation to establish the universality of this trapping probability and its precise value of 0.3558.... for square boundary conditions. |
Slides (Keynote)
|
Slides (PDF)
|
Abstract: In this talk I will discuss advanced tricks to maximize CUDA performance, taking examples from my physics research. The topics to be covered include: how to maximize device bandwidth, the (unofficial) CUDA disassembler decuda, and a discussion of the optimizations performed by the CUDA compiler. I will also mention various gotcha's, pitfalls, tips, ands tricks that I have encountered. An interactive discussion format is encouraged. |
Slides (PDF)
|
Abstract: For many scientific tasks, modern graphics processors (GPUs) are 10 to 100 times faster than ordinary workstation processors (CPUs). Yet a high end graphics card sells for less than $500. How is this exceptional price/performance achieved? The key is massive parallelism: the GPU is designed to execute thousands of threads simultaneously. In this talk I will discuss how GPUs can be applied to parallelizable computational problems, taking examples from my physics research. I will give a brief introduction to CUDA, a C-like language used to program the GPU. |