PBVR
 (Particle Based Volume Rendering)

PBVR 2.0 (December 12, 2022)

・The latest version of PBVR 2.0 is available. In this update, source codes of CS/IS-PBVR's client are integrated, and pre-defined colormaps are updated.

The source codes are available from github.
https://github.com/CCSEPBVR/CS-IS-PBVR

Download

Download List

Purpose and Overview of the Program Development

The development of Peta-scale supercomputers such as the K-computer has dramatically extended the capabilities of numerical simulations. However, simulation data produced in such Peta-scale simulations is becoming too large compared with the network bandwidth, the storage and processing speed on PCs, and therefore, the conventional post-processing approaches, in which the simulation data is transferred and processed using commodity visualization softwares, are facing difficulties.

To resolve this issue, the Center for Computational Science & e-Systems in the Japan Atomic Energy Agency (JAEA) developed a client-sever type remote visualization software PBVR based on the Particle Based Volume Rendering techniques and the KVS library, which are developed at Koyamada laboratory in the Kyoto University.

This software enables the following remote visualization procedures with minimum data transfer.

1. A filter program on a server machine (supercomputers, clusters, workstations) decompose large-scale structured/unstructured volume data into sub-volume data for parallel visualization.
2. A server program on the server machine compresses the volume data to small particle data using massively parallel processing.
3. The small particle data is transferred by the socket communication between the server machine and a client PC.
4. A client program on the client PC renders the particle data with interactive frame rates.

These features enable interactive remote visualization without transferring large-scale volume data.

This software utilizes the massively parallel random number generator, KMATH_RANDOMdeveloped at the RIKEN-AICS, and the GLUI library. The development of the K-computer version is supported by the MEXT, grant for the HPCI Strategic Program-Field No.4: Next-Generation Industrial Innovations.

Main Features

・PBVR supports structured and unstructured grid data formats with AVSFLD/UCD, PLOT3D, STL, and VTK Legacy.

・The server program is optimized for massively parallel environments such as the K computer, FX10, and Linux clusters, and achieved the strong-scaling up to 1024 cores by the dynamic load-balancing. The program works also on Win/Mac/Linux PCs.

・The client program works on Win/Mac/Linux PCs and interactive frame-rate is achieved up to 60 fps.

・The client program and the server program on multi-platform environments can be connected by socket communication through ssh tunnel.

・Depending on the processing environment, the system enables flexible choices of operation modes such as client-server interactive processing, server batch processing + client processing, and stand-alone processing.

・In addition to volume rendering visualization, iso-surface, slice surface, and multivariate visualization are possible by designing multi-dimensional transfer functions.

≪Processing flow of PBVR≫
Processing flow of PBVR
≪GUI of PBVR≫
GUI of PBVR
≪Comparisons of total remote visualization performances of PBVR and the commodity visualization software EnSight using 48 cores of the BX900≫
(Plasma turbulence data with 300 million grids. Server machine: XeonX5570, Client machine: XeonE5+QuadroK5000, Network bandwidth ~3.4MB/s)

PBVR

Ensight

Filter [sec/step]

0.7

-

Image Production [sec/step]

51.4

3873

Data transfer [sec/step]

75.7

Total [sec/step]

127.8

3873

Rendering speed [frame/sec]

60.0

2.7

Client memory [MB]

257

900

Reference

[1] Kawamura, T., Idomura, Y., Miyamura, H., Takemiya, H., Sakamoto, N., and Koyamada, K., “Remote Visualization System based on Particle Based Volume Rendering”, Proceedings of Conference on Visualization and Data Analysis 2015 (2015)

Developer

Computer Science Research and Development Office, Center for Computational Science & e-Systems, Japan Atomic Energy Agency

Contact

ccse-quad(at)ml.jaea.go.jp
※ Substitute @ for (at).