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
|
<HTML>
<BODY>
<H2>Overview</H2>
Example that uses parallel_do to do parallel preorder traversal of a sparse graph.
<P>
Each vertex in the graph is called a "cell".
Each cell has a value.
The value is a matrix.
Some of the cells have operators
that compute the cell's value, using other cell's values as input.
A cell that uses the value of cell x is called a successor of x.
</P><P>
The algorithm works as follows.
<OL>
<LI> Compute the set of cells that have no inputs. This set is called <TT>root_set</TT>.
<LI> Each cell has an associated field <TT>ref_count</TT> that is an atomic integer.
Initialize <TT>ref_count</TT> to the number of inputs for the Cell.
<LI> Update each cell in <TT>root_set</TT>, by applying a <TT>parallel_do</TT> to a <TT>root_set</TT>
<LI> After updating a cell, for each of its successors
<OL>
<LI> Atomically decrement the successor's <TT>ref_count</TT>
<LI> If the count became zero, add the cell to the set of cells to be updated,
by calling <TT>parallel_do_feeder_impl::add</TT>.
</OL>
</OL>
</P><P>
The times printed are for the traversal and update,
and do not include time for computing the root_set.
</P><P>
The example is using custom synchronization via <TT>ref_count</TT> atomic variable.
Correctness checking tools might not take this into account, and report data races
between different tasks that are actually synchronized.
</P>
<B>NOTE: </B>It is important to understand that this example is unlikely to show speedup
if the cell values are changed to type "float". The reason is twofold.
<UL>
<LI> The smaller value type causes each Cell to be significantly smaller than a cache line,
which leads to false sharing conflicts.
<LI> The time to update the cells becomes very small, and consequently the overhead of
parallel_do swamps the useful work.
</UL>
<H2>Files</H2>
<DL>
<DT><A HREF="main.cpp">main.cpp</A>
<DD>Main program which parses command line options and runs the algorithm with different numbers of threads.
<DT><A HREF="parallel_preorder.cpp">parallel_preorder.cpp</A>
<DD>Implementation of the parallel preorder traversal algorithm.
<DT><A HREF="Graph.h">Graph.h</A>
<DD>Interfaces of the Graph and Cell classes.
<DT><A HREF="Graph.cpp">Graph.cpp</A>
<DD>Implementations of the Graph and Cell classes.
<DT><A HREF="Matrix.h">Matrix.h</A>
<DD>The Matrix class definition.
<DT><A HREF="Makefile">Makefile</A>
<DD>Makefile for building example.
</DL>
<H2>Directories</H2>
<DL>
<DT><A HREF="msvs">msvs</A>
<DD>Contains Microsoft* Visual Studio* 2005 workspace for building and running the example (Windows* systems only).
<DT><A HREF="xcode">xcode</A>
<DD>Contains Xcode* IDE workspace for building and running the example (OS X* systems only).
</DL>
<H2>To Build</H2>
General build directions can be found <A HREF=../../index.html#build>here</A>.
<H2>Usage</H2>
<DL>
<DT><TT>parallel_preorder <I>-h</I></TT>
<DD>Prints the help for command line options
<DT><TT>parallel_preorder [<I>n-of-threads</I>=value] [<I>n-of-nodes</I>=value] [<I>n-of-traversals</I>=value] [<I>silent</I>] </TT>
<DT><TT>parallel_preorder [<I>n-of-threads</I> [<I>n-of-nodes</I> [<I>n-of-traversals</I>]]] [<I>silent</I>] </TT>
<DD><I>n-of-threads</I> is the number of threads to use; a range of the form <I>low</I>[:<I>high</I>], where low and optional high are non-negative integers or 'auto' for the TBB default.<BR>
<I>n-of-nodes</I> is a number of nodes in the graph. Default value is 1000.<BR>
<I>n-of-traversals</I> is the number of times to evaluate the graph. Default value is 500.<BR>
<I>silent</I> - no output except elapsed time.<BR>
<DT>To run a short version of this example, e.g., for use with Intel® Parallel Inspector:
<DD>Build a <I>debug</I> version of the example
(see the <A HREF=../../index.html#build>build directions</A>).
<BR>Run it with the desired number of threads and smaller number of traversals, e.g., <TT>parallel_preorder 4 1000 5</TT>.
</DL>
<HR>
<A HREF="../index.html">Up to parent directory</A>
<p></p>
Copyright © 2005-2014 Intel Corporation. All Rights Reserved.
<P></P>
Intel is a registered trademark or trademark of Intel Corporation
or its subsidiaries in the United States and other countries.
<p></p>
* Other names and brands may be claimed as the property of others.
</BODY>
</HTML>
|
