There are great challenges in the development of scientific computation: data sets are growing exponentially, models are becoming increasingly complex, while many of the latest machines are getting more costly and harder to program.
This Summer School will provide an introduction to a new dataflow approach for accelerating computation, which has recently been proven commercially. This approach involves compiling dataflow programs into special-purpose compute engines targeting advanced reconfigurable hardware. It has been used in many scientific computations, including those in geophysics, biophysics, and quantum chemistry. Promising results are reported. The topics addressed include:
- Monte Carlo simulations
- Numerical solutions of ordinary and partial differential equations
- Numerical integration methods
- Finite difference and finite element methods
- Ab initio quantum chemistry
The Summer School will consist of talks given by leading experts from academia and from industry. There will be practicals to enable attendants to have hands-on experience of developing solutions using this approach. Attendants are invited to bring some of their problems, which can be used as case studies to illustrate the development and optimisation of designs. A preliminary schedule of the Summer School is available.
See the Summer School website for more information.