iup-stack/fftw/doc/html/FFTW-MPI-Wisdom.html

172 lines
7.8 KiB
HTML

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<!-- This manual is for FFTW
(version 3.3.10, 10 December 2020).
Copyright (C) 2003 Matteo Frigo.
Copyright (C) 2003 Massachusetts Institute of Technology.
Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.
Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.
Permission is granted to copy and distribute translations of this manual
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation
approved by the Free Software Foundation. -->
<!-- Created by GNU Texinfo 6.7, http://www.gnu.org/software/texinfo/ -->
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<title>FFTW MPI Wisdom (FFTW 3.3.10)</title>
<meta name="description" content="FFTW MPI Wisdom (FFTW 3.3.10)">
<meta name="keywords" content="FFTW MPI Wisdom (FFTW 3.3.10)">
<meta name="resource-type" content="document">
<meta name="distribution" content="global">
<meta name="Generator" content="makeinfo">
<link href="index.html" rel="start" title="Top">
<link href="Concept-Index.html" rel="index" title="Concept Index">
<link href="index.html#SEC_Contents" rel="contents" title="Table of Contents">
<link href="Distributed_002dmemory-FFTW-with-MPI.html" rel="up" title="Distributed-memory FFTW with MPI">
<link href="Avoiding-MPI-Deadlocks.html" rel="next" title="Avoiding MPI Deadlocks">
<link href="An-improved-replacement-for-MPI_005fAlltoall.html" rel="prev" title="An improved replacement for MPI_Alltoall">
<style type="text/css">
<!--
a.summary-letter {text-decoration: none}
blockquote.indentedblock {margin-right: 0em}
div.display {margin-left: 3.2em}
div.example {margin-left: 3.2em}
div.lisp {margin-left: 3.2em}
kbd {font-style: oblique}
pre.display {font-family: inherit}
pre.format {font-family: inherit}
pre.menu-comment {font-family: serif}
pre.menu-preformatted {font-family: serif}
span.nolinebreak {white-space: nowrap}
span.roman {font-family: initial; font-weight: normal}
span.sansserif {font-family: sans-serif; font-weight: normal}
ul.no-bullet {list-style: none}
-->
</style>
</head>
<body lang="en">
<span id="FFTW-MPI-Wisdom"></span><div class="header">
<p>
Next: <a href="Avoiding-MPI-Deadlocks.html" accesskey="n" rel="next">Avoiding MPI Deadlocks</a>, Previous: <a href="FFTW-MPI-Transposes.html" accesskey="p" rel="prev">FFTW MPI Transposes</a>, Up: <a href="Distributed_002dmemory-FFTW-with-MPI.html" accesskey="u" rel="up">Distributed-memory FFTW with MPI</a> &nbsp; [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Concept-Index.html" title="Index" rel="index">Index</a>]</p>
</div>
<hr>
<span id="FFTW-MPI-Wisdom-1"></span><h3 class="section">6.8 FFTW MPI Wisdom</h3>
<span id="index-wisdom-2"></span>
<span id="index-saving-plans-to-disk-2"></span>
<p>FFTW&rsquo;s &ldquo;wisdom&rdquo; facility (see <a href="Words-of-Wisdom_002dSaving-Plans.html">Words of Wisdom-Saving Plans</a>) can
be used to save MPI plans as well as to save uniprocessor plans.
However, for MPI there are several unavoidable complications.
</p>
<span id="index-MPI-I_002fO-1"></span>
<p>First, the MPI standard does not guarantee that every process can
perform file I/O (at least, not using C stdio routines)&mdash;in general,
we may only assume that process 0 is capable of I/O.<a id="DOCF7" href="#FOOT7"><sup>7</sup></a> So, if we
want to export the wisdom from a single process to a file, we must
first export the wisdom to a string, then send it to process 0, then
write it to a file.
</p>
<p>Second, in principle we may want to have separate wisdom for every
process, since in general the processes may run on different hardware
even for a single MPI program. However, in practice FFTW&rsquo;s MPI code
is designed for the case of homogeneous hardware (see <a href="Load-balancing.html">Load balancing</a>), and in this case it is convenient to use the same wisdom
for every process. Thus, we need a mechanism to synchronize the wisdom.
</p>
<p>To address both of these problems, FFTW provides the following two
functions:
</p>
<div class="example">
<pre class="example">void fftw_mpi_broadcast_wisdom(MPI_Comm comm);
void fftw_mpi_gather_wisdom(MPI_Comm comm);
</pre></div>
<span id="index-fftw_005fmpi_005fgather_005fwisdom"></span>
<span id="index-fftw_005fmpi_005fbroadcast_005fwisdom"></span>
<p>Given a communicator <code>comm</code>, <code>fftw_mpi_broadcast_wisdom</code>
will broadcast the wisdom from process 0 to all other processes.
Conversely, <code>fftw_mpi_gather_wisdom</code> will collect wisdom from all
processes onto process 0. (If the plans created for the same problem
by different processes are not the same, <code>fftw_mpi_gather_wisdom</code>
will arbitrarily choose one of the plans.) Both of these functions
may result in suboptimal plans for different processes if the
processes are running on non-identical hardware. Both of these
functions are <em>collective</em> calls, which means that they must be
executed by all processes in the communicator.
<span id="index-collective-function-1"></span>
</p>
<p>So, for example, a typical code snippet to import wisdom from a file
and use it on all processes would be:
</p>
<div class="example">
<pre class="example">{
int rank;
fftw_mpi_init();
MPI_Comm_rank(MPI_COMM_WORLD, &amp;rank);
if (rank == 0) fftw_import_wisdom_from_filename(&quot;mywisdom&quot;);
fftw_mpi_broadcast_wisdom(MPI_COMM_WORLD);
}
</pre></div>
<p>(Note that we must call <code>fftw_mpi_init</code> before importing any
wisdom that might contain MPI plans.) Similarly, a typical code
snippet to export wisdom from all processes to a file is:
<span id="index-fftw_005fmpi_005finit-2"></span>
</p>
<div class="example">
<pre class="example">{
int rank;
fftw_mpi_gather_wisdom(MPI_COMM_WORLD);
MPI_Comm_rank(MPI_COMM_WORLD, &amp;rank);
if (rank == 0) fftw_export_wisdom_to_filename(&quot;mywisdom&quot;);
}
</pre></div>
<div class="footnote">
<hr>
<h4 class="footnotes-heading">Footnotes</h4>
<h5><a id="FOOT7" href="#DOCF7">(7)</a></h3>
<p>In fact,
even this assumption is not technically guaranteed by the standard,
although it seems to be universal in actual MPI implementations and is
widely assumed by MPI-using software. Technically, you need to query
the <code>MPI_IO</code> attribute of <code>MPI_COMM_WORLD</code> with
<code>MPI_Attr_get</code>. If this attribute is <code>MPI_PROC_NULL</code>, no
I/O is possible. If it is <code>MPI_ANY_SOURCE</code>, any process can
perform I/O. Otherwise, it is the rank of a process that can perform
I/O ... but since it is not guaranteed to yield the <em>same</em> rank
on all processes, you have to do an <code>MPI_Allreduce</code> of some kind
if you want all processes to agree about which is going to do I/O.
And even then, the standard only guarantees that this process can
perform output, but not input. See e.g. <cite>Parallel Programming
with MPI</cite> by P. S. Pacheco, section 8.1.3. Needless to say, in our
experience virtually no MPI programmers worry about this.</p>
</div>
<hr>
<div class="header">
<p>
Next: <a href="Avoiding-MPI-Deadlocks.html" accesskey="n" rel="next">Avoiding MPI Deadlocks</a>, Previous: <a href="FFTW-MPI-Transposes.html" accesskey="p" rel="prev">FFTW MPI Transposes</a>, Up: <a href="Distributed_002dmemory-FFTW-with-MPI.html" accesskey="u" rel="up">Distributed-memory FFTW with MPI</a> &nbsp; [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Concept-Index.html" title="Index" rel="index">Index</a>]</p>
</div>
</body>
</html>