MGMS Meeting on Distributed and Grid Computing for the Life Sciences

Abstracts

Maya Natarajan, Entropia Inc.

There is a lot of hype about Grid Computing, the notion of linking computers together, to create a single aggregate computational resource. The largest computational resource within an enterprise is provided by Windows PCs. The concept of distributed computing on desktop machines is inherently simple; but there are many factors essential to ensuring that desktop resources are aggregated into an easily manageable and usable resource (PC grid) provided that there is little or no impact on the PC host. These factors include: a) a platform for fast and secure application enabling; b) the range of applications that can be distributed; c) thinking beyond the "embarrassing" parallel applications; and d) a secure, multidimensional environment where both the desktop user and application scientist use the resource harmoniously. Each of these factors and its impact on PC grid computing will be discussed using a case study.

The Grid and How it is Helping Science

David Boyd, Deputy Director CLRC e-Science Centre

This presentation aims to describe in simple terms what the Grid is, how it works and some of the technology that it uses. This will be done in the context of science projects that are currently underway in the UK based on the Grid.

Application distribution and acceleration in life sciences

Richard Groves, Blackstone Computing

Performance of life science applications on Unix clusters and computes farms may be maximized through the implementation of one or more techniques. These techniques may include optimization of the computational resources available, implementation of embarrassingly parallel techniques, utilization of message passing technology and rewriting the application so that it can run in a truly parallel (SMP) environment. The actual techniques employed will depend upon many factors including the availability of source code, the data requirement of the application, the volume of work and the computational environment at hand. This talk will cover the rationale and methodology taken to create a flexible distributed computing environment and will use examples from computational chemistry to demonstrate the performance enhancements that can be achieved in such an environment.