Exploiting Available Resources

The goal of this project is to explore language and programming paradigms that allow exploitation of available resources on modern desktops and networks.

Parallelism has long been used in scientific computing for so called "grand challenge" problems where it produces enormous benefit. Unfortunately, parallel programming is not easy and successes in scientific computing have come at great cost. Today, parallelism is set to become main stream. For years the software industry has relied on improvements in processor technology to achieve Moore's law which predicts perpetual performance gains. Given the thermal limitation of clock rates on modern processors, for Moore's law to continue we are likely to have to rely on parallelism in the form of multi-core processors.

For parallelism to become main stream it needs to become much easier to exploit. This project is not about achieving higher levels of performance on high-end supercomputers and clusters. It's about bringing parallel computing within reach of everyday programmers, and specifically in the context of e-Research, to smaller research groups that can't employ specialist research programmers to parallelize their applications. From a hardware point of view, parallelism in the form of multi-processor cores, sophisticated graphics processors capable of vector instructions and small personal clusters of PCs are already within reach of even the smallest e-Research project.

Our four specific project areas are:

• Reasoning About Inherent Parallelism
• Measuring Inherent Parallelism
• Simplifying Parallel Programming
• Exploiting Parallel Hardware