Image Processing Reference
In-Depth Information
Simulation Script
CFD Solver
Module
CSM Solver
Module
Solver XY
Module
Steering
Module
Data Management
File
System
Fig. 31.8
Integration of FSSteering into the multi-disciplinary simulation environment Flow-
Simulator. This deep integration allows for simple usage and accessibility to many parameters
and methods
ICARUS, interprets an additional XML file describing data layout and user com-
mands.
Contrary to extending visualization tools, pure native steering frameworks exist,
such as the Steereo [
22
] or INRIA's EPSN [
7
]. They often concentrate on coupling
different simulations and distribute data among them easily. Therefore, missing
analysis tasks as well as the visualization have to be implemented by the user. This
results in a high adaptation overhead for specific problems.
FSSteering [
41
] is a domain-specific framework focussing on multi-disciplinary
CFD workflows. It is integrated into Airbus's FlowSimulator framework (see
Fig.
31.8
). Important parameters as well as the simulationmesh are accessible and can
be changed on-demand without the need to prepare the simulation for this. Due to its
deep integration, instrumentation is possible with minimal changes to the simulation
script and no additional description has to be maintained.
31.3 Techniques for Parallel and Remote Rendering
In most cases, large-scale datasets are only stored where they are generated. A direct
access to storage systems is often prohibited or restricted. However, even if the dataset
can be downloaded, it would be too large to be stored locally and the transmission
time would exceed acceptable waiting periods. Remote and parallel rendering tech-
niques are meaningful tools to avoid those I/O problems. The rendering is directly
performed where the data is accessible. But not only the possibility to process the
results on the same HPC resources where they have been created but also the avail-
ability of massive parallel rendering capacities at today's computing centers helps to
accelerates the delivery of rendered images. Finally, presenting intermediate results
of running simulation in time can also easily be integrated in such distributed and
parallel rendering frameworks.
Search WWH ::
Custom Search