Mercurial > hg > sv-dependency-builds
diff src/fftw-3.3.3/doc/html/Combining-MPI-and-Threads.html @ 10:37bf6b4a2645
Add FFTW3
| author | Chris Cannam |
|---|---|
| date | Wed, 20 Mar 2013 15:35:50 +0000 |
| parents | |
| children |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/src/fftw-3.3.3/doc/html/Combining-MPI-and-Threads.html Wed Mar 20 15:35:50 2013 +0000 @@ -0,0 +1,135 @@ +<html lang="en"> +<head> +<title>Combining MPI and Threads - FFTW 3.3.3</title> +<meta http-equiv="Content-Type" content="text/html"> +<meta name="description" content="FFTW 3.3.3"> +<meta name="generator" content="makeinfo 4.13"> +<link title="Top" rel="start" href="index.html#Top"> +<link rel="up" href="Distributed_002dmemory-FFTW-with-MPI.html#Distributed_002dmemory-FFTW-with-MPI" title="Distributed-memory FFTW with MPI"> +<link rel="prev" href="FFTW-MPI-Performance-Tips.html#FFTW-MPI-Performance-Tips" title="FFTW MPI Performance Tips"> +<link rel="next" href="FFTW-MPI-Reference.html#FFTW-MPI-Reference" title="FFTW MPI Reference"> +<link href="http://www.gnu.org/software/texinfo/" rel="generator-home" title="Texinfo Homepage"> +<!-- +This manual is for FFTW +(version 3.3.3, 25 November 2012). + +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. + --> +<meta http-equiv="Content-Style-Type" content="text/css"> +<style type="text/css"><!-- + pre.display { font-family:inherit } + pre.format { font-family:inherit } + pre.smalldisplay { font-family:inherit; font-size:smaller } + pre.smallformat { font-family:inherit; font-size:smaller } + pre.smallexample { font-size:smaller } + pre.smalllisp { font-size:smaller } + span.sc { font-variant:small-caps } + span.roman { font-family:serif; font-weight:normal; } + span.sansserif { font-family:sans-serif; font-weight:normal; } +--></style> +</head> +<body> +<div class="node"> +<a name="Combining-MPI-and-Threads"></a> +<p> +Next: <a rel="next" accesskey="n" href="FFTW-MPI-Reference.html#FFTW-MPI-Reference">FFTW MPI Reference</a>, +Previous: <a rel="previous" accesskey="p" href="FFTW-MPI-Performance-Tips.html#FFTW-MPI-Performance-Tips">FFTW MPI Performance Tips</a>, +Up: <a rel="up" accesskey="u" href="Distributed_002dmemory-FFTW-with-MPI.html#Distributed_002dmemory-FFTW-with-MPI">Distributed-memory FFTW with MPI</a> +<hr> +</div> + +<h3 class="section">6.11 Combining MPI and Threads</h3> + +<p><a name="index-threads-427"></a> +In certain cases, it may be advantageous to combine MPI +(distributed-memory) and threads (shared-memory) parallelization. +FFTW supports this, with certain caveats. For example, if you have a +cluster of 4-processor shared-memory nodes, you may want to use +threads within the nodes and MPI between the nodes, instead of MPI for +all parallelization. + + <p>In particular, it is possible to seamlessly combine the MPI FFTW +routines with the multi-threaded FFTW routines (see <a href="Multi_002dthreaded-FFTW.html#Multi_002dthreaded-FFTW">Multi-threaded FFTW</a>). However, some care must be taken in the initialization code, +which should look something like this: + +<pre class="example"> int threads_ok; + + int main(int argc, char **argv) + { + int provided; + MPI_Init_thread(&argc, &argv, MPI_THREAD_FUNNELED, &provided); + threads_ok = provided >= MPI_THREAD_FUNNELED; + + if (threads_ok) threads_ok = fftw_init_threads(); + fftw_mpi_init(); + + ... + if (threads_ok) fftw_plan_with_nthreads(...); + ... + + MPI_Finalize(); + } +</pre> + <p><a name="index-fftw_005fmpi_005finit-428"></a><a name="index-fftw_005finit_005fthreads-429"></a><a name="index-fftw_005fplan_005fwith_005fnthreads-430"></a> +First, note that instead of calling <code>MPI_Init</code>, you should call +<code>MPI_Init_threads</code>, which is the initialization routine defined +by the MPI-2 standard to indicate to MPI that your program will be +multithreaded. We pass <code>MPI_THREAD_FUNNELED</code>, which indicates +that we will only call MPI routines from the main thread. (FFTW will +launch additional threads internally, but the extra threads will not +call MPI code.) (You may also pass <code>MPI_THREAD_SERIALIZED</code> or +<code>MPI_THREAD_MULTIPLE</code>, which requests additional multithreading +support from the MPI implementation, but this is not required by +FFTW.) The <code>provided</code> parameter returns what level of threads +support is actually supported by your MPI implementation; this +<em>must</em> be at least <code>MPI_THREAD_FUNNELED</code> if you want to call +the FFTW threads routines, so we define a global variable +<code>threads_ok</code> to record this. You should only call +<code>fftw_init_threads</code> or <code>fftw_plan_with_nthreads</code> if +<code>threads_ok</code> is true. For more information on thread safety in +MPI, see the +<a href="http://www.mpi-forum.org/docs/mpi-20-html/node162.htm">MPI and Threads</a> section of the MPI-2 standard. +<a name="index-thread-safety-431"></a> + + <p>Second, we must call <code>fftw_init_threads</code> <em>before</em> +<code>fftw_mpi_init</code>. This is critical for technical reasons having +to do with how FFTW initializes its list of algorithms. + + <p>Then, if you call <code>fftw_plan_with_nthreads(N)</code>, <em>every</em> MPI +process will launch (up to) <code>N</code> threads to parallelize its transforms. + + <p>For example, in the hypothetical cluster of 4-processor nodes, you +might wish to launch only a single MPI process per node, and then call +<code>fftw_plan_with_nthreads(4)</code> on each process to use all +processors in the nodes. + + <p>This may or may not be faster than simply using as many MPI processes +as you have processors, however. On the one hand, using threads +within a node eliminates the need for explicit message passing within +the node. On the other hand, FFTW's transpose routines are not +multi-threaded, and this means that the communications that do take +place will not benefit from parallelization within the node. +Moreover, many MPI implementations already have optimizations to +exploit shared memory when it is available, so adding the +multithreaded FFTW on top of this may be superfluous. +<a name="index-transpose-432"></a> +<!-- --> + + </body></html> +