1 / 26

Visualization of engineering computations in LitGRID and G ridTechno projects

Visualization of engineering computations in LitGRID and G ridTechno projects. Arnas Ka čeniauskas Laboratory of Parallel Computing arnka @fm.vgtu.lt. Projects. GridTechno (2007-2009): HTP project supported by Lithuanian State Science and Studies Foundation.

inigo
Télécharger la présentation

Visualization of engineering computations in LitGRID and G ridTechno projects

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Visualization of engineering computations in LitGRID and GridTechnoprojects Arnas Kačeniauskas Laboratory of Parallel Computingarnka@fm.vgtu.lt

  2. Projects • GridTechno (2007-2009): • HTP project supported by Lithuanian State Science and Studies Foundation. • Visualization of Engineering Applications in GRID environment. • LitGrid (2007-2011): • National program supported by Ministry of Education and Science. • GRID Visualization System. • BalticGrid-II (2008-2010): • FP7 project. • Visualization Tools.

  3. Pilot Applications • Poly-dispersed particle systems (DEM): • Particle technology, laser technology, fracture mechanics. • Available visualization tools based on VTK and Matlab. • Needs feature extraction (crack propagation in micro-mechanics). • No inherent connections (1D connections for 3D domain can be computed in DEM software)! • 3D cells are needed for interpolation?! • Coupled CFD problems (FEM): • Available visualization tool based on OpenDX. • Requires basic visualization functionality. • Current implementation loads all time steps into RAM at the same time, therefore, it needs a lot of computer memory. • Porous media applications (FDM or FVM): • Available visualization tool based on Matlab. • Can not visualize results from multiple domains obtained from parallel computations.

  4. Particle Systems Poly-dispersed particles colored by radius (VTK)

  5. CFD: Sluice Gate Problem Pressure and velocity field visualized by OpenDX

  6. Flow in Porous Media • Oil filter manufactured by IBS Filtran. 4 processors 1 processsor

  7. User Groups • Visualization is driven by user needs. • Strongly dependent on application, • Dependent on data form, dataset size and its location, • Dependent on hardware resources, • Knowledge level of users is very important! • Experienced users (few): • Has work experience with commercial software, • Might have their own visualization tools, • Need powerful software for large datasets, • Need extended functionality, • Might need some level of freedom. • Beginners (many): • Lack of basic knowledge on visualization. • Need easy, well designed interface for their application.

  8. DataSet1 DataSet2 Source1 Source2 Filter1 Filter2 Filter2 Filter3 Mapper1 Mapper2 Renderer1 Renderer2 User Groups • Visualization Pipelines: • Suitable for strictly defined tasks. • Can be very specialized (adopted to particular application). • Easy to use (suitable for beginners). • Quite easy to develop. • Good basis for visualization tools. • Each application needs development efforts!

  9. Mapper1 Renderer1 Filter1 DataSet1 Universal Source Filter2 Filter4 Mapper2 Renderer2 Filter3 Mapper3 User Groups • Visualization Networks. • Can be made very universal. • Users should have basic knowledge on visualization. • More difficult to develop. • Good basis for visualization system. • New applications might not need additional development! DataSet2

  10. Decision • Develop universal visualization system with basic functionality: • Suitable for experienced and unexperienced users in general, • Any new user can learn it. • Develop several visualisation tools with particular functionality: • Necessary for small number of experienced users that need extended functionality, • Universality can be sacrificed in order to get required efficiency. • Develop several visualisation tools with particular user interface: • Suitable for small number of users that need simplified user interface, • Universality can be sacrificed in order to get required simplicity (limiten number of choices).

  11. Data location in GRID • Network-oriented Visualization: • Data is not located in user PC. • UI (User Interface): • Suitable for small datasets, • Easy, because do not need any special GRID visualization software, • User connects to UI by using protocol suitable for graphics. • SE (Storage Element): • Suitable for large datasets. • Transfer the whole file (data set) if it is not too large. • Transfer a part of the dataset (needs special software)! • WN (Working Nodes): • Computational steering in GRID presents challenge. • Needs interaction between job submission/monitoring and data transfer/visualization.

  12. Distributed Visualization System • Platform: • Java EE 5 platform. • Server: • SVE implemented as service, • Service container (application server) run on UI (usual access to GRID resources). • Client: • Run on user PC, • Java application is more reliable than Java applet.

  13. Distributed Visualization System • Data: • NetCDF(NetworkCommonDataFormat), • HDF5 (Hierarchical Data Format). • Geometry: • VRML, • X3D. • SVE or CVE: • VTKthe best choice for developers, • OpenDX or other MVE. • GUI: • Java Swing, • Viewer: • Xj3D, • VTK based, • Other …

  14. Data Filtering Prepared Data Mapping Geometry Rendering Image Distributed Visualization System Network

  15. Distributed Visualization System • Server sendsgeometry to client: • Sufficient interactivity on aclient: • Rotation, translation, scaling. • Suitable for datasets of medium size: • VRML/X3D file is transferred. • Advantages: • Load is distributed between client and server, • Data remains in the server, • Modest Network bandwidth, • Easy to use and not very difficult to develop. • Disadvantages: • Enough powerful server is required.

  16. WSDL Interface WSDL Compiler Service Server Server Skeleton Client Stub Client SOAP HTTP JAX-WS Runtime e-Service Implementation • Implementation of WEB service: • Java EE 5.0 platform, • Service container/Application server GlassFish v2,

  17. e-Service Implementation • Integrated in Netbeans 6.0 IDE : • WEB service designer, • WEB service tester, • Example including 3 public methods and parameters, • Easy implementation.

  18. e-Service Implementation • WEB service implementation: • In Java EE 5 platform, JAX-WS 2.1 (Java API for XML Web Services) has replacedJAX-RPC API (earlier J2EE platform). WEB services based on XML documents replaced RPC. • JAXB binds XML documens to Java objects during runtime, • Supports Spring 2.0 framework. • Large, zipped X3D file transfer: • Can be performed by using scp (jsch-0.1.37.jar) instead of SOAP over HTTP. • Visualization services in GRID would need: • Stateful resources, • Dynamic and transient nature of services, • Client notification about changes in a service. • OGSA and WSRF standards: • Implemented in Globus Toolkit. • How to implement WSRF in gLite based GRID?

  19. Visualization System • Meniu by Java Swing: • Reading selected datasets from HDF5 file, • Time dependent and time independent data.

  20. Visualization System • Filters window: • ExtractGeometry/ExtracyPolyData filters, • Other filters.

  21. Visualization System • Mappers window: • Glyphs for scalars, • Color Mapping.

  22. Implemented Functionality • Data processing (Filtering): • Extracts the defined part of geometry (vtkExtractGeometry). • Removes some points considering scalar field values (vtkThresholdPoints). • Process vector fields (vtkVectorNorm, vtkExtractVectorComponents). • Creates connections (vtkDelaunay3D, vtkDelaunay2D). • Cuts geometry (vtkCutter). • Visualization (Mapping): • Positions (vtkGlyph3D,vtkIdFilter, vtkLabeledDataMapper). • Connections (vtkExtractEdges, vtkTubeFilter). • Outline (vtkOutlineFilter). • Scalars by isosurfaces (vtkContourFilter, vtkContourGrid) • Scalars by glyphs (vtkGlyph3D) • Vectors by glyphs (vtkGlyph3D) • Color Mapping (vtkScalarsToColors, vtkLookupTable)

  23. Implemented Functionality • Positions and Numbers: • vtkGlyph3D, • vtkLabeledDataMapper. • Positions and Connections: • vtkGlyph3D, • vtkTubeFilter.

  24. Implemented Functionality • Positions and Isosurfaces: • Velocity magnitude, • vtkContourGrid. • Particles: • Colored by radius, • vtkGlyph3D.

  25. Viewer • Xj3D Browser • Supports VRML and X3D, • Fails to draw VTK scalar bars! • Fails to import several vtkRenderers rendering in one window! • Difficult to implement custom interactivity. • VTK importers: • Does not support X3D! • Does not support vertex colors !!! • Easy to implement VTK widgets in remote environment. • Cassandra VTK viewer: • Inherits all VTK problems ...

  26. Future Plans • Performance tests(2008): • Implementing other VTK classes/functionality, • HDF5 read/write speed for different dataset models, • VRML/X3D transfer options and protocols, • Reliable viewer and interactivity. • Powerful GRID visualization (2008-2009): • SE. Transferring part of the dataset (data file). • WN. Computational steering. • Visualization of large data (2009-2010): • Data streaming, • Distributed parallel processing, • Software ray-casting, sort-last rendering.

More Related