File: partitioners.htm

package info (click to toggle)
tbb 4.2~20140122-5
  • links: PTS
  • area: main
  • in suites: jessie, jessie-kfreebsd
  • size: 21,492 kB
  • ctags: 21,278
  • sloc: cpp: 92,813; ansic: 9,775; asm: 1,070; makefile: 1,057; sh: 351; java: 226; objc: 98; pascal: 71; xml: 41
file content (186 lines) | stat: -rwxr-xr-x 7,371 bytes parent folder | download
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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
<!DOCTYPE html
  PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<!-- saved from url=(0014)about:internet -->
<html xmlns:MSHelp="http://www.microsoft.com/MSHelp/" lang="en-us" xml:lang="en-us"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">

<meta name="DC.Type" content="topic">
<meta name="DC.Title" content="Partitioners">
<meta name="DC.subject" content="Partitioners">
<meta name="keywords" content="Partitioners">
<meta name="DC.Relation" scheme="URI" content="../../reference/algorithms.htm">
<meta name="DC.Relation" scheme="URI" content="../../reference/algorithms/partitioners/auto_partitioner_cls.htm">
<meta name="DC.Relation" scheme="URI" content="../../reference/algorithms/partitioners/affinity_partitioner.htm">
<meta name="DC.Relation" scheme="URI" content="../../reference/algorithms/partitioners/simple_partitioner_cls.htm">
<meta name="DC.Relation" scheme="URI" content="parallel_for_func.htm#parallel_for_func">
<meta name="DC.Relation" scheme="URI" content="parallel_reduce_func.htm#parallel_reduce_func">
<meta name="DC.Relation" scheme="URI" content="parallel_scan_func.htm#parallel_scan_func">
<meta name="DC.Format" content="XHTML">
<meta name="DC.Identifier" content="partitioners">
<meta name="DC.Language" content="en-US">
<link rel="stylesheet" type="text/css" href="../../intel_css_styles.css">
<title>Partitioners</title>
<xml>
<MSHelp:Attr Name="DocSet" Value="Intel"></MSHelp:Attr>
<MSHelp:Attr Name="Locale" Value="kbEnglish"></MSHelp:Attr>
<MSHelp:Attr Name="TopicType" Value="kbReference"></MSHelp:Attr>
</xml>
</head>
<body id="partitioners">
 <!-- ==============(Start:NavScript)================= -->
 <script src="..\..\NavScript.js" language="JavaScript1.2" type="text/javascript"></script>
 <script language="JavaScript1.2" type="text/javascript">WriteNavLink(2);</script>
 <!-- ==============(End:NavScript)================= -->
<a name="partitioners"><!-- --></a>

 
  <h1 class="topictitle1">Partitioners</h1>
 
	 
  <div> 
	 <div class="section"><h2 class="sectiontitle">Summary</h2> 
		 
		<p>A partitioner specifies how a loop template should partition its work
		  among threads. 
		</p>
 
	 </div>
 
	 <div class="section"><h2 class="sectiontitle"> Description</h2> 
		 
		<p>The default behavior of the loop templates<samp class="codeph"> parallel_for</samp>,
					<samp class="codeph">parallel_reduce</samp>, and <samp class="codeph">parallel_scan</samp> tries to
				recursively split a range into enough parts to keep processors busy, not necessarily
				splitting as finely as possible. An optional partitioner parameter enables other
				behaviors to be specified, as shown in the table below. The first column of the table shows
				how the formal parameter is declared in the loop templates. An<samp class="codeph">
					affinity_partitioner</samp> is passed by non-const reference because it is
				updated to remember where loop iterations run. </p>
 
		<p> 
		
<div class="tablenoborder"><table cellpadding="4" summary="" width="100%" frame="hsides" border="1" rules="all"><caption><span class="tablecap">Partitioners</span></caption> 
			 <thead align="left"> 
				<tr> 
				  <th class="cellrowborder" valign="top" id="d9145e71"> 
					 <p> Partitioner 
					 </p>
 
				  </th>
 
				  <th class="row-nocellborder" valign="top" id="d9145e77"> 
					 <p> Loop Behavior 
					 </p>
 
				  </th>
 
				</tr>
</thead>
 
			 <tbody> 
				<tr valign="top"> 
				  <td class="cellrowborder" valign="top" headers="d9145e71 "> 
					 <p><samp class="codeph">const auto_partitioner&amp;
						  (default)</samp> 
					 </p>
 
					  
						<div class="Note"><h3 class="NoteTipHead">
					Note</h3><p>In Intel&reg; Threading Building Blocks (Intel&reg; TBB) 2.1, <samp class="codeph">simple_partitioner</samp> was the default. Intel&reg;
											TBB 2.2 changed the default to
												<samp class="codeph">auto_partitioner</samp> to simplify common
											usage of the loop templates. To get the old default,
											compile with the preprocessor symbol
												<samp class="codeph">TBB_DEPRECATED=1</samp>. </p>
</div> 
					  
				  </td>
 
				  <td class="row-nocellborder" valign="top" headers="d9145e77 "> 
					 <p>Performs sufficient splitting to balance load, not necessarily splitting
										as finely as <samp class="codeph">Range::is_divisible</samp> permits.
										When used with classes such as
											<samp class="codeph">blocked_range</samp>, the selection of an
										appropriate grainsize is less important, and often
										acceptable performance can be achieved with the default
										grain size of 1. </p>
 
				  </td>
 
				</tr>
 
				<tr valign="top"> 
				  <td class="cellrowborder" valign="top" headers="d9145e71 "> 
					 <p><samp class="codeph">affinity_partitioner&amp;</samp> 
					 </p>
 
				  </td>
 
				  <td class="row-nocellborder" valign="top" headers="d9145e77 "> 
					 <p> Similar to <samp class="codeph">auto_partitioner</samp>, but improves cache
										affinity by its choice of mapping subranges to worker
										threads. It can improve performance significantly when a
										loop is re-executed over the same data set, and the data set
										fits in cache. </p>
 
				  </td>
 
				</tr>
 
				<tr valign="top"> 
				  <td class="cellrowborder" valign="top" headers="d9145e71 "> 
					 <p><samp class="codeph">const simple_partitioner&amp; 
						</samp> 
					 </p>
 
				  </td>
 
				  <td class="row-nocellborder" valign="top" headers="d9145e77 "> 
					 <p>Recursively splits a range until it is no longer divisible. The
											<samp class="codeph">Range::is_divisible</samp> function is wholly
										responsible for deciding when recursive splitting halts.
										When used with classes such as
											<samp class="codeph">blocked_range</samp>, the selection of an
										appropriate grainsize is critical to enabling concurrency
										while limiting overheads (see the discussion in the blocked_range Template Class section). </p>
 
				  </td>
 
				</tr>
 
			 </tbody>
 
		  </table>
</div>
 
		</p>
 
	 </div>
 
  </div>
 
  
<div class="familylinks">
<div class="parentlink"><strong>Parent topic:</strong>&nbsp;<a href="../../reference/algorithms.htm">Algorithms</a></div>
</div>
<div class="See Also">
<ul class="ullinks">
<li class="ulchildlink"><a href="../../reference/algorithms/partitioners/auto_partitioner_cls.htm">auto_partitioner Class</a><br>
</li>
<li class="ulchildlink"><a href="../../reference/algorithms/partitioners/affinity_partitioner.htm">affinity_partitioner</a><br>
</li>
<li class="ulchildlink"><a href="../../reference/algorithms/partitioners/simple_partitioner_cls.htm">simple_partitioner Class</a><br>
</li>
</ul>

<h2>See Also</h2>
<div class="linklist">
<div><a href="parallel_for_func.htm#parallel_for_func">parallel_for Template Function
		  </a></div>
<div><a href="parallel_reduce_func.htm#parallel_reduce_func">parallel_reduce Template Function
		  </a></div>
<div><a href="parallel_scan_func.htm#parallel_scan_func">parallel_scan Template Function
		  </a></div></div>
</div> 

</body>
</html>