Note: You are experiencing only the raw content of this site, without the intended layout and design. Either your browser has ignored the Cascading Style Sheet (CSS) files for this site, or you are using an outdated browser which does not support Web Standards. Learn more.

EDS' Next Big Thing Blog: Read and Respond to What the EDS Fellows Say About Technology

Read and respond to what the EDS Fellows have to say about the future of technology on EDS' Next Big Thing Blog on eds.com.

We Still Need Supercomputers

by Randy Mears

Compute platforms can be described using several major categories. Viewed from the IT business, we end up with mainframes, supercomputers, minicomputers and microcomputers (further broken down into servers, PCs, handhelds, tablets, etc.). These days the term minicomputer has all but disappeared while mainframes hang on and microcomputers continue to increase their parallelism and ubiquity.

Supercomputers remain the most arcane compute platform. In the days when we were impressed by terms like mega flops and giga flops we could always look to the supercomputer for the limits of the possible. Nowadays we are unimpressed by such measures and look more at what a supercomputer platform can do that other platforms can’t (or shouldn't). To that end, the supercomputer dominates extreme simulations and, unapologetically, chess.

The most popular extreme simulation niche continues to be weather forecasting. As concerns about global warming increase so will our need to simulate weather along with the changes to the environment and geography that could result. Plans for new and improved long-term weather forecasting, along with the side effects, will help keep the term supercomputer alive and well for many years to come.

Published Thursday, July 20, 2006 9:36 PM

Subscribe to this post's comments using RSS

Comments

Posted by Rod Schiffman Friday, July 21, 2006 4:01 AM

I worked on building parallel Super Computers back in the day when it was new. We put together the first machine to do a GigaFlop on standard problems like FFT's, Linpack, Convolutions, etc.

Anyway, it provided a big insight into the work being done on these first machines that broke the barriers of the Cray computers that owned the market at the time. While the grail of weather simulation was talked about, the reality is that there are many problems being worked on using super-computers you just don't hear about.

We worked with Ken Wilson who used a primative super-computer to simulate his atomic theory that won him a nobel prize. There is a lot of work done simulating atomic partical interactions. We also worked with a person doing work for the military enhancing satellite photographs that was classified. When the machine was introduced commercially there was a special kick-off event. Three different groups from NSA came. Of course, they didn't even admit they knew what a computer was, but they were all very interesting in advancing the state of the art in super computers. Groups were also there from Lawrence Livermore, Jet Propulsion Labs and Los Alamos. I can guarantee you they were not there to purchase machines to simulate weather or chess.

The world of massively parallel super-computers is like an iceberg. The stuff you hear about and know about is only the 10% that the public hears about. There's 90% decidated to weapons, national security (code breaking, etc.) and theoritical reseatch types of projects.

Not to denigrate anthing in the blog, but these machines, and the problems being worked on, are a lot bigger than weather simulations.