[University home]

Research Computing

Parallel Express Projects

AVS/Express Multipipe Edition (MPE)

The AVS/Express Multipipe Edition (MPE) product from Advanced Visual Systems is the result of a joint effort undertaken by Research Computing Services, AVS and Knowledge Graphics Technology (KGT). Developed here in Manchester, the project has created a new rendering environment for any AVS/Express visualization application, taking full advantage of parallel, large scale immersive graphics facilities of SGI Multipipe systems and PC (Windows and Linux) clusters.

What is MPE?

AVS/Express MPE running on a Powerwall System

The AVS/Express Multipipe Edition offers users the most comprehensive suite of data visualization and analysis capabilities available today in multipipe or multi-channel display environment. The MPE software is flexible and easy to use, providing a wide range of visualization tools and services for immersive or virtual reality (VR) systems.

Packaged as a high-level tool-kit, MPE software allows rapid construction and tuning of immersive applications, without involving a detailed understanding of the low-level multi-channel graphics programming. This data visualization tool-kit allows users to interactively visualize data in a full 3D stereo immersive display environment. By selecting from a large library of modules, the most appropriate technique can be used during an interactive exploration session.

With AVS/Express, a single visualization application can be moved between multiple systems without having to recompile or change a single line of code. You can be using a desktop PC in the morning, make a discovery, and show your colleagues in a powerful multipipe immersive environment in the afternoon.

As a true development environment, stand-alone multipipe applications can be created for distribution, including full C or C++ based customization.

AVS/Express MPE on the Powerwall Display Researchers in Physics, Chemistry, Engineering and Environmental sciences can make and share fundamental discoveries by exploring large data sets. Commercial organizations, such as Oil & Gas or Pharmecutical software companies can create immersive discovery tools for their industry.

The AVS/Express Multipipe Edition runs on Silicon Graphics® multi-pipe high performance workstations (Irix and IA64) and on PC clusters running under Linux or Windows operating systems. Each pipe may have several display channels and projectors, for large screen or smaller Immersive desktop stereo systems.


Development

Further development of the MPE system will focus on a new improved embedded VR Menu system, a port to the SGI Altix environment and integration with the AVS/Express Parallel Edition (PE) project, also developed here in Manchester.

Project Partners

Active partners in the AVS/Express Multipipe Rendering Project are:

Project Status

The research and development effort is ongoing, with periodic demonstrations and product releases. The software is now in use in the US, Japan and Europe.

A number of visualization applications using this technology are now running in the Manchester VIPL facility.

AVS/Express Parallel Edition (PE)

Visible Woman isosureface computed with PST

Overview

One drawback of both AVS/Express and MPE is the lack of support for parallel computation. The Parallel Edition (PE) aims to remedy this limitation with the development of a framework for heterogeneous distributed computation and an extensive suite of performance visualization modules. The framework will integrate control of data decomposition, distribution, large dataset handling, level of detail, data streaming and asynchronous computation for facilitating computational steering.

This project is a collaboration between JAEA, KGT, AVS and Research Computing Services.

Conferences and Publications

Distributed Data Renderer

The Distributed Data Renderer (DDR) is the latest stage of development for parallelization of AVS/Express. It combines the power of parallel computation and rendering in a tightly integrated environment enabling users to handler large datasets and take advantage of compute clusters that don't necessarily contain supporting graphics hardware.