About | Formal basis | Workpackages | Documents | Meetings | Partners | Contact list |

 »   Workpackages

 »   WP 2.3 Advanced visualization tools

Leader »
Witold Alda, PhD
ACK CYFRONET AGH, Kraków
Co-executors » TASK, WSNC
Start date » 1 Dec, 2002
Ending date » 31 Oct, 2004
[ Timetable |  Team ]
Short task description

Graphic data processing requires access to highly efficient visualization capabilities directly on the user terminal. Visualization tools should be able to gather data coming from HPC (High Performance Computing) computations run on a remote system and then to display them interactively on the local user screen. The visualization process may be run on a remote computer (in that case processed images are transmitted) or alternatively - raw data may be sent to the user to be processed locally.

OpenGL Vizserver software offered by SGI supports remote visualization and enables the users to browse and cooperate interactively with huge data sets from their local stations. OpenGL Vizserver does not run rendering itself but controls the visualization on the host machine and the compressed image data transmission to the user. It is intended to install the OpenGL Vizserver system in two computational centers: in Poznan and in Kraków. Among the tools that may be used for visualization on the host machine within the confines of the project, the following systems will be run and analyzed: AVS from Advanced Visual Systems, Inc., Open Data Explorer from IBM and Vis5D from the University of Wisconsin. Typical applications remaining in the focus of attention cover scientific visualization, medical image data processing and transmission, 3D objects reconstruction and virtual worlds creation.

To accelerate both computations and visualization a mechanism of distribution and parallelisation of the tasks run on clusters should be prepared. It is a natural approach to decompose tasks into subtasks sent to different computers or systems, performing remote computations and gathering the results on the graphic server. This operating mode may be realized using the VisBench library from NCSA and CUMULVS library from the Oak Ridge National Laboratory.