A02-KB104 Creation of Large-scale and High-Performance Technology for Processes of Multidisciplinary Computational Anatomy by Utilizing Supercomputers

From Multidisciplinary Computional Anatomy
Jump to: navigation, search

Member

  • Primary Investigator
    Takahiro Katagiri ( Nagoya University, Information Technology Center, Professor)
  • Corporative Researcher
    Satoshi Ohshima (Kyusyu University, Research Institute for Information Technology, Assistant Professor)
    Hisayasu Kuroda (Ehime University, Graduate School of Science and Engineering, Associate Professor)
    Tatsuya Yokota (Nagoya Institute of Technology, Graduate School of Engineering, Assistant Professor)
    Hidekata Hontani (Nagoya Institute of Technology, Graduate School of Engineering, Professor)

Overview

Models of Multidisciplinary Computational Anatomy (MCA) require file access of several TB per time, hence aggregated size of data reaches several hundreds of TB. This is enormous size of data. Computational complexity is also dramatically increasing according to size of data. As end, it is expected that commodity computers, such as PCs, cannot process it in near future. On the other hand, hardware technologies of high performance computing (HPC), such as the K-computer, are rapidly progressing. In particular, progress of system software for high-performance I/O, parallelization, and HPC, is also progressing. Hence adaptation of the HPC technologies to processes of MCA is one of candidates to make a breakthrough of computation time. In this proposal, we try to create a new technology of algorithm and implementation by collaborating HPC and MCA researches. By utilizing state-of-the-art parallelizations and high-performance implementations for advanced supercomputers, we create large-scale and high-performance technology for MCA, which is also adaptable for next generation exa-scale environment.

Project Design

A02-KB104-E.png