The Visual Supercomputer is the umbrella name for SGI's new range of high-performance visualisation solutions targeted initially at engineering analysis, scientific visualisation, manufacturing, styling, rendering and Reality Center® applications.

The solutions are SGI's response to the most important recent trends in visualisation. For example, as model sizes are increasing, the clustered computing solutions used by many organisations are adding complexity to the visualisation problem. Distributed visualisation is also growing in importance, as is the need for remote collaboration.

The Visual Supercomputer is designed to address the five key steps in the visualisation process - building a model, rendering it, transmitting the results (locally or via a LAN or WAN), displaying them and interacting with them. SGI's strategy is to optimise performance at each of these stages, with a particular focus on distributed visualisation (through Visual Area Networking) and remote collaboration (through the Wide Area Visualisation Environment).

The solutions are being deployed through the addition of competitively priced, tightly coupled X86-based Graphics Nodes to SGI® Altix® clusters including SGI Altix 4700, SGI Altix XE 310 and SGI Altix ICE - with programmable graphics processing units (GPUs) being used for computation only. CPU- and/or GPU-assisted rendering and remote visualisation middleware and applications provide users with multi-paradigm rendering options and libraries - offering greater choice, and broadening the market for the new solutions. Workflow optimisation within visualisation environments is provided by SGI Professional Services.

What are the benefits?
Taking scientific visualisation as an example, a major benefit of the Visual Supercomputer is that while many scientific users have access to significant computing power, this does not necessarily enable them to gain insight as efficiently as possible. Compute-only solutions fail to solve the entire visualisation problem, because an organisation's people, data, compute and visualisation resources are rarely collocated.

By comparison, the SGI approach enables a customer to purchase an X86/Linux®-based solution delivering outstanding price/performance and scalability for almost any visualisation application, with the added benefit of application-transparent remote visualisation.

Many scientists like to think of their computing environment as a "Grid", which virtualises all aspects of computing to the extent that they can simply make a request to the Grid and then receive a result. In supercomputing environments (especially those involving Terabytes of data), storing, analysing and visualising this data locally can be impractical; whereas with the Visual Supercomputer and remote visualisation, SGI can provide scientists with real-time access to huge simulation results via their desktop - enabling them to interact with and control the visualisation application remotely, even over a WAN.

In computer-aided engineering, meanwhile, SGI is able to offer a choice of visualisation techniques, highly differentiated server platforms, and its deep domain expertise in visualisation to provide end-to-end solutions for CAE users that address their workflows in a tightly integrated fashion, so improving return on investment. And in rendering, the Visual Supercomputer enables SGI to offer bundled applications utilising CPUs or GPUs for rendering, and a robust, single source for the entire render farm (including servers, switches, storage and software).

Image courtesy of the University of Toronto Institure for Aerospace Studies.

FURTHER INFORMATION
www.sgi.com/visualization