<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="generator" content="Docutils 0.4: http://docutils.sourceforge.net/" />
<title>GraphicsMagick 1.2 vs ImageMagick 6.3.6 Benchmark Report</title>
<style type="text/css">
/* magick.css -- default style sheet */
BODY, P, TD, TH, UL, OL {
font-family : Verdana, Arial, Helvetica, sans-serif;
font-size : medium;
}
H1, H2, H3, H4 {
font-family : Verdana, Arial, Helvetica, sans-serif;
}
H1 {
font-size : 120%;
}
H2 {
font-size : 110%;
}
H3, H4 {
font-size : 105%;
}
A:LINK {
text-decoration : underline;
color : #0085C0;
}
A:VISITED {
text-decoration : underline;
color : Purple;
}
A:HOVER {
text-decoration : underline;
color : #0085C0;
}
A.nav:LINK {
text-decoration : none;
color : #0085C0;
}
A.nav:VISITED {
text-decoration : none;
color : Purple;
}
A.nav:HOVER {
text-decoration : underline;
color : #0085C0;
}
A.leftcolnav:LINK {
text-decoration : none;
color : #0085C0;
}
A.leftcolnav:HOVER {
text-decoration : none;
background : #999999;
color: black;
}
A.leftcolnav:VISITED {
text-decoration : none;
color : Purple;
}
.stripes {
font-size : smaller;
}
.navtopon {
background-color : #999999;
font-size : smaller;
text-align : center;
font-weight : bold;
}
.navtopoff {
font-size : smaller;
text-align : center;
font-weight : bold;
}
.gutter {
margin-top : 2mm;
margin-right : 3mm;
margin-bottom : 3mm;
margin-left : 5mm;
}
.bannergutter {
margin-top : 2mm;
margin-right : 3mm;
}
.pagebodygutter {
margin-top : 2mm;
}
.pagecentergutter {
margin-left : 20mm;
margin-right: 20mm;
margin-top : 5mm;
margin-bottom: 2mm;
}
.leftcolgutter {
margin-top : 2mm;
margin-right : 1mm;
margin-left : 3mm;
}
.textcolgutter {
margin-bottom : 1.5mm;
}
.textpromo {
font-size : medium;
margin-top : 2mm;
margin-right : 1mm;
margin-left : 2mm;
}
.newscolgutter {
margin-left : 1mm;
margin-top : 2mm;
margin-bottom : 3mm;
font-size : medium;
}
.itemgutter {
margin-left : 1.5mm;
font-size : small;
}
.logogutter {
margin-top : 3mm;
margin-bottom : 2mm;
margin-left : 3mm;
}
.navtext {
font-size : smaller;
}
.navlevel2 {
font-size : 9pt;
}
.navlevel3 {
font-size : 8pt;
}
.navlevel2on {
font-size : 9pt;
background-color : #cccccc;
}
.navlevel3on {
font-size : 8pt;
background-color : #cccccc;
}
.navlevel1on {
background-color : #cccccc;
}
.note {
font-style : italic;
color : gray;
}
.textwhite {
color : white;
}
.small {
font-size : smaller;
}
.tiny {
font-size : smaller;
}
.confirmation {
color : Lime;
}
.warning {
color : #ffcc66;
}
.error {
color : Red;
}
.disclaimer {
font-size : smaller;
color : Gray;
margin-top : 1em;
margin-bottom : 1em;
}
.bgwhite {
background-color: White;
}
.1colgutter {
margin-top: 2mm;
margin-right: 3mm;
margin-left: 3mm;
}
.1colnavgutter {
margin-right: 1mm;
margin-left: 1mm;
}
.largeboldfont {
font-size : 110%;
font-weight: bold;
}
.redfont {
color : red;
}
</style>
</head>
<body>
<div class="document" id="graphicsmagick-1-2-vs-imagemagick-6-3-6-benchmark-report">
<h1 class="title">GraphicsMagick 1.2 vs ImageMagick 6.3.6 Benchmark Report</h1>
<!-- This text is in reStucturedText format, so it may look a bit odd. -->
<!-- See http://docutils.sourceforge.net/rst.html for details. -->
<div class="contents local topic">
<ul class="simple">
<li><a class="reference" href="#purpose" id="id1" name="id1">Purpose</a></li>
<li><a class="reference" href="#executive-summary" id="id2" name="id2">Executive Summary</a></li>
<li><a class="reference" href="#input-files" id="id3" name="id3">Input Files</a></li>
<li><a class="reference" href="#file-conversion-results" id="id4" name="id4">File Conversion Results</a><ul>
<li><a class="reference" href="#bmp-format" id="id5" name="id5">BMP Format</a></li>
<li><a class="reference" href="#dpx-format" id="id6" name="id6">DPX Format</a></li>
<li><a class="reference" href="#gif-format" id="id7" name="id7">GIF Format</a></li>
<li><a class="reference" href="#jpeg-format" id="id8" name="id8">JPEG Format</a></li>
<li><a class="reference" href="#pcx-format" id="id9" name="id9">PCX Format</a></li>
<li><a class="reference" href="#pdf-format" id="id10" name="id10">PDF Format</a></li>
<li><a class="reference" href="#pnm-format" id="id11" name="id11">PNM Format</a></li>
<li><a class="reference" href="#postscript-format" id="id12" name="id12">Postscript Format</a></li>
<li><a class="reference" href="#sgi-format" id="id13" name="id13">SGI Format</a></li>
<li><a class="reference" href="#tiff-format" id="id14" name="id14">TIFF Format</a></li>
</ul>
</li>
<li><a class="reference" href="#image-processing-results" id="id15" name="id15">Image Processing Results</a><ul>
<li><a class="reference" href="#blur-0-0x1-0-up-to-162-faster" id="id16" name="id16">-blur 0.0x1.0 (up to 162% faster)</a></li>
<li><a class="reference" href="#gamma-1-2-up-to-219-faster" id="id17" name="id17">-gamma 1.2 (up to 219% faster)</a></li>
<li><a class="reference" href="#resize-50-only-34-faster" id="id18" name="id18">-resize 50% (only 34% faster)</a></li>
<li><a class="reference" href="#rotate-90-up-to-134-faster" id="id19" name="id19">-rotate 90 (up to 134% faster)</a></li>
</ul>
</li>
<li><a class="reference" href="#execution-overhead-results-3-2x-less-overhead" id="id20" name="id20">Execution Overhead Results (3.2X less overhead!)</a></li>
</ul>
</div>
<p>GraphicsMagick 1.2 has been benchmarked against the most recent
ImageMagick delivered by the FreeBSD Ports system at that time
(ImageMagick 6.3.6). Testing was done on a dual-CPU Intel Xeon 2.4GHz
system, with 2GB RAM and running FreeBSD 7.0. The software involved is
configured and compiled using similar options. The software involved in
the testing identified itself as:</p>
<pre class="literal-block">
GraphicsMagick 1.2.1 2008-05-06 Q16 http://www.GraphicsMagick.org/
ImageMagick 6.3.6 03/30/08 Q16 http://www.imagemagick.org
</pre>
<div class="section">
<h1><a class="toc-backref" href="#id1" id="purpose" name="purpose">Purpose</a></h1>
<p>The purpose of the benchmark was to measure the raw throughput of the
sofware with medium sized (600 DPI) files. We consider 600 DPI to be
"medium" since today's printers typically support 1200 or 2400 DPI. Raw
throughput is important since testing showed that reading and writing
files often dominates image processing time. We threw in a little bit of
image processing as well. The benchmark uses Ghostscript and the
'golfer.ps' and 'tiger.ps' sample files to render test input TIFF files
at 600 DPI.</p>
<p>The benchmark is available via anonymous ftp so that you can try it on
your own computer. Retrieve it from
"<a class="reference" href="ftp://ftp.graphicsmagick.org/pub/GraphicsMagick/benchmark/magick-bench.tar.gz">ftp://ftp.graphicsmagick.org/pub/GraphicsMagick/benchmark/magick-bench.tar.gz</a>".</p>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id2" id="executive-summary" name="executive-summary">Executive Summary</a></h1>
<p>GraphicsMagick is often 2X to 200X faster at converting from one file to
another than ImageMagick 6.3.6 is on garden variety hardware. Typical
image processing tasks are faster as well. If time and money are
important to you, then you should consider using GraphicsMagick rather
than ImageMagick. If time and money are not important to you, then you
should continue using ImageMagick.</p>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id3" id="input-files" name="input-files">Input Files</a></h1>
<p>The following input files are used (alphabetical order):</p>
<table border="1" class="docutils">
<colgroup>
<col width="41%" />
<col width="59%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head">Name</th>
<th class="head">Description</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>Black/white (8 bits/sample) in "RGB" format</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>Black/white (1 bit/sample) in bilevel format</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>Grayscale (8 bits/sample)in "RGB" format</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>Grayscale (8 bits/sample) in compact format</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>Graycale (16 bits/sample) in a compact format.</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>Grayscale, indexed (4 bits/sample)</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>Black/white, indexed (1 bit/sample)</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>Color (RGB) indexed (8 bits/sample)</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>Color (RGB) truecolor (8 bits/sample)</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>Color (RGB) truecolor (16 bits/sample)</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>Color (RGBA) truecolor (8 bits/sample)</td>
</tr>
</tbody>
</table>
<p>The 'golfer-600' files have pixel dimensions of 4750x5792, and the
'tiger-600' files have pixel dimensions of 4542x4725.</p>
<p>The benchmark is run via a simple script which runs a 'convert' command to convert the
input file to an output file with the specified format.</p>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id4" id="file-conversion-results" name="file-conversion-results">File Conversion Results</a></h1>
<div class="section">
<h2><a class="toc-backref" href="#id5" id="bmp-format" name="bmp-format">BMP Format</a></h2>
<p>BMP is a popular format on Windows systems. Since it is uncompressed, it
should be fast. GraphicsMagick is clearly much faster at writing BMP than
ImageMagick is.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>0.66s</td>
<td>0.88s</td>
<td>1.30s</td>
<td>3.11s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.46s</td>
<td>0.59s</td>
<td>0.90s</td>
<td>1.07s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>0.61s</td>
<td>0.93s</td>
<td>1.33s</td>
<td>3.05s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.60s</td>
<td>0.89s</td>
<td>1.13s</td>
<td>3.02s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.66s</td>
<td>0.85s</td>
<td>1.19s</td>
<td>2.93s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.64s</td>
<td>0.49s</td>
<td>0.62s</td>
<td>1.50s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.65s</td>
<td>0.42s</td>
<td>0.68s</td>
<td>0.76s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.50s</td>
<td>0.45s</td>
<td>0.75s</td>
<td>2.29s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.43s</td>
<td>0.78s</td>
<td>0.98s</td>
<td>2.52s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.58s</td>
<td>0.84s</td>
<td>1.08s</td>
<td>2.65s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>0.62s</td>
<td>0.84s</td>
<td>1.14s</td>
<td>2.48s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id6" id="dpx-format" name="dpx-format">DPX Format</a></h2>
<p>DPX format is used by the motion picture industry. It is rarely used
outside that industry. Since a typical motion picture contains 144K image
frames (and millions of frames may be processed), processing speed is
quite important. Even though GraphicsMagick's DPX implementation
prioritizes flexibility and correctness over speed, it compares well with
ImageMagick.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>0.87s</td>
<td>0.67s</td>
<td>1.32s</td>
<td>2.87s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.90s</td>
<td>0.74s</td>
<td>5.24s</td>
<td>0.61s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>0.88s</td>
<td>0.68s</td>
<td>1.22s</td>
<td>3.04s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.84s</td>
<td>0.65s</td>
<td>0.91s</td>
<td>3.05s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.79s</td>
<td>1.03s</td>
<td>0.86s</td>
<td>3.21s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.89s</td>
<td>1.05s</td>
<td>5.21s</td>
<td>1.45s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.91s</td>
<td>0.75s</td>
<td>5.40s</td>
<td>0.80s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.65s</td>
<td>0.89s</td>
<td>1.05s</td>
<td>2.46s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.54s</td>
<td>0.68s</td>
<td>0.92s</td>
<td>2.44s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.67s</td>
<td>1.00s</td>
<td>0.91s</td>
<td>2.75s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>0.69s</td>
<td>0.73s</td>
<td>0.89s</td>
<td>2.33s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id7" id="gif-format" name="gif-format">GIF Format</a></h2>
<p>GIF format is quite popular on the Internet. It seems that GraphicsMagick
is sometimes 3X to 50X faster than ImageMagick at outputting GIF files.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.19s</td>
<td>0.63s</td>
<td>41.21s</td>
<td>3.40s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.88s</td>
<td>0.58s</td>
<td>45.67s</td>
<td>1.26s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.14s</td>
<td>0.60s</td>
<td>50.16s</td>
<td>3.32s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.96s</td>
<td>0.60s</td>
<td>52.96s</td>
<td>3.35s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>1.06s</td>
<td>0.52s</td>
<td>52.99s</td>
<td>3.33s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.97s</td>
<td>0.48s</td>
<td>1.02s</td>
<td>1.43s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.94s</td>
<td>0.51s</td>
<td>1.09s</td>
<td>0.72s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.72s</td>
<td>0.40s</td>
<td>0.93s</td>
<td>2.30s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>12.75s</td>
<td>0.48s</td>
<td>34.63s</td>
<td>2.84s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>12.70s</td>
<td>0.66s</td>
<td>34.96s</td>
<td>2.78s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>12.57s</td>
<td>0.50s</td>
<td>73.41s</td>
<td>3.12s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id8" id="jpeg-format" name="jpeg-format">JPEG Format</a></h2>
<p>JPEG is a quite popular format for still photos and on the Internet.
GraphicsMagick writes JPEG files faster than ImageMagick.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>2.26s</td>
<td>0.61s</td>
<td>3.73s</td>
<td>2.74s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>2.16s</td>
<td>0.65s</td>
<td>3.41s</td>
<td>0.72s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.67s</td>
<td>0.59s</td>
<td>2.80s</td>
<td>2.81s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>1.37s</td>
<td>0.66s</td>
<td>2.70s</td>
<td>2.73s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>1.48s</td>
<td>0.56s</td>
<td>2.77s</td>
<td>2.64s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>1.49s</td>
<td>0.71s</td>
<td>2.63s</td>
<td>1.49s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>2.15s</td>
<td>0.66s</td>
<td>3.53s</td>
<td>0.81s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>2.21s</td>
<td>0.52s</td>
<td>2.91s</td>
<td>2.48s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>2.14s</td>
<td>0.47s</td>
<td>2.93s</td>
<td>2.31s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>2.12s</td>
<td>0.72s</td>
<td>3.02s</td>
<td>2.50s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>2.24s</td>
<td>0.55s</td>
<td>3.00s</td>
<td>2.08s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id9" id="pcx-format" name="pcx-format">PCX Format</a></h2>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.48s</td>
<td>0.47s</td>
<td>1.85s</td>
<td>2.70s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.52s</td>
<td>0.53s</td>
<td>1.58s</td>
<td>0.64s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.15s</td>
<td>0.58s</td>
<td>1.51s</td>
<td>2.73s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>1.07s</td>
<td>0.46s</td>
<td>1.49s</td>
<td>2.55s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>1.03s</td>
<td>0.52s</td>
<td>1.39s</td>
<td>2.64s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.61s</td>
<td>0.54s</td>
<td>0.71s</td>
<td>1.45s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.56s</td>
<td>0.52s</td>
<td>0.87s</td>
<td>0.75s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.44s</td>
<td>0.51s</td>
<td>0.77s</td>
<td>2.29s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.86s</td>
<td>0.42s</td>
<td>1.31s</td>
<td>2.14s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.93s</td>
<td>0.56s</td>
<td>1.50s</td>
<td>2.20s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>1.13s</td>
<td>0.53s</td>
<td>1.55s</td>
<td>1.95s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id10" id="pdf-format" name="pdf-format">PDF Format</a></h2>
<p>PDF format is important since it is used for documents and by the
pre-press industry. GraphicsMagick writes PDF files much faster than
ImageMagick.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.78s</td>
<td>0.55s</td>
<td>4.38s</td>
<td>2.84s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>1.68s</td>
<td>0.59s</td>
<td>4.07s</td>
<td>0.76s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.71s</td>
<td>0.52s</td>
<td>4.45s</td>
<td>2.76s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>1.51s</td>
<td>0.53s</td>
<td>4.25s</td>
<td>2.77s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>1.48s</td>
<td>0.58s</td>
<td>4.23s</td>
<td>2.73s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>1.69s</td>
<td>0.58s</td>
<td>4.40s</td>
<td>1.48s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>1.80s</td>
<td>0.49s</td>
<td>4.07s</td>
<td>0.80s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>1.24s</td>
<td>0.51s</td>
<td>2.85s</td>
<td>2.26s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>2.70s</td>
<td>0.43s</td>
<td>6.85s</td>
<td>2.54s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>2.68s</td>
<td>0.67s</td>
<td>7.11s</td>
<td>2.56s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>2.80s</td>
<td>0.57s</td>
<td>9.25s</td>
<td>2.51s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id11" id="pnm-format" name="pnm-format">PNM Format</a></h2>
<p>The PNM format (from pbmplus/netpbm) is designed for blazing speed since
it is often used as part of image processing pipelines. Unfortunately the
ImageMagick implementation does not offer blazing speed and is 200X
slower than GraphicsMagick.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>0.78s</td>
<td>0.42s</td>
<td>98.05s</td>
<td>3.78s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.49s</td>
<td>0.57s</td>
<td>98.00s</td>
<td>1.62s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>0.70s</td>
<td>0.57s</td>
<td>97.89s</td>
<td>3.88s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.56s</td>
<td>0.54s</td>
<td>98.11s</td>
<td>3.80s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.54s</td>
<td>0.60s</td>
<td>99.53s</td>
<td>4.37s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.63s</td>
<td>0.64s</td>
<td>98.09s</td>
<td>2.47s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.59s</td>
<td>0.50s</td>
<td>98.04s</td>
<td>1.78s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.56s</td>
<td>0.79s</td>
<td>76.47s</td>
<td>3.20s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.58s</td>
<td>0.51s</td>
<td>76.62s</td>
<td>2.95s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.55s</td>
<td>0.93s</td>
<td>77.71s</td>
<td>3.44s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>0.49s</td>
<td>0.79s</td>
<td>76.51s</td>
<td>2.98s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id12" id="postscript-format" name="postscript-format">Postscript Format</a></h2>
<p>Sending jobs to Postscript printers can be slow, and now we know why.
GraphicsMagick is 10-40X more efficient at outputting Level I Postscript
than ImageMagick is.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.06s</td>
<td>0.56s</td>
<td>10.73s</td>
<td>2.73s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.55s</td>
<td>0.57s</td>
<td>10.45s</td>
<td>0.62s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>0.83s</td>
<td>0.64s</td>
<td>75.58s</td>
<td>3.07s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.65s</td>
<td>0.62s</td>
<td>75.31s</td>
<td>2.71s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.64s</td>
<td>0.67s</td>
<td>75.23s</td>
<td>2.72s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.75s</td>
<td>0.67s</td>
<td>75.09s</td>
<td>1.52s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.63s</td>
<td>0.48s</td>
<td>10.54s</td>
<td>0.74s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.57s</td>
<td>0.52s</td>
<td>58.38s</td>
<td>2.28s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.98s</td>
<td>0.78s</td>
<td>72.56s</td>
<td>2.56s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>1.10s</td>
<td>0.89s</td>
<td>72.48s</td>
<td>2.79s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>1.20s</td>
<td>0.75s</td>
<td>72.78s</td>
<td>2.54s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id13" id="sgi-format" name="sgi-format">SGI Format</a></h2>
<p>SGI format is a legacy format which is still used for heavy-duty image
processing. GraphicsMagick is several times faster at dealing with SGI
format than ImageMagick is.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>0.99s</td>
<td>0.75s</td>
<td>2.75s</td>
<td>2.94s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>1.01s</td>
<td>0.67s</td>
<td>2.47s</td>
<td>0.85s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.01s</td>
<td>0.66s</td>
<td>2.69s</td>
<td>3.03s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.82s</td>
<td>0.67s</td>
<td>2.57s</td>
<td>2.95s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.80s</td>
<td>0.71s</td>
<td>2.40s</td>
<td>3.04s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.92s</td>
<td>0.72s</td>
<td>2.47s</td>
<td>1.65s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>1.02s</td>
<td>0.67s</td>
<td>2.54s</td>
<td>0.99s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.88s</td>
<td>0.53s</td>
<td>3.57s</td>
<td>2.70s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.77s</td>
<td>0.53s</td>
<td>3.54s</td>
<td>2.55s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.84s</td>
<td>0.68s</td>
<td>3.72s</td>
<td>3.05s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>0.90s</td>
<td>0.60s</td>
<td>4.42s</td>
<td>2.33s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id14" id="tiff-format" name="tiff-format">TIFF Format</a></h2>
<p>TIFF is one of the world's most important image file formats. Since it is
often used serious image processing, it needs to be as efficient as
possible. Clearly GraphicsMagick is much faster than ImageMagick at
processing TIFF files.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>0.74s</td>
<td>0.49s</td>
<td>1.25s</td>
<td>2.76s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.60s</td>
<td>0.46s</td>
<td>0.89s</td>
<td>0.73s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>0.65s</td>
<td>0.68s</td>
<td>1.91s</td>
<td>2.80s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.56s</td>
<td>0.58s</td>
<td>1.70s</td>
<td>2.88s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.56s</td>
<td>0.63s</td>
<td>1.32s</td>
<td>2.67s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.49s</td>
<td>0.61s</td>
<td>1.55s</td>
<td>1.45s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>0.58s</td>
<td>0.50s</td>
<td>1.04s</td>
<td>0.77s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.36s</td>
<td>0.60s</td>
<td>0.77s</td>
<td>2.32s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.51s</td>
<td>0.63s</td>
<td>0.84s</td>
<td>2.52s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.46s</td>
<td>1.05s</td>
<td>0.86s</td>
<td>2.73s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>0.74s</td>
<td>0.74s</td>
<td>1.27s</td>
<td>2.37s</td>
</tr>
</tbody>
</table>
</div>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id15" id="image-processing-results" name="image-processing-results">Image Processing Results</a></h1>
<p>For these tests, we convert between two TIFF files with a simple image
processing step added in. The pupose of these tests is to show that the
performance advantages of GraphicsMagick carry over to simple image
processing tasks as well.</p>
<div class="section">
<h2><a class="toc-backref" href="#id16" id="blur-0-0x1-0-up-to-162-faster" name="blur-0-0x1-0-up-to-162-faster">-blur 0.0x1.0 (up to 162% faster)</a></h2>
<p>This test is the same as a TIFF to TIFF conversion except that
'-blur 0.0x1.0' is included. GraphicsMagick seems to be 2X faster than
ImageMagick for -blur.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>19.22s</td>
<td>1.25s</td>
<td>49.89s</td>
<td>3.46s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>19.25s</td>
<td>0.98s</td>
<td>36.89s</td>
<td>1.32s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>19.29s</td>
<td>1.18s</td>
<td>50.13s</td>
<td>3.41s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>19.24s</td>
<td>0.92s</td>
<td>37.31s</td>
<td>3.12s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>19.09s</td>
<td>1.12s</td>
<td>37.10s</td>
<td>3.29s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>19.15s</td>
<td>1.17s</td>
<td>37.38s</td>
<td>2.11s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>19.31s</td>
<td>0.92s</td>
<td>37.04s</td>
<td>1.46s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>14.61s</td>
<td>1.13s</td>
<td>28.63s</td>
<td>2.88s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>14.36s</td>
<td>0.96s</td>
<td>28.17s</td>
<td>2.91s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>14.32s</td>
<td>1.42s</td>
<td>28.30s</td>
<td>3.03s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>14.60s</td>
<td>0.98s</td>
<td>28.41s</td>
<td>3.00s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id17" id="gamma-1-2-up-to-219-faster" name="gamma-1-2-up-to-219-faster">-gamma 1.2 (up to 219% faster)</a></h2>
<p>This test is the same as a TIFF to TIFF conversion except that
'-gamma 1.2' is included. Clearly GraphicsMagick is much faster at this
task.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.03s</td>
<td>0.57s</td>
<td>1.77s</td>
<td>2.81s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>0.99s</td>
<td>0.52s</td>
<td>1.48s</td>
<td>0.71s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.08s</td>
<td>0.63s</td>
<td>2.54s</td>
<td>2.92s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>0.88s</td>
<td>0.63s</td>
<td>2.41s</td>
<td>2.89s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>0.97s</td>
<td>0.61s</td>
<td>2.41s</td>
<td>2.79s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>0.97s</td>
<td>0.56s</td>
<td>2.59s</td>
<td>1.41s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>1.02s</td>
<td>0.48s</td>
<td>1.62s</td>
<td>0.78s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>0.72s</td>
<td>0.59s</td>
<td>1.84s</td>
<td>2.43s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>0.80s</td>
<td>0.66s</td>
<td>1.83s</td>
<td>2.56s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.75s</td>
<td>1.07s</td>
<td>1.75s</td>
<td>2.85s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>1.13s</td>
<td>0.66s</td>
<td>1.90s</td>
<td>2.71s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id18" id="resize-50-only-34-faster" name="resize-50-only-34-faster">-resize 50% (only 34% faster)</a></h2>
<p>This test is the same as a TIFF to TIFF conversion except that '-resize
50%' is included so that there is some more serious image procesing. We
have done nothing at all to try to improve image resize performance
(considerable speed-up is possible). Image resize currently has severe
performance problems (compared to what is possible) due to the way the
algorithm accesses memory. Regardless, GraphicsMagick is still somewhat
faster.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>9.38s</td>
<td>0.95s</td>
<td>10.78s</td>
<td>3.15s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>7.38s</td>
<td>1.02s</td>
<td>10.55s</td>
<td>0.96s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>9.46s</td>
<td>0.86s</td>
<td>10.84s</td>
<td>3.04s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>9.40s</td>
<td>0.84s</td>
<td>10.61s</td>
<td>3.06s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>9.55s</td>
<td>0.83s</td>
<td>10.73s</td>
<td>2.94s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>7.40s</td>
<td>0.93s</td>
<td>8.79s</td>
<td>2.08s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>7.60s</td>
<td>0.82s</td>
<td>8.48s</td>
<td>1.57s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>5.53s</td>
<td>0.90s</td>
<td>6.91s</td>
<td>2.88s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>7.21s</td>
<td>0.71s</td>
<td>8.28s</td>
<td>2.51s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>7.14s</td>
<td>1.03s</td>
<td>8.41s</td>
<td>2.67s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>6.29s</td>
<td>0.79s</td>
<td>7.63s</td>
<td>2.46s</td>
</tr>
</tbody>
</table>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id19" id="rotate-90-up-to-134-faster" name="rotate-90-up-to-134-faster">-rotate 90 (up to 134% faster)</a></h2>
<p>This test is the same as a TIFF to TIFF conversion except that
'-rotate 90' is included so that there is some simple image processing.
Rotating by 90 degrees is a common requirement. For this system,
GraphicsMagick is typically 2X faster for this task.</p>
<table border="1" class="docutils">
<colgroup>
<col width="54%" />
<col width="11%" />
<col width="12%" />
<col width="11%" />
<col width="12%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head" rowspan="2">Name</th>
<th class="head" colspan="2">GraphicsMagick</th>
<th class="head" colspan="2">ImageMagick</th>
</tr>
<tr><th class="head">User</th>
<th class="head">System</th>
<th class="head">User</th>
<th class="head">System</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>golfer-600-bilevel-truecolor.tiff</td>
<td>1.28s</td>
<td>1.02s</td>
<td>2.82s</td>
<td>3.20s</td>
</tr>
<tr><td>golfer-600-bilevel.tiff</td>
<td>1.46s</td>
<td>1.19s</td>
<td>2.63s</td>
<td>1.02s</td>
</tr>
<tr><td>golfer-600-gray-truecolor.tiff</td>
<td>1.25s</td>
<td>1.15s</td>
<td>3.40s</td>
<td>3.29s</td>
</tr>
<tr><td>golfer-600-gray.tiff</td>
<td>1.18s</td>
<td>1.01s</td>
<td>3.42s</td>
<td>3.12s</td>
</tr>
<tr><td>golfer-600-gray16.tiff</td>
<td>1.11s</td>
<td>1.15s</td>
<td>3.47s</td>
<td>2.98s</td>
</tr>
<tr><td>golfer-600-palette16.tiff</td>
<td>1.55s</td>
<td>1.13s</td>
<td>4.15s</td>
<td>2.11s</td>
</tr>
<tr><td>golfer-600-pallet2.tiff</td>
<td>1.59s</td>
<td>1.08s</td>
<td>3.51s</td>
<td>1.17s</td>
</tr>
<tr><td>tiger-600-palette256.tiff</td>
<td>1.15s</td>
<td>1.04s</td>
<td>3.28s</td>
<td>2.63s</td>
</tr>
<tr><td>tiger-600-rgb.tiff</td>
<td>1.15s</td>
<td>0.86s</td>
<td>2.39s</td>
<td>3.01s</td>
</tr>
<tr><td>tiger-600-rgb16.tiff</td>
<td>0.96s</td>
<td>1.42s</td>
<td>2.60s</td>
<td>3.00s</td>
</tr>
<tr><td>tiger-600-rgba.tiff</td>
<td>1.23s</td>
<td>1.11s</td>
<td>2.67s</td>
<td>2.96s</td>
</tr>
</tbody>
</table>
</div>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id20" id="execution-overhead-results-3-2x-less-overhead" name="execution-overhead-results-3-2x-less-overhead">Execution Overhead Results (3.2X less overhead!)</a></h1>
<p>It is common to execute GraphicsMagick and ImageMagick utilities in
scripts. In this case, the utilities may be started over and over again
for thousands of files. If the operation requested takes little time to
execute then the time required to start the utility will dominate the
time. In order to evaluate the execution overhead, this small script
("bench"):</p>
<pre class="literal-block">
#!/bin/bash
i=1
count=1000
time while test $i -lt $count
do
eval "$@"
let i=i+1
done
</pre>
<p>Is used to execute a specified command 1000 times and report the amount
of time it took. For the purposes of this test, we take the 'rose' image
which is included in GraphicsMagick and ImageMagick, write it to an input
file, and then measure how long it takes to rotate it by 90 degrees and
write to another file:</p>
<pre class="literal-block">
/usr/local/bin/gm convert magick:rose input.tiff
./bench /usr/local/bin/convert input.tiff -rotate 90 output.tiff
./bench /usr/local/bin/gm convert input.tiff -rotate 90 output.tiff
</pre>
<p>Which results in these timings for ImageMagick:</p>
<blockquote>
<table border="1" class="docutils">
<colgroup>
<col width="40%" />
<col width="60%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head">Type</th>
<th class="head">Time</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>real</td>
<td>1m53.885s</td>
</tr>
<tr><td>user</td>
<td>1m30.421s</td>
</tr>
<tr><td>sys</td>
<td>0m17.356s</td>
</tr>
</tbody>
</table>
</blockquote>
<p>and these timings for GraphicsMagick:</p>
<blockquote>
<table border="1" class="docutils">
<colgroup>
<col width="40%" />
<col width="60%" />
</colgroup>
<thead valign="bottom">
<tr><th class="head">Type</th>
<th class="head">Time</th>
</tr>
</thead>
<tbody valign="top">
<tr><td>real</td>
<td>0m36.002s</td>
</tr>
<tr><td>user</td>
<td>0m22.440s</td>
</tr>
<tr><td>sys</td>
<td>0m12.008s</td>
</tr>
</tbody>
</table>
</blockquote>
<p>From which it can be seen that GraphicsMagick has 3.2X less overhead than
ImageMagick when executed on small files.</p>
<hr class="docutils" />
<div class="line-block">
<div class="line">Copyright (C) 2008 GraphicsMagick Group</div>
</div>
<p>This program is covered by multiple licenses, which are described in
Copyright.txt. You should have received a copy of Copyright.txt with this
package; otherwise see <a class="reference" href="http://www.graphicsmagick.org/www/Copyright.html">http://www.graphicsmagick.org/www/Copyright.html</a>.</p>
</div>
</div>
<div class="footer">
<hr class="footer" />
Generated on: 2008-05-06 18:42 UTC.
</div>
</body>
</html>
