• 検索結果がありません。

GPUのメモリ階層に対する最適化技術の考察

N/A
N/A
Protected

Academic year: 2021

シェア "GPUのメモリ階層に対する最適化技術の考察"

Copied!
1
0
0

読み込み中.... (全文を見る)

全文

(1)情報処理学会論文誌. プログラミング. Vol.8 No.2 3 (July 2015). 発表概要. GPU のメモリ階層に対する最適化技術の考察 窪田 昌史1,a) 2015年1月13日発表. GPU は,CUDA や OpenCL などの既存の C,C++,Fortran 言語をベースとしたプログラミング環境 を用いて高い性能が得られることから,HPC 分野で急速に普及してきた.しかし,性能チューニングの際 に,並列性だけでなく GPU の複雑なメモリ階層をも考慮しなくてはならないことが,GPU プログラミン グの問題点として指摘されている.本発表では,配列変数のアクセス順序の変更,変数の生存区間解析と いったプログラムの最適化技術が,GPU プログラミングにおいて活用できることを,実例を交えて報告 する.. A Study on Optimization Techniques for Memory Hierarchy of GPUs Atsushi Kubota1,a) Presented: January 13, 2015. GPU has been widely accepted in HPC because high performance can be attained by programming environments such as CUDA and OpenCL, which are based on conventional programming languages including C, C++ and Fortran. However, it is pointed out that GPU programming is difficult because complex memory hierarchy of GPUs as well as parallelism must be taken into account for performance tuning. We present optimization techniques for programs such as changes of order of accessing array elements and live range analysis for variables can be utilized in GPU programming and demonstrate performance tuning for several examples.. 1 a). 広島市立大学 Hiroshima City University [email protected]. c 2015 Information Processing Society of Japan . 3.

(2)

参照

関連したドキュメント

Using an “energy approach” introduced by Bronsard and Kohn [11] to study slow motion for Allen-Cahn equation and improved by Grant [25] in the study of Cahn-Morral systems, we

Based on the asymptotic expressions of the fundamental solutions of 1.1 and the asymptotic formulas for eigenvalues of the boundary-value problem 1.1, 1.2 up to order Os −5 ,

, 6, then L(7) 6= 0; the origin is a fine focus of maximum order seven, at most seven small amplitude limit cycles can be bifurcated from the origin.. Sufficient

In order to achieve the minimum of the lowest eigenvalue under a total mass constraint, the Stieltjes extension of the problem is necessary.. Section 3 gives two discrete examples

0.1. Additive Galois modules and especially the ring of integers of local fields are considered from different viewpoints. Leopoldt [L] the ring of integers is studied as a module

Using variational techniques we prove an eigenvalue theorem for a stationary p(x)-Kirchhoff problem, and provide an estimate for the range of such eigenvalues1. We employ a

This problem becomes more interesting in the case of a fractional differential equation where it closely resembles a boundary value problem, in the sense that the initial value

In this article we prove the following result: if two 2-dimensional 2-homogeneous rational vector fields commute, then either both vector fields can be explicitly integrated to