root / src / pauxil / HPL_indxg2l.c @ 1
Historique | Voir | Annoter | Télécharger (6,84 ko)
| 1 |
/*
|
|---|---|
| 2 |
* -- High Performance Computing Linpack Benchmark (HPL)
|
| 3 |
* HPL - 2.0 - September 10, 2008
|
| 4 |
* Antoine P. Petitet
|
| 5 |
* University of Tennessee, Knoxville
|
| 6 |
* Innovative Computing Laboratory
|
| 7 |
* (C) Copyright 2000-2008 All Rights Reserved
|
| 8 |
*
|
| 9 |
* -- Copyright notice and Licensing terms:
|
| 10 |
*
|
| 11 |
* Redistribution and use in source and binary forms, with or without
|
| 12 |
* modification, are permitted provided that the following conditions
|
| 13 |
* are met:
|
| 14 |
*
|
| 15 |
* 1. Redistributions of source code must retain the above copyright
|
| 16 |
* notice, this list of conditions and the following disclaimer.
|
| 17 |
*
|
| 18 |
* 2. Redistributions in binary form must reproduce the above copyright
|
| 19 |
* notice, this list of conditions, and the following disclaimer in the
|
| 20 |
* documentation and/or other materials provided with the distribution.
|
| 21 |
*
|
| 22 |
* 3. All advertising materials mentioning features or use of this
|
| 23 |
* software must display the following acknowledgement:
|
| 24 |
* This product includes software developed at the University of
|
| 25 |
* Tennessee, Knoxville, Innovative Computing Laboratory.
|
| 26 |
*
|
| 27 |
* 4. The name of the University, the name of the Laboratory, or the
|
| 28 |
* names of its contributors may not be used to endorse or promote
|
| 29 |
* products derived from this software without specific written
|
| 30 |
* permission.
|
| 31 |
*
|
| 32 |
* -- Disclaimer:
|
| 33 |
*
|
| 34 |
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
| 35 |
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
| 36 |
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
| 37 |
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE UNIVERSITY
|
| 38 |
* OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
| 39 |
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
| 40 |
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
| 41 |
* DATA OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
| 42 |
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
| 43 |
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
| 44 |
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
| 45 |
* ---------------------------------------------------------------------
|
| 46 |
*/
|
| 47 |
/*
|
| 48 |
* Include files
|
| 49 |
*/
|
| 50 |
#include "hpl.h" |
| 51 |
|
| 52 |
#ifdef STDC_HEADERS
|
| 53 |
int HPL_indxg2l
|
| 54 |
( |
| 55 |
const int IG, |
| 56 |
const int INB, |
| 57 |
const int NB, |
| 58 |
const int SRCPROC, |
| 59 |
const int NPROCS |
| 60 |
) |
| 61 |
#else
|
| 62 |
int HPL_indxg2l
|
| 63 |
( IG, INB, NB, SRCPROC, NPROCS ) |
| 64 |
const int IG; |
| 65 |
const int INB; |
| 66 |
const int NB; |
| 67 |
const int SRCPROC; |
| 68 |
const int NPROCS; |
| 69 |
#endif
|
| 70 |
{
|
| 71 |
/*
|
| 72 |
* Purpose
|
| 73 |
* =======
|
| 74 |
*
|
| 75 |
* HPL_indxg2l computes the local index of a matrix entry pointed to by
|
| 76 |
* the global index IG. This local returned index is the same in all
|
| 77 |
* processes.
|
| 78 |
*
|
| 79 |
* Arguments
|
| 80 |
* =========
|
| 81 |
*
|
| 82 |
* IG (input) const int
|
| 83 |
* On entry, IG specifies the global index of the matrix entry.
|
| 84 |
* IG must be at least zero.
|
| 85 |
*
|
| 86 |
* INB (input) const int
|
| 87 |
* On entry, INB specifies the size of the first block of the
|
| 88 |
* global matrix. INB must be at least one.
|
| 89 |
*
|
| 90 |
* NB (input) const int
|
| 91 |
* On entry, NB specifies the blocking factor used to partition
|
| 92 |
* and distribute the matrix. NB must be larger than one.
|
| 93 |
*
|
| 94 |
* SRCPROC (input) const int
|
| 95 |
* On entry, if SRCPROC = -1, the data is not distributed but
|
| 96 |
* replicated, in which case this routine returns IG in all
|
| 97 |
* processes. Otherwise, the value of SRCPROC is ignored.
|
| 98 |
*
|
| 99 |
* NPROCS (input) const int
|
| 100 |
* On entry, NPROCS specifies the total number of process rows
|
| 101 |
* or columns over which the matrix is distributed. NPROCS must
|
| 102 |
* be at least one.
|
| 103 |
*
|
| 104 |
* ---------------------------------------------------------------------
|
| 105 |
*/
|
| 106 |
/*
|
| 107 |
* .. Local Variables ..
|
| 108 |
*/
|
| 109 |
int i, j;
|
| 110 |
/* ..
|
| 111 |
* .. Executable Statements ..
|
| 112 |
*/
|
| 113 |
if( ( IG < INB ) || ( SRCPROC == -1 ) || ( NPROCS == 1 ) ) |
| 114 |
/*
|
| 115 |
* IG belongs to the first block, or the data is not distributed, or
|
| 116 |
* there is just one process in this dimension of the grid.
|
| 117 |
*/
|
| 118 |
return( IG );
|
| 119 |
/*
|
| 120 |
* IG = INB - NB + ( l * NPROCS + MYROC ) * NB + X with 0 <= X < NB,
|
| 121 |
* thus IG is to be found in the block (IG-INB+NB) / NB = l*NPROCS+MYROC
|
| 122 |
* with 0 <= MYROC < NPROCS. The local index to be returned depends on
|
| 123 |
* whether IG resides in the process owning the first partial block of
|
| 124 |
* size INB (MYROC=0). To determine this cheaply, let i = (IG-INB) / NB,
|
| 125 |
* so that if NPROCS divides i+1, i.e. MYROC=0, we have i+1 = l*NPROCS.
|
| 126 |
* If we set j = i / NPROCS, it follows that j = l-1. Therefore, i+1 is
|
| 127 |
* equal to (j+1) * NPROCS. Conversely, if NPROCS does not divide i+1,
|
| 128 |
* then i+1 = l*NPROCS + MYROC with 1 <= MYROC < NPROCS. It follows that
|
| 129 |
* j=l and thus (j+1)*NPROCS > i+1.
|
| 130 |
*/
|
| 131 |
j = ( i = ( IG - INB ) / NB ) / NPROCS; |
| 132 |
/*
|
| 133 |
* When IG resides in the process owning the first partial block of size
|
| 134 |
* INB (MYROC = 0), then the result IL can be written as:
|
| 135 |
* IL = INB - NB + l * NB + X = IG + ( l - (l * NPROCS + MYROC) ) * NB.
|
| 136 |
* Using the above notation, we have i+1 = l*NPROCS + MYROC = l*NPROCS,
|
| 137 |
* i.e l = ( i+1 ) / NPROCS = j+1, since NPROCS divides i+1, therefore
|
| 138 |
* IL = IG + ( j + 1 - ( i + 1 ) ) * NB.
|
| 139 |
*
|
| 140 |
* Otherwise when MYROC >= 1, the result IL can be written as:
|
| 141 |
* IL = l * NB + X = IG - INB + ( ( l+1 ) - ( l * NPROCS + MYROC ) )*NB.
|
| 142 |
* We still have i+1 = l*NPROCS+MYROC. Since NPROCS does not divide i+1,
|
| 143 |
* we have j = (l*NPROCS+MYROC-1) / NPROCS = l, i.e
|
| 144 |
* IL = IG - INB + ( j + 1 - ( i + 1 ) ) * NB.
|
| 145 |
*/
|
| 146 |
return( NB * (j - i) +
|
| 147 |
( ( i + 1 - ( j + 1 )*NPROCS ) ? IG - INB : IG ) ); |
| 148 |
/*
|
| 149 |
* End of HPL_indxg2l
|
| 150 |
*/
|
| 151 |
} |