|
|
Is Grid a solution for researchers? A number of successful applications in the areas of Bioinformatics, Financial Simulation and data-mining at NUS in recent years have convinced us that Grid is indeed a cutting edge solution for research computation. In 2005, the NUS Grid (also known as Tera-scale Campus Grid or TCG@NUS) was created using the cycle harvesting technology to harness idle compute cycles for large scale research computation. Today the NUS Grid consists of 1400 PC nodes and 150 multi-core server nodes providing more than 1800 processor cores. During the first two years of implementation, researchers used the NUS Grid to run large-scale bioinformatics applications such as BLAST, AUTODOCK, Modeller and HMMER, reducing computation time from days and weeks to hours. Subsequently, more applications in financial engineering and data-mining have also benefitted from the NUS Grid. Examples such as running Monte Carlo-based simulation from 50 days to 1-day on NUS Grid were experienced by Dr Ding of NUS Business School. Patsnap, a start-up company incubated by NUS Enterprise, successfully took <10 days to run 1.6 terabytes of patent data which would have taken a standard computer an estimated 2 years run time. How then is NUS Grid similar to or different from the well known community Grids today (SETI@HOME, FOLDING@HOME or World Community Grid)? First let’s look at the similarity. All these Grids adopt cycle harvesting technologies to harness unused compute cycles and channel them to perform useful tasks. Applications that run on them use the same divide-and-conquer methodology by chopping a large computational task into many small tasks that can be performed in parallel on the Grid nodes. Size-wise, NUS Grid is relatively small compared to the community Grids. However, NUS Grid has the advantage in terms of the interconnection performance among its nodes. Our faster campus network, with at least 10 times more bandwidth than home PC connections used in the community Grids, allows data intensive applications to be dispatched faster on the NUS Grid. Another significant advantage is our inter-connected large memory sever nodes in the pool of compute resources with up to 8GB of memory, thus enabling the execution of larger memory applications and widen the applications supported. As we anticipate the multi-core server nodes to grow in the coming years, and by adding such computing capability on the NUS Grid, it will truly offer greater competitive edge to our research community in enabling new discoveries. Please feel free to contact us at if you are keen to explore NUS Grid for your research applications. |
|
