Nanomaterial applications for computational Grid

1
Nano-material applications for computational Grid
Hongsuk Yi Supercomputing Center KISTI
2
Objective of 2003

• Installation of Nano-material application (VASP)
  program on Unix/Linux Clusters
• Installation of Globus, MPICH-G2, Open PBS on 16
  CPU cluster
• Installation of VASP, Intel F90 Compiler, PGF90
  compiler
• Development of Globus linked application programs
• Test of parallel versions of VASP
• Implement New Materials Design Systems based on
  High Throughput Computing Grid computations
• Collaboration with PSE research team sharing R
  D informations
• Development of HTC Grid Web Portal Site

3
Scope of Nano-material Grid Project

• Construction of computational Nano-material
  systems for computational Grid
• Demonstrate HTC (High Throughput Computing)
  capability on various computing systems including
  KISTI Grid Testbed
• Scalability study for the HPC (High Performance
  Computing) performance with the KISTI Grid
  Testbed
• Study of New Physical Phenomena in nano-scale.
• Improve R D efficiency in the Grid system
• Developing programs for the job distribution in
  grid
• Construction of new problem solving environment

4
Road map of Nano-material computational Grid

• Nano-Simulation from HPC to HTC Grid

System Size
5
Grid Activities
6
High Performance Comuting Challenges

• Potential Energy Surface of Semiconductors
• Improve the accuracy by combining ab initio
  simulation methods (VASP)
• HPC Challenge problems setting and Performance
  evaluation using the GridRPC and/or Pacx-MPI
  parallelization methods

7
High Performance Computing

• VASP requires a very fast network speed
• Change from HPC to HTC

8
Potential Energy Surface of K/InAs(110) surface
9
Partitioning Applicationfor Grid Computing
High-Throughput Computing on KISTI Grid Testbed