Scalable isosurface visualization of massive datasets on commodity off-the-shelf clusters

Area of Science:

• Computer Science
• Data Visualization
• Scientific Computing

Background:

• Massive datasets from tomographic imaging and simulations require advanced visualization tools.
• Interactive and exploratory visualizations are crucial for analyzing large volumetric data.
• Existing frameworks may face scalability challenges with increasing data sizes.

Purpose of the Study:

• To develop a scalable, end-to-end parallel and progressive visualization framework for large volumetric data.
• To enhance the backend scalability of isosurface extraction for massive datasets.
• To introduce efficient methods for isosurface compression and progressive transmission.

Main Methods:

• Developed a scalable isosurface visualization framework on commodity clusters.
• Implemented a fully parallel and out-of-core isosurface extraction algorithm.
• Utilized parallel disks and I/O-optimal external interval trees for efficient data handling.
• Introduced an isosurface compression scheme for progressive extraction, transmission, and storage.

Main Results:

• Achieved scalability through parallel and out-of-core processing with parallel disks.
• Minimized I/O operations by statically partitioning data and using external interval trees.
• Demonstrated an efficient isosurface compression scheme.
• Enabled interactive browsing of extracted isosurfaces via parallel rendering and specialized hardware (Metabuffer).

Conclusions:

• The proposed framework offers an end-to-end solution for visualizing massive volumetric data.
• The parallel, out-of-core isosurface extraction algorithm significantly improves backend scalability.
• The compression scheme enhances efficiency in data transmission and storage for progressive visualization.