A research component is being developed to meet both short-term and long-term needs of SciDAC science applications.  The Institute for Ultra-Scale Visualization team will develop and evaluate parallel visualization strategies on diverse platforms including general-purpose clusters, dedicated visualization clusters, and especially general-purpose massively parallel supercomputers.  This  research component will complement rather than duplicate other visualization CET efforts.  The outreach and education component of the Institute will allow us to set up venues for the important communication and collaboration with other visualization CET efforts as well as the application scientists who will be the users of the deployed technology.  Our team has expertise covering the full spectrum of parallel visualization technology, and we have identified the following five critical technology focus areas:

• Parallel rendering framework and API
  

• Parallel I/O
  

• Parallel visualization algorithms
  

• GPU abstraction and interfaces
  

• Parallel visualization tools

A tightly coordinated effort on these research topics will lead to the deployment of scalable, usable parallel visualization systems at DOE Leadership Computing Facility (LCF) and Flagship Computing Facility (FCF), and thus a viable solution to petascale visualization.  We will also enable ultra-scale visualization on scientists' desktop by utilizing the high-performance computing resources provided by LCF and FCF over network.

In addition to parallel visualization, the following research areas are also under investigation:

• In-Situ Visualization

• Time-varying multivaraite data visualization

• Uncertainty visualization

• Visualization interfaces

• Feature extraction and tracking

• Vector field visualization

• AMR data visualization

• Distance and Collaborative visualization

• Comparative visualization