Sustainable Adaptive Grid Supercomputing: Multiscale Simulation of Semiconductor Processing across the Pacific
Session: Grid Applications
Event Type: Paper
Date: Wednesday, Nov 15
Time: 2:00pm - 2:30pm
Session Chair: Thilo Kielmann
Author(s): Hiroshi Takemiya, Yoshio Tanaka, Satoshi Sekiguchi, Shuji Ogata, Rajiv K. Kalia, Aiichiro Nakano, Priya Vashishta
Location: 20-21
Abstract:
We propose a reservation-based sustainable adaptive Grid supercomputing paradigm for tightly coupled computations of considerable scale and duration on a Grid. The paradigm is demonstrated for an adaptive multiscale simulation, in which accurate but compute-intensive QM simulations are embedded within a classical MD simulation only when and where high fidelity is required. Key technical innovations include an embedded divide-and-conquer algorithmic framework to maximally expose data and computation localities for scalability, a buffered-cluster hybridization scheme to adaptively adjust MD/QM boundaries for accuracy, and a hybrid GridRPC+MPI programming approach to combine flexibility, fault tolerance, and efficiency. We have achieved an automated execution of the simulation on a Pan-Pacific Grid in total of 150,000 processor-hours, in which the number of processors changes on demand and resources are allocated/migrated according to both reservations and unexpected faults. The simulation results of SIMOX reveal a strong dependence of the oxygen penetration depth on the incident position.
Thilo Kielmann (Chair)
Vrije Universiteit
Hiroshi Takemiya
National Institute of Advanced Industrial Science and Technology
Yoshio Tanaka
National Institute of Advanced Industrial Science and Technology
Satoshi Sekiguchi
National Institute of Advanced Industrial Science and Technology
Shuji Ogata
Nagoya Institute of Technology
Rajiv K. Kalia
University of Southern California
Aiichiro Nakano
University of Southern California
Priya Vashishta
University of Southern California