Content

•  
  Introduction
•  
   Problem Description
•  
   Areas of Interest
•  
   Goals and Contributions
•  
   Thesis Structure
•  
  Foundations
•  
  Clusters Now and Then
•  
   The Rise of Clusters
•  
   Hybrid Cluster Systems
•  
   The Arminius Hybrid Cluster at PC2
•  
   Software for Clusters
•  
   Application - Simulation
•  
   Systems, Models and Simulation
•  
   Distributed Simulations on Clusters
•  
  Application - Interactive Simulations and VR
•  
   Definitions
•  
   IS/VR on Hybrid Clusters
•  
   The VisSim Project
•  
   Subsystems for Traditional and Hybrid Clusters
•  
   Hardware APIs and NVIDIA CUDA
•  
   CUDA - Architecture
•  
   CUDA - Programming Model
•  
   CUDA - API and Interoperability
•  
  Related Work
•  
   Computational Steering
•  
   Existing Systems for Traditional CS
•  
   Approaches towards CS for Interactive Simulations
•  
   Remote Visualization
•  
  Classes of Remote Visualization
•  
   Client-Side Rendering
•  
   Server-Side Rendering for 2D and Administration
•  
   Server-Side Rendering for 3D Applications
•  
   Limitations and Problems
•  
  Two Essential Subsystems for IS/VR on Hybrid Cluster Systems
•  
   Traditional Usage of Hybrid Cluster Systems
•  
   Hybrid Cluster Systems for IS/VR
•  
   CS for IS/VR - Idea and Requirements
•  
   Advanced RV Techniques for IS/VR - Idea and Requirements
•  
   Computational Steering of IS/VR
•  
  A Concept for Extended Computational Steering of IS/VR
•  
  New CS Models
•  
   Collaborative Computational Steering
•  
   Synchronized Computational Steering
•  
   Concurrent Computational Steering
•  
   Combinations of the Models
•  
   Requirements
•  
   Conceptual Design and Classifications
•  
   Actors in Computational Steering
•  
   Classification of Simulation Data
•  
   Passing of Volatile State Data
•  
   Scheduling of Volatile State Data
•  
   Dynamic Mapping of Volatile State Data
•  
   Decoupled Handling of Shared Parameters
•  
  Architecture of the Framework
•  
   Communication Server
•  
   Publish/Subscribe Service
•  
   Exemplary Workflow
•  
   Consistency, Performance Estimation and Latency Optimization
•  
   Best Effort and Dropping of Latecomers
•  
   Clustering of Communication Servers
•  
  Prototype - The CSIS Framework
•  
   Commuvit
•  
   Architecture
•  
   Variable Mapping
•  
   D-Bus
•  
   The CSIS Server - Integrating Commuvit and D-Bus
•  
   The CSIS Steering Library
•  
  Computational Steering of a Distributed Driving Simulator
•  
   The Virtual Night Drive Simulator
•  
  Modularizing the VND Simulator
•  
   The Input Component
•  
   The Simulation Component
•  
   The Visualization Component
•  
   The Audio Component
•  
   Exemplary Setup of the VND with CS
•  
   Example for the Passing of Volatile State Data
•  
   Example for the Passing of Shared Parameters
•  
   CS-Enhanced Virtual Night Drive - Conclusion
•  
   Distributed Shader-Based Visualization through CS
•  
  Computational Steering for an Interactive Material Flow Simulation
•  
  The d3FACT insight Material Flow Simulation
•  
   Initialization and Execution of Simulations
•  
   Modeling with Buildingblocks and Subblocks
•  
   The Simulation, Tokens and Visualization
•  
   Communication Limitations
•  
   Computational Steering in d3FACT insight
•  
   Adapting the MFS Kernel and the Visualization
•  
   Interfaces
•  
   Data Exchange
•  
   Example Scenario
•  
   CS-Enhanced d3FACT insight - Conclusion
•  
   Conclusion - Computational Steering of IS/VR
•  
   Remote Visualization for IS/VR
•  
   Limitations of Remote Visualization for IS/VR
•  
   Slow Compression for High Resolutions
•  
   Reading Image Data From the Graphics Hardware
•  
   Rendering to Multiple Targets
•  
   Programmable Graphics Hardware and GPGPU
•  
   Invire - A Concept for an Interactive Remote Visualization System
•  
   The Architecture
•  
   Data Transfer
•  
   Image Readback
•  
   Compression
•  
   Run Length Encoding
•  
   Difference Compression with Index
•  
   Parallel Difference Compression with Index
•  
   JPEG
•  
   Parallel JPEG Compression
•  
   Remote Visualization on Hybrid Clusters
•  
  Prototype - The Invire Framework
•  
   Invire Plugin
•  
   Integration into Host Applications
•  
   Remote Control of the Application
•  
  Invire Client
•  
   Reception, Decompression and Displaying of the Frames
•  
   Controlling the Remote Application
•  
  Invire Library
•  
   The Frame Object
•  
   Compression/Decompression - General
•  
   Compression - CUDA-Based DIC
•  
   Decompression - CUDA-Based DIC
•  
   Compression - CUDA-Based JPEG
•  
   Benchmarking the RV Framework
•  
   The Reference Framework VirtualGL
•  
   Quality Assessment with the SSIM Index
•  
   The Benchmarked System and the Sample Applications
•  
  Benchmarking the Overall System Performance
•  
   Rotating Teapot
•  
   Virtual Night Driver
•  
  Benchmarking Three Different JPEG Implementations
•  
   JPEG - Compression Times
•  
   Quality Assessment of the Lossy Compression Methods
•  
   Benchmarking Conclusion
•  
   Conclusion - Remote Visualization for IS/VR
•  
  Conclusion
•  
   Contributions
•  
   Conclusion
•  
   Outlook
•  
   Bibliography
•  
   Acronyms