Cornell Theory Centre - Day One

Yesterday was day one at this brief training course I'm fortunate enough to be attending at the Cornell Theory Centre (CTC). CTC is a high-performance computing and interdisciplinary research center located on the Ithaca campus of Cornell University, which is buried in beautiful snow... people actually live like this???

The coolest part for me of the first day was a tour of CTC's CAVE. A CAVE provides a three-dimensional, stereo immersive virtual reality environment for viewing scientific, engineering, architectural and art applications. In the demo we had, we donned 3D glasses and walked inside a model of a protein molecule, dove down into topographical maps, flew through weather pattern maps, and, coolest of all, traversed a Quake map - in 3D! People, I have seen the future of gaming! I can't wait to play Halo 3D!

The next most exciting event of the day was actually touching the 68th fastest supercomputer in the world! And it is built on top of Windows! It's called “Velocity 2“ and its a Dell/Windows cluster - the Dell PowerEdge 2650 cluster runs Microsoft Windows Advanced Server and consists of 320 dual-processor servers with 2.4 GHz Intel® Xeon™ processors, 2 GB Ram/Node, and 72 GB Disk/Node (RAID 0).

CTC currently have 18 Dell/Windows clusters with over 1500 processors and their CTO explained the ROI they have experienced by moving 100% of their supercomputing capability from IBM over to Windows during the last few years. CTC claim their benchmarking has proven that the performance on Windows is superior and/or equal to IBM SP or Linux. And the ROI they have experienced since moving from IBM SP to Windows is amazing. Huge reductions in staff, millions saved in maintenance, etc.

They key message however was that the researchers at Cornell can now focus on their own scientific discipline and not have to spend 50% of their time also being computational scientists. All of the tools they use on their desktop can now run on the supercomputer, allowing them to stop writing code and to be able to test their models faster as well as cheaper.

We also looked at a demo of using Excel as a front-end to do risk analysis on a portfolio of callable bonds. Excel invokes a web service which runs on one of CTC's clusters, allowing a hedge fund to calculate its Value-at-Risk, etc.

With the price point of High-Performance Computing dropping so quickly, it's going to be amazing to witness what new applications it massive crunch will be applied to over the next decade.

Quote from CTC's CTO, David Lifka: “What would it take to build a Windows-based number one system? Money, power, cooling and space. There is no reason why this platform won't scale.“