📄 node60.html
字号:
<html><!DOCTYPE HTML PUBLIC "-//W3O//DTD W3 HTML 2.0//EN">
<!Converted with LaTeX2HTML 95.1 (Fri Jan 20 1995) by Nikos Drakos (nikos@cbl.leeds.ac.uk), CBLU, University of Leeds >
<HEAD>
<TITLE>5.9 Modularity</TITLE>
</HEAD>
<BODY>
<meta name="description" value="5.9 Modularity">
<meta name="keywords" value="book">
<meta name="resource-type" value="document">
<meta name="distribution" value="global">
<P>
<BR> <HR><a href="msgs0.htm#2" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/tppmsgs/msgs0.htm#2"><img ALIGN=MIDDLE src="asm_color_tiny.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/asm_color_tiny.gif" alt="[DBPP]"></a> <A NAME=tex2html2655 HREF="node59.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node59.html"><IMG ALIGN=MIDDLE ALT="previous" SRC="previous_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/previous_motif.gif"></A> <A NAME=tex2html2663 HREF="node61.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node61.html"><IMG ALIGN=MIDDLE ALT="next" SRC="next_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/next_motif.gif"></A> <A NAME=tex2html2661 HREF="node51.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node51.html"><IMG ALIGN=MIDDLE ALT="up" SRC="up_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/up_motif.gif"></A> <A NAME=tex2html2665 HREF="node1.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node1.html"><IMG ALIGN=MIDDLE ALT="contents" SRC="contents_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/contents_motif.gif"></A> <A NAME=tex2html2666 HREF="node133.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node133.html"><IMG ALIGN=MIDDLE ALT="index" SRC="index_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/index_motif.gif"></A> <a href="msgs0.htm#3" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/tppmsgs/msgs0.htm#3"><img ALIGN=MIDDLE src="search_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/search_motif.gif" alt="[Search]"></a> <BR>
<B> Next:</B> <A NAME=tex2html2664 HREF="node61.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node61.html">5.10 Performance Issues</A>
<B>Up:</B> <A NAME=tex2html2662 HREF="node51.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node51.html">5 Compositional C++
</A>
<B> Previous:</B> <A NAME=tex2html2656 HREF="node59.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node59.html">5.8 Mapping</A>
<BR><HR><P>
<H1><A NAME=SECTION03290000000000000000>5.9 Modularity</A></H1>
<P>
<A NAME=secccmod> </A>
<P>
<A NAME=7589> </A>
Example <A HREF="node59.html#exccmxc" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node59.html#exccmxc">5.5</A> showed how CC++
constructs can be used to
<A NAME=7591> </A>
implement parallel and concurrent composition. The basic ideas are
<A NAME=7592> </A>
straightforward: each component
<A NAME=7593> </A>
of a multicomponent program is
<A NAME=7594> </A>
implemented as a distinct task. Each such task is passed an array of
<tt> proc_t</tt> objects representing its computational resources. The
task creates a distinct set of processor objects and performs mapping
and communication with respect to these <tt> proc_t</tt> objects and
processor objects.
<P>
<A NAME=7597> </A>
In this section, we discuss the techniques used to implement <em>
<A NAME=7598> </A>
sequential
</em> composition in CC++
programs. Recall that in
sequential composition, different program components execute in
sequence on all processors (Section <A HREF="node41.html#secseqcomp" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node41.html#secseqcomp">4.2.2</A>). These
program components may themselves communicate and synchronize, but
they cannot create new tasks. Hence, each process executes the same
program, and the entire computation moves sequentially from one
parallel operation to the next. This is the single program multiple
data (SPMD) programming model discussed in Section <A HREF="node9.html#sec1other" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node9.html#sec1other">1.3.2</A>.
<P>
A CC++
implementation of an SPMD program comprises two components.
The <em> initialization
</em> component creates the processor objects
in which computation will be performed and the communication
structures (such as channels) required by the program components
called in the sequential composition. The <em> execution
</em>
component performs the actual computation, using the structures
created during the initialization phase. The execution component can
be structured in two different ways (Figure <A HREF="node60.html#figccseqc" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node60.html#figccseqc">5.7</A>). In
the first approach, the top-level program is structured as a sequence
of calls to routines that each use a <tt> parfor</tt> statement to create
a thread on each processor object. In the second approach a single
<tt> parfor</tt> statement creates a set of long-lived threads (one per
processor object) that each make a sequence of calls to the various
routines involved in the sequential composition. The first approach
can lead to simpler programs, but the latter tends to be more
efficient.
<P>
<P><A NAME=8486> </A><IMG BORDER=0 ALIGN=BOTTOM ALT="" SRC="img859.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/img859.gif">
<BR><STRONG>Figure 5.7:</STRONG> <em> Two alternative approaches to the implementation of sequential
<A NAME=8480> </A>
composition in CC++
. The two figures are timelines, with the parent
<A NAME=8481> </A>
thread shown as a solid dark line when active and a dashed dark line
when suspended. In both cases, an initialization phase creates the
four processor objects in which computation will occur. In the first
approach, a set of long-lived threads is then created; each of these
threads executes both components before terminating. In the second
approach, control returns to the parent thread after the first
component executes; the parent thread then creates a new set of
threads for the second component.</em><A NAME=figccseqc> </A><BR>
<P>
<P>
<BR><HR>
<b> Example <IMG BORDER=0 ALIGN=BOTTOM ALT="" SRC="img861.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/img861.gif">.<IMG BORDER=0 ALIGN=BOTTOM ALT="" SRC="img860.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/img860.gif"> Finite Difference</b>:<A NAME=exccfd> </A>
<P>
<A NAME=7613> </A>
We apply the two approaches to the SPMD finite difference computation
<A NAME=7614> </A>
used to illustrate sequential composition in
Section <A HREF="node41.html#secseqcomp" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node41.html#secseqcomp">4.2.2</A>. This computation is structured as a
sequence of calls to a finite difference routine that performs
nearest-neighbor communication and a reduction routine used to detect
termination; the latter routine performs global communication.
<P>
An implementation of this algorithm using the first approach is
illustrated in Program <A HREF="node60.html#progccfd1" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node60.html#progccfd1">5.13</A>. The execution component is
structured as a while loop containing a sequential composition of
parallel <tt> finite_difference</tt> and <tt> global_maximum</tt> routines.
Concurrency and communication are encapsulated in these routines,
which use <tt> parfor</tt> and other parallel constructs to create threads
of control within the processor objects created in the initialization
phase.
<P>
The second approach is illustrated in Program <A HREF="node60.html#progccfd2" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node60.html#progccfd2">5.14</A>.
Here, a single <tt> parfor</tt> statement creates one thread of control in
each processor object. This thread makes a sequence of calls to
routines that call local <tt> finite_difference</tt> and <tt>
global_maximum</tt> routines. The resulting program is more complex but
potentially more efficient, as it avoids the cost of repeatedly
creating and destroying the threads used to perform the SPMD
computation.
<P>
<BR><HR>
<P>
<P><A NAME=progccfd1> </A><IMG BORDER=0 ALIGN=BOTTOM ALT="" SRC="img862.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/img862.gif"><P>
<P>
<P><A NAME=progccfd2> </A><IMG BORDER=0 ALIGN=BOTTOM ALT="" SRC="img863.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/img863.gif"><P>
<P>
<BR> <HR><a href="msgs0.htm#2" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/tppmsgs/msgs0.htm#2"><img ALIGN=MIDDLE src="asm_color_tiny.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/asm_color_tiny.gif" alt="[DBPP]"></a> <A NAME=tex2html2655 HREF="node59.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node59.html"><IMG ALIGN=MIDDLE ALT="previous" SRC="previous_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/previous_motif.gif"></A> <A NAME=tex2html2663 HREF="node61.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node61.html"><IMG ALIGN=MIDDLE ALT="next" SRC="next_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/next_motif.gif"></A> <A NAME=tex2html2661 HREF="node51.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node51.html"><IMG ALIGN=MIDDLE ALT="up" SRC="up_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/up_motif.gif"></A> <A NAME=tex2html2665 HREF="node1.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node1.html"><IMG ALIGN=MIDDLE ALT="contents" SRC="contents_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/contents_motif.gif"></A> <A NAME=tex2html2666 HREF="node133.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node133.html"><IMG ALIGN=MIDDLE ALT="index" SRC="index_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/index_motif.gif"></A> <a href="msgs0.htm#3" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/tppmsgs/msgs0.htm#3"><img ALIGN=MIDDLE src="search_motif.gif" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/search_motif.gif" alt="[Search]"></a> <BR>
<B> Next:</B> <A NAME=tex2html2664 HREF="node61.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node61.html">5.10 Performance Issues</A>
<B>Up:</B> <A NAME=tex2html2662 HREF="node51.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node51.html">5 Compositional C++
</A>
<B> Previous:</B> <A NAME=tex2html2656 HREF="node59.html" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/node59.html">5.8 Mapping</A>
<BR><HR><P>
<P><ADDRESS>
<I>© Copyright 1995 by <A href="msgs0.htm#6" tppabs="http://www.dit.hcmut.edu.vn/books/system/par_anl/tppmsgs/msgs0.htm#6">Ian Foster</a></I>
</ADDRESS>
</BODY>
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -