Tamagawa University Academic Repository >
00. 大学 >
04. 工学部紀要 >

このアイテムを引用あるいはリンクする場合は次の識別子を使用してください: http://hdl.handle.net/11078/46

タイトル: GPUを用いた並列共用スーパーコンピュータの構築
その他のタイトル: Shared Parallel GPU Supercomputer Construction
著者: 白﨑, 博公
キーワード: Supercomputer
Parallel computing
Linux server
Cent OS
発行日: 2013-03-31
抄録: In this study, the parallel GPU supercomputer is constructed. Using the GPU, the high-performance computing (HPC) system can be comparatively constructed cheaply and the calculation performance per energy consumption (Green) becomes better or acceptable. Here, we construct the programming language CUDA (Compute Unified Device Architecture) environment for the NVIDIA GPU. The computer is composed of two Intel Xeon (6 cores) CPUs and four Nvidia Tesla C2070 (448 CUDA cores) GPUs. The OS is 64 bit Linux (Cent OS) which makes it possible to access remotely from other PCs. We install the MATLAB, Open ACC and so on for use with the GPU. Next, we calculate the electromagnetic field by the FDTD method, and show that the calculation time becomes more than 20 times.
URI: http://hdl.handle.net/11078/46
ISSN: 0371-5981
出現コレクション:04. 工学部紀要


ファイル 記述 サイズフォーマット
3_2013_43-49.pdf4.39 MBAdobe PDF見る/開く



Valid XHTML 1.0! DSpace Software Copyright © 2002-2010  Duraspace - ご意見をお寄せください