A Distributed Memory Hierarchy and Data Management for Interactive Scene Navigation and Modification on Tiled Display...

Area of Science:

• Computer Graphics
• Data Management
• High-Performance Computing

Background:

• Interactive navigation and modification of massive 3D models are hindered by data layout changes impacting out-of-core rendering performance.
• Existing methods struggle with maintaining data coherency and efficient performance during simultaneous operations on large datasets.

Purpose of the Study:

• To propose a novel distributed data management approach for simultaneous interactive navigation and modification of massive 3D data.
• To leverage tiled display infrastructure as a cluster of storage-compute-display (SCD) nodes for enhanced performance.

Main Methods:

• Developed a distributed memory hierarchy tailored for interactive rendering applications.
• Introduced a new data partitioning strategy for distributed storage across SCD nodes to minimize data load variance.
• Implemented a data management technique that leverages screen-space and object-space adjacency without data duplication.

Main Results:

• The proposed method effectively manages massive 3D data for simultaneous interactive navigation and modification.
• Empirical evaluations on large datasets demonstrated superior performance compared to state-of-the-art methods.
• Reduced latency and improved coherency in data access were achieved through optimized data distribution.

Conclusions:

• The novel distributed data management approach significantly enhances interactive performance for massive 3D models.
• The method provides an efficient solution for handling large-scale 3D data on tiled display clusters.
• This work offers a scalable and effective strategy for complex 3D data manipulation and visualization.