Changeset 4475 for docs


Ignore:
Timestamp:
Feb 7, 2015, 12:50:46 AM (4 years ago)
Author:
cameron
Message:

Discussion of performance results

File:
1 edited

Legend:

Unmodified
Added
Removed
  • docs/Working/icGrep/evaluation.tex

    r4474 r4475  
    3737most of the world's major language families as a test corpus.   For each program
    3838under test, we perform searches for each regular expression against each XML document.
    39 Searches were repeated n times.  Table \ref{tbl:property_test} shows the results.
    40 
    41 \begin{table}
    42 \input{table-prop.tex}
    43 \caption{Performance of Matching Property and Property Combinations}\label{tbl:property_test}
    44 \end{table}
     39Results are presented in \Figure \ref{fig:property_test}.  Performance is reported
     40in CPU cycles per byte on an Intel Core i7 machine.   The results were grouped
     41by the percentage of matching lines found in the XML document, grouped in
     425\% increments.  ICgrep shows dramatically better performance, particularly
     43when searching for rare items.
     44As shown in the figure, pcre2grep and ugrep both show
     45increased performance (reduced CPU cycles per byte) with increasing percentage
     46of matches found.  In essence, each match found allows these programs
     47to skip the full processing of the rest of the line.   On the other
     48hand, icGrep shows a slight drop-off in performance with the number
     49of matches found.   This is primarily due to property classes that
     50include large numbers of codepoints.   These classes require more
     51bitstream equations for calculation and also have a greater probability
     52of matching.   Nevertheless, the performance of icGrep in matching
     53the defined property expressions is stable and well ahead of the competitors
     54in all cases.
    4555
    4656\begin{figure}
     57\begin{center}
    4758\pgfplotstableread[col sep = comma]{data/icgrep-scatter.csv}\icgrep
    4859\pgfplotstableread[col sep = comma]{data/ugrep541-scatter.csv}\ugrep
     
    5465x tick label style={ /pgf/number format/1000 sep=},
    5566% x buffer=sort,
    56 ylabel={Cycles Per Byte},
    57 xlabel={Match Percentage},
     67ylabel={CPU Cycles Per Byte},
     68xlabel={Percentage of Matching Lines},
    5869minor y tick num={1},
    5970xmax=100
     
    6879
    6980\end{tikzpicture}
    70 
     81\end{center}
     82\caption{Comparative Matching Performance}\label{fig:property_test}
    7183\end{figure}
    7284
Note: See TracChangeset for help on using the changeset viewer.