Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Shimokawabe, Takashi*; Endo, Toshio*; Onodera, Naoyuki; Aoki, Takayuki*
Proceedings of 2017 IEEE International Conference on Cluster Computing (IEEE Cluster 2017) (Internet), p.525 - 529, 2017/09
Stencil-based applications such as CFD have succeeded in obtaining high performance on GPU supercomputers. The problem sizes of these applications are limited by the GPU device memory capacity, which is typically smaller than the host memory. On GPU supercomputers, a locality improvement technique using temporal blocking method with memory swapping between host and device enables large computation beyond the device memory capacity. Our high-productivity stencil framework automatically applies temporal blocking to boundary exchange required for stencil computation and supports automatic memory swapping provided by a MPI/CUDA wrapper library. The framework-based application for the airflow in an urban city maintains 80% performance even with the twice larger than the GPU memory capacity and have demonstrated good weak scalability on the TSUBAME 2.5 supercomputer.
Inamura, Yasuhiro; Ito, Takayoshi*; Suzuki, Jiro*; Nakatani, Takeshi
JAEA-Testing 2016-001, 91 Pages, 2016/06
JAEA-Testing-2016-001.pdf:2.38MB
Manyo Library is a software framework for developing analysis software of neutron scattering data produced at MLF, J-PARC. This software framework is required to work on many instruments in MLF and to include base functions applied to various scientific purposes at beam lines. This framework mainly consists of data containers, which enable to store 1, 2 and 3 dimensional axes data for neutron scattering. Data containers have many functions to calculate four arithmetic operations with errors distribution between containers, to store the meta-data about measurements and to read or write text file. Since Manyo Library is built in C++ language, we' ve introduced the technology to call C++ function from Python environment into the framework. As results, we have already developed a lot of software for data reduction, analysis and visualization, which are utilized widely in beam lines at MLF. This document is the manual for the beginner to touch this framework.
*; *
Nihon Keisan Kogakkai Rombunshu, 1, p.27 - 33, 1999/00
no abstracts in English