annotate paper/fft_benchmark.tex @ 2:b7c8a956c10b

write benchmark and conclusion
author Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
date Wed, 06 Nov 2013 01:16:42 +0900
parents 9e88a388ec83
children 208ca954df56
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
1 \section{Benchmark}\label{fft_benchmark}
2
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
2 続いて,フーリエ変換と周波数フィルタによる画像処理を行う例題を用いて benchmark を行った。
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
3 512*512 の画像を High Pass Filter で変換する例題である。
0
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
4
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
5 実験環境
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
6 \begin{itemize}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
7 \item OS : MacOS 10.9
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
8 \item CPU : 2*2.66GHz 6-CoreIntel Xeon
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
9 \item Memory : 16GB
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
10 \item Compiler : Apple clang version 4.1 (based on LLVM 3.1svn)
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
11 \item GPU : AMD ATI Radeon HD 5870 1024MB
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
12 \end{itemize}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
13
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
14 \begin{tiny}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
15 \begin{table}[h]
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
16 \begin{center}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
17 \caption{Run Time}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
18 \label{table:fft_runtime}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
19 \small
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
20 \begin{tabular}[t]{c||r}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
21 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
22 &Run Time \\
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
23 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
24 1 CPU&453 ms \\
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
25 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
26 2 CPU&261 ms \\
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
27 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
28 4 CPU&178 ms \\
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
29 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
30 8 CPU&117 ms \\
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
31 \hline
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
32 \end{tabular}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
33 \end{center}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
34 \end{table}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
35 \end{tiny}
9e88a388ec83 first commit
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents:
diff changeset
36
2
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
37 表\ref{table:fft_runtime}は使用する CPU のコア数を変更し比較を行った。
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
38 1 CPU を利用した場合と比較して,2 CPU では約 1.7 倍,GPU では約 4.8 倍の速度向上が見られる。
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
39 しかしながら,8 CPU を利用した場合,4 CPU を利用した場合と比較して速度はあがっているが速度上昇率は約 1.5 倍に落ちている。
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
40 これはアムダールの法則から,並列化率が低いために速度向上が頭打ちになっていると考えられる。
b7c8a956c10b write benchmark and conclusion
Shohei KOKUBO <e105744@ie.u-ryukyu.ac.jp>
parents: 0
diff changeset
41 並列化率が低いのは,iterate で登録された Task が終了されるまで次の Task を実行することが出来ず,表\ref{table:data_parallel_index}のような index 割り当てだと Task の終了時間にばらつきが出て CPU の Utilization が低くなってることが考えられる。