GPGPUを用いたPFM解析の高速化

DOI Web Site

書誌事項

タイトル別名
  • Accelerating PFM analysis using GPGPU

抄録

Various things are replaced by simulation, but processing speed becomes the problem now. One of the technique to solve this is a parallel calculation. There is the thing which converts used GPU into image processing, and performs it that I prepare a CPU taking a normal step for a parallel calculation and perform it. Above all, the GPGPU parallel calculation using GPU has difficult making, but it is cheaper and can make an efficient parallel operation machine. I aimed at the speedup of the parallel calculation by GPGPU as a model by PFM analysis during the simulation in this article. It is the calculation of the matrix and the PFM analysis of the semiellipse surface crack model for the analysis that I performed in an article. I prepared for a program of the sequential processing to use only the CPU for and a parallel operation program to use including GPGPU and compared these. As a result, I was able to handle GPGPU program with the speed of about 12 times by about 65 times, PFM analysis by a matrix calculation. The parallel operation using GPGPU was found to function effectively from this result. I will inspect the versatility of the GPGPU parallel operation machine in future by speeding up general-purpose PFM analysis in a model.

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ