RAI has years of expertise creating highly customized visualization systems for our customers:

• Mesh visualizations such as those employed in the automotive, aerospace, and manufacturing fields to create visual models of their products
• Seismic and volume analysis in the oil and gas companies
• Materials and volume simulations like those used by science labs
• Photorealistic simulations using ray tracing for television and film

The RAI VIZ has many features which are completely unique in the visualization node market:

• RAI GPFS a premier parallel cluster file system to ensure that the heavy data storage and huge data bandwidth requirements of a visualization cluster will be reliably met.

• Carefully chosen graphics hardware and drivers.
• Validated performance engineering, providing extensive testing, benchmarking, and optimization before leaving the RAI labs to be installed and benchmarked again onsite by an experienced cluster professional.
• The RAI VIZ can be rapidly deployed as a stand-alone visualization system, scalable from 3 to 32 nodes, or an RAI Visualization Solution can be easily added to an existing cluster solution  providing a fully integrated high performance computational and visualization solution.
• XWC, a powerful suite of management and monitoring tools, is a part of each RAI VIZ Solution.

RAI recommends the inclusion of storage and visualization nodes within the compute node cluster fabric to,

• Maximize performance by minimizing load and storage time
• Avoid the delays of moving large data sets to the desktop
• Minimize network bandwidth utilization by remoting post-rendered pixels to the desktop
• Speeds multiple iterations as the compute, storage and visualization resources exist on the high bandwidth interconnect

 RAI has also spent time helping customers to remote their visualizations to their engineers, thus allowing them to use their clusters in a centralized location and support engineering groups around the world. Many of our larger installations use CEI's EnSight DR (Distributed Rendering) in combination with OpenGL and Turbo VNC to accomplish the remote visualization as illustrated in the diagram below.

·         The Server of Server node arbitrates requests from Clients and maximizes the performance and efficiency of the cluster  

·         Rendering nodes render, rotate and animate the model at full speeds and resolutions

·         Compositing the view of the model back to the Client sends only the pixels and model elements that are needed for viewing; maximizing efficiency and natively supporting pan and zoom

·         EnSight DR Clients can either be on or off the cluster (e.g. a Windows Desktop PC)