A Framework for Assisting Large Data Visualization on Fugaku Environment
Only 129 seats left
Tuesday, June 29, 2021 3:00 PM to 4:00 PM · 1 hr. (Africa/Abidjan)
Information
Contributors:
Abstract:
During the operational period of the K computer, several PBR (Particle Based Rendering) based applications have been developed for assisting large data visualization. One of the reasons was the initial lack of the Mesa 3D graphics library, and the other reason was probably the visibility sorting free nature of the PBR technique which facilitates the parallel rendering and the handling of irregular mesh data. In this poster, we present a framework to assist large data visualization on the Fugaku environment to provide the PBR for those users who are accustomed to use it. This is based on KVS (Kyoto Visualization System) library, which has a native support for PBR as well as other traditional rendering techniques. Large data visualization is handled via hybrid MPI/OpenMP parallelism and sort-last image compositing based on the 234Compositor. The proposed visualization framework is still in the initial development phase, we expect that will soon serve as an effective large data visualization alternative, on the Fugaku environment, especially to those users accustomed to use the PBR for their visual analysis of the simulation results.
Abstract:
During the operational period of the K computer, several PBR (Particle Based Rendering) based applications have been developed for assisting large data visualization. One of the reasons was the initial lack of the Mesa 3D graphics library, and the other reason was probably the visibility sorting free nature of the PBR technique which facilitates the parallel rendering and the handling of irregular mesh data. In this poster, we present a framework to assist large data visualization on the Fugaku environment to provide the PBR for those users who are accustomed to use it. This is based on KVS (Kyoto Visualization System) library, which has a native support for PBR as well as other traditional rendering techniques. Large data visualization is handled via hybrid MPI/OpenMP parallelism and sort-last image compositing based on the 234Compositor. The proposed visualization framework is still in the initial development phase, we expect that will soon serve as an effective large data visualization alternative, on the Fugaku environment, especially to those users accustomed to use the PBR for their visual analysis of the simulation results.
