miller/doc/performance.html
2015-05-14 14:05:45 -04:00

217 lines
7.9 KiB
HTML

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<!-- PAGE GENERATED FROM template.html and content-for-performance.html BY poki. -->
<!-- PLEASE MAKE CHANGES THERE AND THEN RE-RUN poki. -->
<head>
<meta http-equiv="Content-type" content="text/html;charset=UTF-8"/>
<meta name="description" content="Miller documentation"/>
<meta name="viewport" content="width=device-width, initial-scale=1.0"/> <!-- mobile-friendly -->
<title> Performance </title>
<link rel="stylesheet" type="text/css" href="css/miller.css"/>
<link rel="stylesheet" type="text/css" href="css/poki-callbacks.css"/>
</head>
<!-- ================================================================ -->
<script type="text/javascript">
var gaJsHost = (("https:" == document.location.protocol) ? "https://ssl." : "http://www.");
document.write(unescape("%3Cscript src='" + gaJsHost + "google-analytics.com/ga.js' type='text/javascript'%3E%3C/script%3E"));
</script>
<script type="text/javascript">
try {
var pageTracker = _gat._getTracker("UA-15651652-1");
pageTracker._trackPageview();
} catch(err) {}
</script>
<script type="text/javascript">
function toggle(divName) {
var eleDiv = document.getElementById(divName);
if (eleDiv != null) {
if (eleDiv.style.display == "block") {
eleDiv.style.display = "none";
} else {
eleDiv.style.display = "block";
}
}
}
</script>
<!--
The background image is from a screenshot of a Google search for "data analysis
tools", lightened and sepia-toned. Over this was placed a Mac Terminal app with
very light-grey font and translucent background, in which a few statistical
Miller commands were run with pretty-print-tabular output format.
-->
<body background="pix/sepia-overlay.jpg">
<!-- ================================================================ -->
<table width="100%">
<tr>
<!-- navbar -->
<td width="15%">
<!--
<img src="pix/mlr.jpg" />
<img style="border-width:1px; color:black;" src="pix/mlr.jpg" />
-->
<div class="pokinav">
<center><titleinbody>Miller</titleinbody></center>
<!-- PAGE LIST GENERATED FROM template.html BY poki -->
<br/>User info:
<br/>&bull;&nbsp;<a href="index.html">About</a>
<br/>&bull;&nbsp;<a href="file-formats.html">File formats</a>
<br/>&bull;&nbsp;<a href="feature-comparison.html">Miller features in the context of the Unix toolkit</a>
<br/>&bull;&nbsp;<a href="record-heterogeneity.html">Record-heterogeneity</a>
<br/>&bull;&nbsp;<a href="performance.html"><b>Performance</b></a>
<br/>&bull;&nbsp;<a href="etymology.html">Why call it Miller?</a>
<br/>&bull;&nbsp;<a href="originality.html">How original is Miller?</a>
<br/>&bull;&nbsp;<a href="reference.html">Reference</a>
<br/>&bull;&nbsp;<a href="data-examples.html">Data examples</a>
<br/>&bull;&nbsp;<a href="to-do.html">Things to do</a>
<br/>Developer info:
<br/>&bull;&nbsp;<a href="build.html">Compiling, portability, dependencies, and testing</a>
<br/>&bull;&nbsp;<a href="whyc.html">Why C?</a>
<br/>&bull;&nbsp;<a href="contact.html">Contact information</a>
<br/>&bull;&nbsp;<a href="https://github.com/johnkerl/miller">GitHub repo</a>
<br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/>
<br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/> <br/>
<br/> <br/> <br/> <br/> <br/> <br/>
</div>
</td>
<!-- page body -->
<td>
<div style="overflow-y:scroll;height:1500px">
<center> <titleinbody> Performance </titleinbody> </center>
<p/>
<!-- BODY COPIED FROM content-for-performance.html BY poki -->
<div class="pokitoc">
<center><b>Contents:</b></center>
&bull;&nbsp;<a href="#Data">Data</a><br/>
&bull;&nbsp;<a href="#Comparands">Comparands</a><br/>
&bull;&nbsp;<a href="#Raw_results">Raw results</a><br/>
&bull;&nbsp;<a href="#Analysis">Analysis</a><br/>
&bull;&nbsp;<a href="#Conclusion">Conclusion</a><br/>
</div>
<p/>
<a id="Data"/><h1>Data</h1>
Test data were of the form
<table><tr><td>
<p/>
<div class="pokipanel">
<pre>
a=pan,b=pan,i=1,x=0.3467901443380824,y=0.7268028627434533
a=eks,b=pan,i=2,x=0.7586799647899636,y=0.5221511083334797
a=wye,b=wye,i=3,x=0.20460330576630303,y=0.33831852551664776
a=eks,b=wye,i=4,x=0.38139939387114097,y=0.13418874328430463
a=wye,b=pan,i=5,x=0.5732889198020006,y=0.8636244699032729
</pre>
</div>
<p/>
</td><td>
<p/>
<div class="pokipanel">
<pre>
a,b,i,x,y
pan,pan,1,0.3467901443380824,0.7268028627434533
eks,pan,2,0.7586799647899636,0.5221511083334797
wye,wye,3,0.20460330576630303,0.33831852551664776
eks,wye,4,0.38139939387114097,0.13418874328430463
wye,pan,5,0.5732889198020006,0.8636244699032729
</pre>
</div>
<p/>
</td></tr></table>
for DKVP and CSV, respectively, where fields <tt>a</tt> and <tt>b</tt> take one of five text values,
uniformly distributed; <tt>i</tt> is a 1-up line counter; <tt>x</tt> and <tt>y</tt>
are independent uniformly distributed floating-point numbers in the unit
interval.
<p>Data files of one million lines (totalling about 50MB for CSV and 60MB for
DKVP) were used. In experiments not shown here, I also varied the file sizes;
the size-dependent results were the expected, completely unsurprising
linearities and so I produced no file-size-dependent plots for your viewing pleasure.
<a id="Comparands"/><h1>Comparands</h1>
The <tt>cat</tt>, <tt>cut</tt>, <tt>awk</tt>, <tt>sed</tt>, <tt>sort</tt> tools
were compared to <tt>mlr</tt> on an 8-core Darwin laptop; RAM capacity was
nowhere near challenged . The <tt>catc</tt> program is a simple line-oriented
line-printer (<a
href="https://github.com/johnkerl/miller/blob/master/c/tools/catc.c">source
here</a>) which is intermediate between Miller (which is record-aware as well
as line-aware) and <tt>cat</tt> (which is only byte-aware).
<a id="Raw_results"/><h1>Raw results</h1>
Note that for CSV data, the command is <tt>mlr --csv ...</tt> rather than <tt>mlr ...</tt>.
<p/>
<div class="pokipanel">
<pre>
Mac Mac Comparand
DKVP CSV
seconds seconds
0.016 0.013 cat
0.189 0.189 catc
3.657 4.388 awk -F, '{print}'
2.027 1.795 mlr cat
2.292 1.940 cut -d , -f 1,4
3.540 4.516 awk -F, '{print $1,$4}'
1.600 1.390 mlr cut -f a,x
1.694 1.648 mlr cut -x -f a,x
0.845 0.643 sed -e 's/x=/EKS=/' -e 's/b=/BEE=/'
2.076 1.842 mlr rename x,EKS,b,BEE
5.643 5.031 awk -F, '{gsub("x=","",$4);gsub("y=","",$5);print $4+$5}'
4.019 3.679 mlr put '$z=$x+$y'
2.481 2.628 mlr stats1 -a mean -f x,y -g a,b
2.587 2.389 mlr stats2 -a corr -f x,y -g a,b
23.247 14.466 sort -t, -k 1,2
3.023 5.658 mlr sort -f a,b
17.224 15.523 sort -t, -k 4,5
5.807 5.194 mlr sort -n x,y
</pre>
</div>
<p/>
<a id="Analysis"/><h1>Analysis</h1>
<ul>
<li/> As expected, <tt>cat</tt> is very fast &mdash; it needs only stream bytes as quickly as possible; it doesn&rsquo;t even need to touch individual bytes.
<li/> My <tt>catc</tt> is also faster than Miller: it needs to read and write lines, but it doesn&rsquo;t segment lines into records; in fact it does no iteration over bytes in each line.
<li/> Miller does not outperform <tt>sed</tt>, which is string-oriented rather than record-oriented.
<li/> For the tools which do need to pick apart fields (<tt>cut</tt>,
<tt>awk</tt>, <tt>sort</tt>), Miller is comparable or outperforms. As noted above, this effect
persists linearly across file sizes.
<li/> For univariate and bivariate statistics, I didn&rsquo;t attempt to
compare to other tools wherein such computations are less straightforward;
rather, I attempted only to show that Miller&rsquo;s processing time here is comparable to its own processing time for other problems.
</ul>
<a id="Conclusion"/><h1>Conclusion</h1>
For record-oriented data transformations, Miller meets or beats the Unix
toolkit in many contexts. Field renames in particular are worth doing as a
pre-pipe or post-pipe using <tt>sed</tt>.
</div>
</td>
</table>
</body>
</html>