cannam@167: <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
cannam@167: <html>
cannam@167: <!-- This manual is for FFTW
cannam@167: (version 3.3.8, 24 May 2018).
cannam@167:
cannam@167: Copyright (C) 2003 Matteo Frigo.
cannam@167:
cannam@167: Copyright (C) 2003 Massachusetts Institute of Technology.
cannam@167:
cannam@167: Permission is granted to make and distribute verbatim copies of this
cannam@167: manual provided the copyright notice and this permission notice are
cannam@167: preserved on all copies.
cannam@167:
cannam@167: Permission is granted to copy and distribute modified versions of this
cannam@167: manual under the conditions for verbatim copying, provided that the
cannam@167: entire resulting derived work is distributed under the terms of a
cannam@167: permission notice identical to this one.
cannam@167:
cannam@167: Permission is granted to copy and distribute translations of this manual
cannam@167: into another language, under the above conditions for modified versions,
cannam@167: except that this permission notice may be stated in a translation
cannam@167: approved by the Free Software Foundation. -->
cannam@167: <!-- Created by GNU Texinfo 6.3, http://www.gnu.org/software/texinfo/ -->
cannam@167: <head>
cannam@167: <title>FFTW 3.3.8: MPI Plan Creation</title>
cannam@167:
cannam@167: <meta name="description" content="FFTW 3.3.8: MPI Plan Creation">
cannam@167: <meta name="keywords" content="FFTW 3.3.8: MPI Plan Creation">
cannam@167: <meta name="resource-type" content="document">
cannam@167: <meta name="distribution" content="global">
cannam@167: <meta name="Generator" content="makeinfo">
cannam@167: <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
cannam@167: <link href="index.html#Top" rel="start" title="Top">
cannam@167: <link href="Concept-Index.html#Concept-Index" rel="index" title="Concept Index">
cannam@167: <link href="index.html#SEC_Contents" rel="contents" title="Table of Contents">
cannam@167: <link href="FFTW-MPI-Reference.html#FFTW-MPI-Reference" rel="up" title="FFTW MPI Reference">
cannam@167: <link href="MPI-Wisdom-Communication.html#MPI-Wisdom-Communication" rel="next" title="MPI Wisdom Communication">
cannam@167: <link href="MPI-Data-Distribution-Functions.html#MPI-Data-Distribution-Functions" rel="prev" title="MPI Data Distribution Functions">
cannam@167: <style type="text/css">
cannam@167: <!--
cannam@167: a.summary-letter {text-decoration: none}
cannam@167: blockquote.indentedblock {margin-right: 0em}
cannam@167: blockquote.smallindentedblock {margin-right: 0em; font-size: smaller}
cannam@167: blockquote.smallquotation {font-size: smaller}
cannam@167: div.display {margin-left: 3.2em}
cannam@167: div.example {margin-left: 3.2em}
cannam@167: div.lisp {margin-left: 3.2em}
cannam@167: div.smalldisplay {margin-left: 3.2em}
cannam@167: div.smallexample {margin-left: 3.2em}
cannam@167: div.smalllisp {margin-left: 3.2em}
cannam@167: kbd {font-style: oblique}
cannam@167: pre.display {font-family: inherit}
cannam@167: pre.format {font-family: inherit}
cannam@167: pre.menu-comment {font-family: serif}
cannam@167: pre.menu-preformatted {font-family: serif}
cannam@167: pre.smalldisplay {font-family: inherit; font-size: smaller}
cannam@167: pre.smallexample {font-size: smaller}
cannam@167: pre.smallformat {font-family: inherit; font-size: smaller}
cannam@167: pre.smalllisp {font-size: smaller}
cannam@167: span.nolinebreak {white-space: nowrap}
cannam@167: span.roman {font-family: initial; font-weight: normal}
cannam@167: span.sansserif {font-family: sans-serif; font-weight: normal}
cannam@167: ul.no-bullet {list-style: none}
cannam@167: -->
cannam@167: </style>
cannam@167:
cannam@167:
cannam@167: </head>
cannam@167:
cannam@167: <body lang="en">
cannam@167: <a name="MPI-Plan-Creation"></a>
cannam@167: <div class="header">
cannam@167: <p>
cannam@167: Next: <a href="MPI-Wisdom-Communication.html#MPI-Wisdom-Communication" accesskey="n" rel="next">MPI Wisdom Communication</a>, Previous: <a href="MPI-Data-Distribution-Functions.html#MPI-Data-Distribution-Functions" accesskey="p" rel="prev">MPI Data Distribution Functions</a>, Up: <a href="FFTW-MPI-Reference.html#FFTW-MPI-Reference" accesskey="u" rel="up">FFTW MPI Reference</a> [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Concept-Index.html#Concept-Index" title="Index" rel="index">Index</a>]</p>
cannam@167: </div>
cannam@167: <hr>
cannam@167: <a name="MPI-Plan-Creation-1"></a>
cannam@167: <h4 class="subsection">6.12.5 MPI Plan Creation</h4>
cannam@167:
cannam@167: <a name="Complex_002ddata-MPI-DFTs"></a>
cannam@167: <h4 class="subsubheading">Complex-data MPI DFTs</h4>
cannam@167:
cannam@167: <p>Plans for complex-data DFTs (see <a href="2d-MPI-example.html#g_t2d-MPI-example">2d MPI example</a>) are created by:
cannam@167: </p>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005f1d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005f2d-1"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005f3d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fmany_005fdft"></a>
cannam@167: <div class="example">
cannam@167: <pre class="example">fftw_plan fftw_mpi_plan_dft_1d(ptrdiff_t n0, fftw_complex *in, fftw_complex *out,
cannam@167: MPI_Comm comm, int sign, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: fftw_complex *in, fftw_complex *out,
cannam@167: MPI_Comm comm, int sign, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_3d(ptrdiff_t n0, ptrdiff_t n1, ptrdiff_t n2,
cannam@167: fftw_complex *in, fftw_complex *out,
cannam@167: MPI_Comm comm, int sign, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft(int rnk, const ptrdiff_t *n,
cannam@167: fftw_complex *in, fftw_complex *out,
cannam@167: MPI_Comm comm, int sign, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_many_dft(int rnk, const ptrdiff_t *n,
cannam@167: ptrdiff_t howmany, ptrdiff_t block, ptrdiff_t tblock,
cannam@167: fftw_complex *in, fftw_complex *out,
cannam@167: MPI_Comm comm, int sign, unsigned flags);
cannam@167: </pre></div>
cannam@167:
cannam@167: <a name="index-MPI-communicator-2"></a>
cannam@167: <a name="index-collective-function-4"></a>
cannam@167: <p>These are similar to their serial counterparts (see <a href="Complex-DFTs.html#Complex-DFTs">Complex DFTs</a>)
cannam@167: in specifying the dimensions, sign, and flags of the transform. The
cannam@167: <code>comm</code> argument gives an MPI communicator that specifies the set
cannam@167: of processes to participate in the transform; plan creation is a
cannam@167: collective function that must be called for all processes in the
cannam@167: communicator. The <code>in</code> and <code>out</code> pointers refer only to a
cannam@167: portion of the overall transform data (see <a href="MPI-Data-Distribution.html#MPI-Data-Distribution">MPI Data Distribution</a>)
cannam@167: as specified by the ‘<samp>local_size</samp>’ functions in the previous
cannam@167: section. Unless <code>flags</code> contains <code>FFTW_ESTIMATE</code>, these
cannam@167: arrays are overwritten during plan creation as for the serial
cannam@167: interface. For multi-dimensional transforms, any dimensions <code>>
cannam@167: 1</code> are supported; for one-dimensional transforms, only composite
cannam@167: (non-prime) <code>n0</code> are currently supported (unlike the serial
cannam@167: FFTW). Requesting an unsupported transform size will yield a
cannam@167: <code>NULL</code> plan. (As in the serial interface, highly composite sizes
cannam@167: generally yield the best performance.)
cannam@167: </p>
cannam@167: <a name="index-advanced-interface-6"></a>
cannam@167: <a name="index-FFTW_005fMPI_005fDEFAULT_005fBLOCK-2"></a>
cannam@167: <a name="index-stride-3"></a>
cannam@167: <p>The advanced-interface <code>fftw_mpi_plan_many_dft</code> additionally
cannam@167: allows you to specify the block sizes for the first dimension
cannam@167: (<code>block</code>) of the n<sub>0</sub> × n<sub>1</sub> × n<sub>2</sub> × … × n<sub>d-1</sub>
cannam@167: input data and the first dimension
cannam@167: (<code>tblock</code>) of the n<sub>1</sub> × n<sub>0</sub> × n<sub>2</sub> ×…× n<sub>d-1</sub>
cannam@167: transposed data (at intermediate
cannam@167: steps of the transform, and for the output if
cannam@167: <code>FFTW_TRANSPOSED_OUT</code> is specified in <code>flags</code>). These must
cannam@167: be the same block sizes as were passed to the corresponding
cannam@167: ‘<samp>local_size</samp>’ function; you can pass <code>FFTW_MPI_DEFAULT_BLOCK</code>
cannam@167: to use FFTW’s default block size as in the basic interface. Also, the
cannam@167: <code>howmany</code> parameter specifies that the transform is of contiguous
cannam@167: <code>howmany</code>-tuples rather than individual complex numbers; this
cannam@167: corresponds to the same parameter in the serial advanced interface
cannam@167: (see <a href="Advanced-Complex-DFTs.html#Advanced-Complex-DFTs">Advanced Complex DFTs</a>) with <code>stride = howmany</code> and
cannam@167: <code>dist = 1</code>.
cannam@167: </p>
cannam@167: <a name="MPI-flags"></a>
cannam@167: <h4 class="subsubheading">MPI flags</h4>
cannam@167:
cannam@167: <p>The <code>flags</code> can be any of those for the serial FFTW
cannam@167: (see <a href="Planner-Flags.html#Planner-Flags">Planner Flags</a>), and in addition may include one or more of
cannam@167: the following MPI-specific flags, which improve performance at the
cannam@167: cost of changing the output or input data formats.
cannam@167: </p>
cannam@167: <ul>
cannam@167: <li> <a name="index-FFTW_005fMPI_005fSCRAMBLED_005fOUT-2"></a>
cannam@167: <a name="index-FFTW_005fMPI_005fSCRAMBLED_005fIN-2"></a>
cannam@167: <code>FFTW_MPI_SCRAMBLED_OUT</code>, <code>FFTW_MPI_SCRAMBLED_IN</code>: valid for
cannam@167: 1d transforms only, these flags indicate that the output/input of the
cannam@167: transform are in an undocumented “scrambled” order. A forward
cannam@167: <code>FFTW_MPI_SCRAMBLED_OUT</code> transform can be inverted by a backward
cannam@167: <code>FFTW_MPI_SCRAMBLED_IN</code> (times the usual 1/<i>N</i> normalization).
cannam@167: See <a href="One_002ddimensional-distributions.html#One_002ddimensional-distributions">One-dimensional distributions</a>.
cannam@167:
cannam@167: </li><li> <a name="index-FFTW_005fMPI_005fTRANSPOSED_005fOUT-2"></a>
cannam@167: <a name="index-FFTW_005fMPI_005fTRANSPOSED_005fIN-2"></a>
cannam@167: <code>FFTW_MPI_TRANSPOSED_OUT</code>, <code>FFTW_MPI_TRANSPOSED_IN</code>: valid
cannam@167: for multidimensional (<code>rnk > 1</code>) transforms only, these flags
cannam@167: specify that the output or input of an n<sub>0</sub> × n<sub>1</sub> × n<sub>2</sub> × … × n<sub>d-1</sub>
cannam@167: transform is
cannam@167: transposed to n<sub>1</sub> × n<sub>0</sub> × n<sub>2</sub> ×…× n<sub>d-1</sub>
cannam@167: . See <a href="Transposed-distributions.html#Transposed-distributions">Transposed distributions</a>.
cannam@167:
cannam@167: </li></ul>
cannam@167:
cannam@167: <a name="Real_002ddata-MPI-DFTs"></a>
cannam@167: <h4 class="subsubheading">Real-data MPI DFTs</h4>
cannam@167:
cannam@167: <a name="index-r2c-4"></a>
cannam@167: <p>Plans for real-input/output (r2c/c2r) DFTs (see <a href="Multi_002ddimensional-MPI-DFTs-of-Real-Data.html#Multi_002ddimensional-MPI-DFTs-of-Real-Data">Multi-dimensional MPI DFTs of Real Data</a>) are created by:
cannam@167: </p>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fr2c_005f2d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fr2c_005f2d-1"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fr2c_005f3d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fr2c"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fc2r_005f2d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fc2r_005f2d-1"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fc2r_005f3d"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fdft_005fc2r"></a>
cannam@167: <div class="example">
cannam@167: <pre class="example">fftw_plan fftw_mpi_plan_dft_r2c_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: double *in, fftw_complex *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_r2c_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: double *in, fftw_complex *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_r2c_3d(ptrdiff_t n0, ptrdiff_t n1, ptrdiff_t n2,
cannam@167: double *in, fftw_complex *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_r2c(int rnk, const ptrdiff_t *n,
cannam@167: double *in, fftw_complex *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_c2r_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: fftw_complex *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_c2r_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: fftw_complex *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_c2r_3d(ptrdiff_t n0, ptrdiff_t n1, ptrdiff_t n2,
cannam@167: fftw_complex *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_dft_c2r(int rnk, const ptrdiff_t *n,
cannam@167: fftw_complex *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: </pre></div>
cannam@167:
cannam@167: <p>Similar to the serial interface (see <a href="Real_002ddata-DFTs.html#Real_002ddata-DFTs">Real-data DFTs</a>), these
cannam@167: transform logically n<sub>0</sub> × n<sub>1</sub> × n<sub>2</sub> × … × n<sub>d-1</sub>
cannam@167: real data to/from n<sub>0</sub> × n<sub>1</sub> × n<sub>2</sub> × … × (n<sub>d-1</sub>/2 + 1)
cannam@167: complex
cannam@167: data, representing the non-redundant half of the conjugate-symmetry
cannam@167: output of a real-input DFT (see <a href="Multi_002ddimensional-Transforms.html#Multi_002ddimensional-Transforms">Multi-dimensional Transforms</a>).
cannam@167: However, the real array must be stored within a padded n<sub>0</sub> × n<sub>1</sub> × n<sub>2</sub> × … × [2 (n<sub>d-1</sub>/2 + 1)]
cannam@167:
cannam@167: array (much like the in-place serial r2c transforms, but here for
cannam@167: out-of-place transforms as well). Currently, only multi-dimensional
cannam@167: (<code>rnk > 1</code>) r2c/c2r transforms are supported (requesting a plan
cannam@167: for <code>rnk = 1</code> will yield <code>NULL</code>). As explained above
cannam@167: (see <a href="Multi_002ddimensional-MPI-DFTs-of-Real-Data.html#Multi_002ddimensional-MPI-DFTs-of-Real-Data">Multi-dimensional MPI DFTs of Real Data</a>), the data
cannam@167: distribution of both the real and complex arrays is given by the
cannam@167: ‘<samp>local_size</samp>’ function called for the dimensions of the
cannam@167: <em>complex</em> array. Similar to the other planning functions, the
cannam@167: input and output arrays are overwritten when the plan is created
cannam@167: except in <code>FFTW_ESTIMATE</code> mode.
cannam@167: </p>
cannam@167: <p>As for the complex DFTs above, there is an advance interface that
cannam@167: allows you to manually specify block sizes and to transform contiguous
cannam@167: <code>howmany</code>-tuples of real/complex numbers:
cannam@167: </p>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fmany_005fdft_005fr2c"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fmany_005fdft_005fc2r"></a>
cannam@167: <div class="example">
cannam@167: <pre class="example">fftw_plan fftw_mpi_plan_many_dft_r2c
cannam@167: (int rnk, const ptrdiff_t *n, ptrdiff_t howmany,
cannam@167: ptrdiff_t iblock, ptrdiff_t oblock,
cannam@167: double *in, fftw_complex *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_many_dft_c2r
cannam@167: (int rnk, const ptrdiff_t *n, ptrdiff_t howmany,
cannam@167: ptrdiff_t iblock, ptrdiff_t oblock,
cannam@167: fftw_complex *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: </pre></div>
cannam@167:
cannam@167: <a name="MPI-r2r-transforms"></a>
cannam@167: <h4 class="subsubheading">MPI r2r transforms</h4>
cannam@167:
cannam@167: <a name="index-r2r-4"></a>
cannam@167: <p>There are corresponding plan-creation routines for r2r
cannam@167: transforms (see <a href="More-DFTs-of-Real-Data.html#More-DFTs-of-Real-Data">More DFTs of Real Data</a>), currently supporting
cannam@167: multidimensional (<code>rnk > 1</code>) transforms only (<code>rnk = 1</code> will
cannam@167: yield a <code>NULL</code> plan):
cannam@167: </p>
cannam@167: <div class="example">
cannam@167: <pre class="example">fftw_plan fftw_mpi_plan_r2r_2d(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: double *in, double *out,
cannam@167: MPI_Comm comm,
cannam@167: fftw_r2r_kind kind0, fftw_r2r_kind kind1,
cannam@167: unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_r2r_3d(ptrdiff_t n0, ptrdiff_t n1, ptrdiff_t n2,
cannam@167: double *in, double *out,
cannam@167: MPI_Comm comm,
cannam@167: fftw_r2r_kind kind0, fftw_r2r_kind kind1, fftw_r2r_kind kind2,
cannam@167: unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_r2r(int rnk, const ptrdiff_t *n,
cannam@167: double *in, double *out,
cannam@167: MPI_Comm comm, const fftw_r2r_kind *kind,
cannam@167: unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_many_r2r(int rnk, const ptrdiff_t *n,
cannam@167: ptrdiff_t iblock, ptrdiff_t oblock,
cannam@167: double *in, double *out,
cannam@167: MPI_Comm comm, const fftw_r2r_kind *kind,
cannam@167: unsigned flags);
cannam@167: </pre></div>
cannam@167:
cannam@167: <p>The parameters are much the same as for the complex DFTs above, except
cannam@167: that the arrays are of real numbers (and hence the outputs of the
cannam@167: ‘<samp>local_size</samp>’ data-distribution functions should be interpreted as
cannam@167: counts of real rather than complex numbers). Also, the <code>kind</code>
cannam@167: parameters specify the r2r kinds along each dimension as for the
cannam@167: serial interface (see <a href="Real_002dto_002dReal-Transform-Kinds.html#Real_002dto_002dReal-Transform-Kinds">Real-to-Real Transform Kinds</a>). See <a href="Other-Multi_002ddimensional-Real_002ddata-MPI-Transforms.html#Other-Multi_002ddimensional-Real_002ddata-MPI-Transforms">Other Multi-dimensional Real-data MPI Transforms</a>.
cannam@167: </p>
cannam@167: <a name="MPI-transposition"></a>
cannam@167: <h4 class="subsubheading">MPI transposition</h4>
cannam@167: <a name="index-transpose-5"></a>
cannam@167:
cannam@167: <p>FFTW also provides routines to plan a transpose of a distributed
cannam@167: <code>n0</code> by <code>n1</code> array of real numbers, or an array of
cannam@167: <code>howmany</code>-tuples of real numbers with specified block sizes
cannam@167: (see <a href="FFTW-MPI-Transposes.html#FFTW-MPI-Transposes">FFTW MPI Transposes</a>):
cannam@167: </p>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005ftranspose-1"></a>
cannam@167: <a name="index-fftw_005fmpi_005fplan_005fmany_005ftranspose-1"></a>
cannam@167: <div class="example">
cannam@167: <pre class="example">fftw_plan fftw_mpi_plan_transpose(ptrdiff_t n0, ptrdiff_t n1,
cannam@167: double *in, double *out,
cannam@167: MPI_Comm comm, unsigned flags);
cannam@167: fftw_plan fftw_mpi_plan_many_transpose
cannam@167: (ptrdiff_t n0, ptrdiff_t n1, ptrdiff_t howmany,
cannam@167: ptrdiff_t block0, ptrdiff_t block1,
cannam@167: double *in, double *out, MPI_Comm comm, unsigned flags);
cannam@167: </pre></div>
cannam@167:
cannam@167: <a name="index-new_002darray-execution-2"></a>
cannam@167: <a name="index-fftw_005fmpi_005fexecute_005fr2r-1"></a>
cannam@167: <p>These plans are used with the <code>fftw_mpi_execute_r2r</code> new-array
cannam@167: execute function (see <a href="Using-MPI-Plans.html#Using-MPI-Plans">Using MPI Plans</a>), since they count as (rank
cannam@167: zero) r2r plans from FFTW’s perspective.
cannam@167: </p>
cannam@167: <hr>
cannam@167: <div class="header">
cannam@167: <p>
cannam@167: Next: <a href="MPI-Wisdom-Communication.html#MPI-Wisdom-Communication" accesskey="n" rel="next">MPI Wisdom Communication</a>, Previous: <a href="MPI-Data-Distribution-Functions.html#MPI-Data-Distribution-Functions" accesskey="p" rel="prev">MPI Data Distribution Functions</a>, Up: <a href="FFTW-MPI-Reference.html#FFTW-MPI-Reference" accesskey="u" rel="up">FFTW MPI Reference</a> [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Concept-Index.html#Concept-Index" title="Index" rel="index">Index</a>]</p>
cannam@167: </div>
cannam@167:
cannam@167:
cannam@167:
cannam@167: </body>
cannam@167: </html>