Simon Brooke
f616992191
More unit tests now, and they all pass. More work needed, but this is very promising.
171 lines
11 KiB
HTML
171 lines
11 KiB
HTML
<!DOCTYPE html PUBLIC ""
|
||
"">
|
||
<html><head><meta charset="UTF-8" /><title>Dali performance</title><link rel="stylesheet" type="text/css" href="css/default.css" /><link rel="stylesheet" type="text/css" href="css/highlight.css" /><script type="text/javascript" src="js/highlight.min.js"></script><script type="text/javascript" src="js/jquery.min.js"></script><script type="text/javascript" src="js/page_effects.js"></script><script>hljs.initHighlightingOnLoad();</script></head><body><div id="header"><h2>Generated by <a href="https://github.com/weavejester/codox">Codox</a></h2><h1><a href="index.html"><span class="project-title"><span class="project-name">Walkmap</span> <span class="project-version">0.1.0-SNAPSHOT</span></span></a></h1></div><div class="sidebar primary"><h3 class="no-link"><span class="inner">Project</span></h3><ul class="index-link"><li class="depth-1 "><a href="index.html"><div class="inner">Index</div></a></li></ul><h3 class="no-link"><span class="inner">Topics</span></h3><ul><li class="depth-1 current"><a href="dali-performance.html"><div class="inner"><span>Dali performance</span></div></a></li><li class="depth-1 "><a href="intro.html"><div class="inner"><span>Introduction to walkmap</span></div></a></li></ul><h3 class="no-link"><span class="inner">Namespaces</span></h3><ul><li class="depth-1"><div class="no-link"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>walkmap</span></div></div></li><li class="depth-2 branch"><a href="walkmap.core.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>core</span></div></a></li><li class="depth-2 branch"><a href="walkmap.edge.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>edge</span></div></a></li><li class="depth-2 branch"><a href="walkmap.geometry.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>geometry</span></div></a></li><li class="depth-2 branch"><a href="walkmap.ocean.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>ocean</span></div></a></li><li class="depth-2 branch"><a href="walkmap.path.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>path</span></div></a></li><li class="depth-2 branch"><a href="walkmap.polygon.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>polygon</span></div></a></li><li class="depth-2 branch"><a href="walkmap.stl.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>stl</span></div></a></li><li class="depth-2 branch"><a href="walkmap.superstructure.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>superstructure</span></div></a></li><li class="depth-2 branch"><a href="walkmap.svg.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>svg</span></div></a></li><li class="depth-2 branch"><a href="walkmap.utils.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>utils</span></div></a></li><li class="depth-2"><a href="walkmap.vertex.html"><div class="inner"><span class="tree"><span class="top"></span><span class="bottom"></span></span><span>vertex</span></div></a></li></ul></div><div class="document" id="content"><div class="doc"><div class="markdown"><h1><a href="#dali-performance" name="dali-performance"></a>Dali performance</h1>
|
||
<p>Notes written while trying to characterise the performance problem in Dali.</p>
|
||
<h2><a href="#hypothesis-one-its-the-way-i-format-the-polygons-thats-the-issue" name="hypothesis-one-its-the-way-i-format-the-polygons-thats-the-issue"></a>Hypothesis one: it’s the way I format the polygons that’s the issue</h2>
|
||
<p>Firstly, with both versions of <code>stl->svg</code> using the same version of <code>facet->svg-poly</code>, i.e. this one:</p>
|
||
<pre><code>(defn- facet->svg-poly
|
||
[facet]
|
||
[:polygon
|
||
{:points (s/join " " (map #(str (:x %) "," (:y %)) (:vertices facet)))}])
|
||
</code></pre>
|
||
<p>we get this performance using the smaller <code>isle_of_man</code> map:</p>
|
||
<pre><code>walkmap.svg=> (def ^:dynamic *preferred-svg-render* :hiccup)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def hiccup (binary-stl-file->svg "resources/isle_of_man.stl" "resources/isle_of_man.svg")))
|
||
20-05-25 09:21:43 mason INFO [walkmap.svg:82] - Generating SVG for :hiccup renderer
|
||
20-05-25 09:21:43 mason INFO [walkmap.svg:96] - Emitting SVG with :hiccup renderer
|
||
"Elapsed time: 86.904891 msecs"
|
||
#'walkmap.svg/hiccup
|
||
walkmap.svg=> (def ^:dynamic *preferred-svg-render* :dali)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def dali (binary-stl-file->svg "resources/isle_of_man.stl" "resources/isle_of_man.svg")))
|
||
20-05-25 09:22:17 mason INFO [walkmap.svg:82] - Generating SVG for :dali renderer
|
||
20-05-25 09:22:17 mason INFO [walkmap.svg:96] - Emitting SVG with :dali renderer
|
||
"Elapsed time: 890.863814 msecs"
|
||
#'walkmap.svg/dali
|
||
</code></pre>
|
||
<p>If we switch the Dali render to use my original version of <code>facet->svg-poly</code>, i.e. this one:</p>
|
||
<pre><code>(defn- dali-facet->svg-poly
|
||
[facet]
|
||
(vec
|
||
(cons
|
||
:polygon
|
||
(map #(vec (list (:x %) (:y %))) (:vertices facet)))))
|
||
</code></pre>
|
||
<p>we get this performance:</p>
|
||
<pre><code>walkmap.svg=> (def ^:dynamic *preferred-svg-render* :hiccup)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def hiccup (binary-stl-file->svg "resources/isle_of_man.stl" "resources/isle_of_man.svg")))
|
||
20-05-25 09:35:33 mason INFO [walkmap.svg:82] - Generating SVG for :hiccup renderer
|
||
20-05-25 09:35:33 mason INFO [walkmap.svg:96] - Emitting SVG with :hiccup renderer
|
||
"Elapsed time: 84.09972 msecs"
|
||
#'walkmap.svg/hiccup
|
||
walkmap.svg=> (def ^:dynamic *preferred-svg-render* :dali)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def dali (binary-stl-file->svg "resources/isle_of_man.stl" "resources/isle_of_man.svg")))
|
||
20-05-25 09:35:41 mason INFO [walkmap.svg:82] - Generating SVG for :dali renderer
|
||
20-05-25 09:35:41 mason INFO [walkmap.svg:96] - Emitting SVG with :dali renderer
|
||
"Elapsed time: 874.292007 msecs"
|
||
#'walkmap.svg/dali
|
||
</code></pre>
|
||
<p>No significant difference in performance.</p>
|
||
<p>If we generate but don’t render, we get this:</p>
|
||
<pre><code>walkmap.svg=> (def ^:dynamic *preferred-svg-render* :hiccup)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def hiccup (binary-stl-file->svg "resources/isle_of_man.stl")))
|
||
20-05-25 09:37:44 mason INFO [walkmap.svg:82] - Generating SVG for :hiccup renderer
|
||
"Elapsed time: 52.614707 msecs"
|
||
#'walkmap.svg/hiccup
|
||
walkmap.svg=> (def ^:dynamic *preferred-svg-render* :dali)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (def dali (binary-stl-file->svg "resources/isle_of_man.stl")))
|
||
20-05-25 09:38:07 mason INFO [walkmap.svg:82] - Generating SVG for :dali renderer
|
||
"Elapsed time: 49.891043 msecs"
|
||
#'walkmap.svg/dali
|
||
</code></pre>
|
||
<p>This implies that the problem is not in the way polygons are formatted.</p>
|
||
<p>The difference between the two versions of <code>facet->svg-poly</code> is as follows:</p>
|
||
<h3><a href="#new-version-works-with-both-hiccup-and-dali-" name="new-version-works-with-both-hiccup-and-dali-"></a>New version, works with both Hiccup and Dali:</h3>
|
||
<pre><code>walkmap.svg=> (def stl (decode-binary-stl "resources/isle_of_man.stl"))
|
||
#'walkmap.svg/stl
|
||
walkmap.svg=> (def facet (first (:facets stl)))
|
||
#'walkmap.svg/facet
|
||
walkmap.svg=> (pprint facet)
|
||
{:normal {:x -0.0, :y 0.0, :z 1.0},
|
||
:vertices
|
||
[{:x 3.0, :y 1.0, :z 1.0}
|
||
{:x 2.0, :y 3.0, :z 1.0}
|
||
{:x 0.0, :y 0.0, :z 1.0}],
|
||
:abc 0}
|
||
nil
|
||
walkmap.svg=> (pprint (facet->svg-poly facet))
|
||
[:polygon {:points "3.0,1.0 2.0,3.0 0.0,0.0"}]
|
||
nil
|
||
</code></pre>
|
||
<p>In other words, the new version constructs the <code>:points</code> attribute of the <code>:polygon</code> tag by string concatenation, and the renderer just needs to output it.</p>
|
||
<h3><a href="#older-version-works-with-dali-only-" name="older-version-works-with-dali-only-"></a>Older version, works with Dali only:</h3>
|
||
<pre><code>walkmap.svg=> (pprint (dali-facet->svg-poly facet))
|
||
[:polygon [3.0 1.0] [2.0 3.0] [0.0 0.0]]
|
||
nil
|
||
</code></pre>
|
||
<p>This means that the renderer is actually doing more work, since it has to compose the <code>:points</code> attribute itself; nevertheless there doesn’t seem to be an increased time penalty.</p>
|
||
<h3><a href="#conclusion" name="conclusion"></a>Conclusion</h3>
|
||
<p>It doesn’t seem that formatting the polygons is the issue.</p>
|
||
<h2><a href="#hypothesis-two-dali-renderer-scales-non-linearly-with-number-of-objects-drawn" name="hypothesis-two-dali-renderer-scales-non-linearly-with-number-of-objects-drawn"></a>Hypothesis two: Dali renderer scales non-linearly with number of objects drawn</h2>
|
||
<p>To test this, we need some otherwise-similar test files with different numbers of objects:</p>
|
||
<pre><code>walkmap.svg=> (count (:facets stl))
|
||
4416
|
||
walkmap.svg=> (def small-stl (assoc stl :facets (take 400 (:facets stl))))
|
||
#'walkmap.svg/small-stl
|
||
walkmap.svg=> (count (:facets small-stl))
|
||
400
|
||
walkmap.svg=> (def large-stl (decode-binary-stl "../the-great-game/resources/maps/heightmap.stl"))
|
||
#'walkmap.svg/large-stl
|
||
walkmap.svg=> (count (:facets large-stl))
|
||
746585
|
||
walkmap.svg=> (def ^:dynamic *preferred-svg-render* :dali)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (dali.io/render-svg (stl->svg small-stl) "dali-small.svg"))
|
||
20-05-25 10:12:25 mason INFO [walkmap.svg:92] - Generating SVG for :dali renderer
|
||
"Elapsed time: 32.55506 msecs"
|
||
nil
|
||
walkmap.svg=> (def ^:dynamic *preferred-svg-render* :hiccup)
|
||
#'walkmap.svg/*preferred-svg-render*
|
||
walkmap.svg=> (time (spit "hiccup-small.svg" (hiccup.core/html (stl->svg small-stl))))
|
||
20-05-25 10:14:07 mason INFO [walkmap.svg:92] - Generating SVG for :hiccup renderer
|
||
"Elapsed time: 10.026369 msecs"
|
||
</code></pre>
|
||
<p>So we have</p>
|
||
<table>
|
||
<thead>
|
||
<tr>
|
||
<th> </th>
|
||
<th>Dali </th>
|
||
<th> </th>
|
||
<th>Hiccup </th>
|
||
<th> </th>
|
||
<th> </th>
|
||
</tr>
|
||
</thead>
|
||
<tbody>
|
||
<tr>
|
||
<td># of facets </td>
|
||
<td>time (msecs) </td>
|
||
<td>objets/msec </td>
|
||
<td>time (msecs) </td>
|
||
<td>objets/msec </td>
|
||
<td>ratio (Dali/Hiccup) </td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|
||
<table>
|
||
<tbody>
|
||
<tr>
|
||
<td>400 </td>
|
||
<td>32.55506 </td>
|
||
<td>12.29 </td>
|
||
<td>10.026369 </td>
|
||
<td>39.89 </td>
|
||
<td>3.35 </td>
|
||
</tr>
|
||
<tr>
|
||
<td>4416 </td>
|
||
<td>874.292007 </td>
|
||
<td>5.05 </td>
|
||
<td>84.09972 </td>
|
||
<td>52.51 </td>
|
||
<td>10.40 </td>
|
||
</tr>
|
||
<tr>
|
||
<td>746585 </td>
|
||
<td>29,695,695.61 </td>
|
||
<td>0.03 </td>
|
||
<td>16724.848222 </td>
|
||
<td>44.64 </td>
|
||
<td>1775.54 </td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|
||
<h3><a href="#conclusion" name="conclusion"></a>Conclusion</h3>
|
||
<p>What we’re seeing is that Hiccup renders more or less linearly by the number of objects (bear in mind that all of these objects are triangles, so essentially equally complex to render), whereas trhe performance of Dali degrades significantly as the number of objects increases.</p></div></div></div></body></html> |