IBM said it has created a $US29 million supercomputing research initiative, and marked its debut by announcing that it will release the source code for software that helps users create visual representations of data.
The Deep Computing Institute is designed to bring together academics and representatives of industry who will jointly work on real-world business and scientific issues that can be addressed through supercomputing, said William Pulleyblank, director of Mathematical Sciences at IBM Research, who also is now directing the new institute.
"We really want this to be something that we do for the global scientific community," Pulleyblank said.
The release of the source code for Visualisation Data Explorer is intended to be the first in an ongoing effort to open discussion and development. Data Explorer is used for visualisation applications in areas such as computational chemistry, engineering analysis and medical imagery.
The software is used in diverse applications - to visualise weather patterns in an effort to improve forecasting accuracy, provide oil-flow simulations to boost drilling success and to put together more than 5,000 ceiling fragments from a 1,200 year old Peruvian temple.
Developers create different software modules for Data Explorer, depending on the data being used, so there are various ways to scale and display data, Pulleyblank said.
With ever more digital information readily accessible on the Internet, more can be done technologically than ever before, he said, adding that weather forecasting is a prime example. Satellites now provide weather data leading to more precise forecasts.
Weather data can be provided to electrical utility companies which can then make decisions about likely demand and need. The airline industry can take advantage of such knowledge, as well.
"If you know in advance that thunderheads are going to go through an area from 2:30 to 3:00, that's a good time to plan a way around them," Pulleyblank said.
Agricultural interests also benefit from such advance warning, he said, noting that some possible applications aren't as readily evident. One that recently came to IBM's attention that Pulleyblank said amused company researchers was from a major auto-racing team that wanted to know track conditions on race day.
Part of the objective in releasing software source code and making certain programs available to the open-source community is for developers to work on applications that might not have been previously considered, he said.
The software runs on a range of machines, including Unix-based supercomputers and workstations and PCs or servers using Windows or Linux operating systems. The source code can be found at http://www.research.ibm.com/dci/software.html starting today.