Clusters have become the mainstays of computing in science, research and engineering within HPC environments. The amount of data generated by today's cluster systems is expanding exponentially. More and more scientists and engineers are relying on visualization techniques to turn this data into information, and to gain better insights. Vast amounts of data span the computing world, from simulations of a variety of physical systems to huge quantities of telemetric data, an example being seismic processing and analysis in the Oil and Gas industry.
ISVs and research institutions from a wide array of disciplines continue to advance the state of the art with new algorithms for rendering the raw data. HP has been complementing this work by researching ways of supplying vast amounts of scalable computational, rendering and storage power needed by these users at an affordable cost.
Distributed clusters with commodity processors and graphics cards coupled by high-speed networks provide the affordable computational and visualization power. HP, through this collaboration, is investigating techniques for distributing once monolithic rendering codes across the CPUs and GPUs of such a cluster.
Central to the current investigation is the Parallel Compositing Library. This Library greatly reduces the effort needed to distribute rendering across a set of machines and develop parallel graphics applications.
|
Printable version
