1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
"http://www.w3.org/TR/html4/loose.dtd">
<html >
<head><title>Optimisation</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="generator" content="TeX4ht (http://www.tug.org/tex4ht/)">
<meta name="originator" content="TeX4ht (http://www.tug.org/tex4ht/)">
<!-- 3,html -->
<meta name="src" content="nipguide.tex">
<link rel="stylesheet" type="text/css" href="nipguide.css">
</head><body
>
<!--l. 1527--><div class="crosslinks"><p class="noindent">[<a
href="nipguideap1.html" >next</a>] [<a
href="nipguidese34.html" >prev</a>] [<a
href="nipguidese34.html#tailnipguidese34.html" >prev-tail</a>] [<a
href="nipguidese34.html#tailnipguidese35.html">tail</a>] [<a
href="nipguidech6.html#nipguidese36.html" >up</a>] </p></div>
<h3 class="sectionHead"><span class="titlemark">6.14 </span> <a
id="x46-870006.14"></a>Optimisation</h3>
<!--l. 1528--><p class="noindent" ><a name="nip_label_sec:optimise"></a>
<!--l. 1530--><p class="indent" > <span
class="phvr7t-x-x-80">nip2 </span>performs three useful optimisations on expressions.
First, it finds and removes common sub-expressions in
functions. So for example:
<div class="verbatim" id="verbatim-89">
if a + b < 12 then a + b else b
</div>
<!--l. 1535--><p class="nopar" >
<!--l. 1537--><p class="noindent" >will only evaluate <span
class="phvr7t-x-x-80">a + b </span>once. This can save a lot of time if <span
class="phvr7t-x-x-80">a</span>
or <span
class="phvr7t-x-x-80">b </span>is a large image.
<!--l. 1541--><p class="indent" > Second, <span
class="phvr7t-x-x-80">nip2 </span>detects arithmetic operations on <span
class="phvr7t-x-x-80">unsigned</span>
<span
class="phvr7t-x-x-80">char </span>images, and replaces them with look-up tables. For
example:
<div class="verbatim" id="verbatim-90">
a = vips_image "campin.v"
 <br />b = a ⋆ (a - 1) ⋆⋆ 0.5
</div>
<!--l. 1547--><p class="nopar" >
<!--l. 1549--><p class="noindent" >Provided <span
class="phvr7t-x-x-80">campin.v </span>is an 8 bit image image, this expression
will evaluate with a single call to <span
class="phvr7t-x-x-80">im</span><span
class="phvr7t-x-x-80">_maplut()</span>.
<!--l. 1553--><p class="indent" > Finally, <span
class="phvr7t-x-x-80">nip2 </span>has a VIPS operation cache. It memorises the
arguments to the last few hundred calls to VIPS, and the
result each call gave. Before calling VIPS again, it
checks to see if there is a previous call with the same
arguments and if there is, uses the result it obtained last
time.
<!--l. 1558--><div class="crosslinks"><p class="noindent">[<a
href="nipguideap1.html" >next</a>] [<a
href="nipguidese34.html" >prev</a>] [<a
href="nipguidese34.html#tailnipguidese34.html" >prev-tail</a>] [<a
href="nipguidese35.html" >front</a>] [<a
href="nipguidech6.html#nipguidese36.html" >up</a>] </p></div>
<!--l. 1558--><p class="indent" > <a
id="tailnipguidese35.html"></a>
</body></html>
|
