PC World: Hewlett-Packard is releasing to the open source community the computer code for a software programming interface that helps manage large data sets in high performance computing environments.

HP is announcing at the LinuxWorld Conference & Expo, starting here today, that it's sharing the code for its High-Performance Computing (HPC) Parallel Compositing Library, a tool to help researchers access unused computing power in their data centers to "visualize complex data sets," said Doug Small, marketing director for Linux and open source at HP.

Parallel Compositing is an application programming interface developed by HP in collaboration with universities and other industry experts. Its purpose is to combine the unused graphics card capacity of a number of high-performance computers working together in a cluster to simplify the data sets created by computer simulations and other large data sets in such fields as pharmaceuticals, seismology, and medicine, said Small.

"One of the challenges in high-performance computing is visualizing the data sets, getting the horsepower to accomplish that," he said. "With parallel compositing, we're breaking up the work across all those [graphics cards] to use the processing power to work on the visualization of the data."

One user of the technology, cancer research center MD Anderson, spent $2.2 million with HP for a Linux-based HPC that uses Parellel Compositing for computer modeling and other research in bioinformatics, epidemiology and radiation treatments, Small said. HP, which touts Parallel Compositing as a cost-effective approach, is releasing the code to spur technology development around it.

Read more...