Distributed visualization is a kind of distributed computing. In that kind of computing many computers are used to render parts of bigger project. Main reason for that is complexity of rendering which prevents us from using single PC. Rendering some animations on single PC can take not just hours, but days or months. In order to finish rendering as quick as possible we use multiple computers. Only way to render on distributed network is to split animation into frames and render them separately. Every machine will get one or few frames to render. After renderirng all frames should be collected and joined together.
|
|
Visualization on cluster is simpler than visualization on grid. Command qsub is used to job submission, and qstat to veiw the status of the job. Communication between nodes is much faster, so there is no need for downloading results. All results are being automaticly downloaded to your home directory or to directory where they were run from. That depends of server configuration, but the path can be set up. |
|
Read more...
|
|
If you don't have enough local resources, grid is ideal solution for you. As mentioned in the distributive visualization article your input has to be split into parts and sent to grid by using the scripts.
Veppar project uses SEEgrid infrastructure where every job must be submitted with special scripts.Here is one example of those JDL scripts which are used for submission. |
|
Read more...
|
|
All pictures in this gallery are used for testing infrastructure of grid, and clusters.
The first picture represents light harvesting complex which was get from PDB database with code 1RZG found at the Protein Data Bank web pages. Every atom from database file is represented as sphere (spacefill method), and uses unique color which makes it different from other atoms. H uses light blue color, C uses white color, O uses red color, N is blue. Yellow colour represents sulphur. Every atom has it's own radius. All mentioned atom atributes were get from Damir Zucic's molecular visualization software named Garlic. All PDB data was exported to relational database (developed at Unveirsity of Split, Faculty of Natural Sciences, at the Department of Physics) specially designed for easier and faster visualization and analysis of membrane proteins, or other molecules. Data salected from databse is exported into povray, and rendered on grid or clusters, depending on it's complexity.
|
|
Read more...
|
|
|
|
|
|
