1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
|
/*
* Copyright (C) 2010. See COPYRIGHT in top-level directory.
*/
/* One-Sided MPI 2-D Strided Get Test
*
* Author: James Dinan <dinan@mcs.anl.gov>
* Date : December, 2010
*
* This code performs N strided get operations from a 2d patch of a shared
* array. The array has dimensions [X, Y] and the subarray has dimensions
* [SUB_X, SUB_Y] and begins at index [0, 0]. The input and output buffers are
* specified using an MPI indexed type.
*/
#include <stdio.h>
#include <stdlib.h>
#include <mpi.h>
#define XDIM 8
#define YDIM 1024
#define SUB_XDIM 8
#define SUB_YDIM 256
int main(int argc, char **argv) {
int i, j, rank, nranks, peer, bufsize, errors;
double *win_buf, *loc_buf;
MPI_Win buf_win;
MPI_Aint idx_loc[SUB_YDIM];
int idx_rem[SUB_YDIM];
int blk_len[SUB_YDIM];
MPI_Datatype loc_type, rem_type;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
MPI_Comm_size(MPI_COMM_WORLD, &nranks);
bufsize = XDIM * YDIM * sizeof(double);
MPI_Alloc_mem(bufsize, MPI_INFO_NULL, &win_buf);
MPI_Alloc_mem(bufsize, MPI_INFO_NULL, &loc_buf);
if (rank == 0)
printf("MPI RMA Strided Get Test:\n");
for (i = 0; i < XDIM*YDIM; i++)
*(win_buf + i) = 1.0 + rank;
MPI_Win_create(win_buf, bufsize, 1, MPI_INFO_NULL, MPI_COMM_WORLD, &buf_win);
peer = (rank+1) % nranks;
// Build the datatype
for (i = 0; i < SUB_YDIM; i++) {
MPI_Get_address(&loc_buf[i*XDIM], &idx_loc[i]);
idx_rem[i] = i*XDIM;
blk_len[i] = SUB_XDIM;
}
MPI_Type_indexed(SUB_YDIM, blk_len, idx_rem, MPI_DOUBLE, &loc_type);
MPI_Type_indexed(SUB_YDIM, blk_len, idx_rem, MPI_DOUBLE, &rem_type);
MPI_Type_commit(&loc_type);
MPI_Type_commit(&rem_type);
// Perform get operation
MPI_Win_lock(MPI_LOCK_EXCLUSIVE, peer, 0, buf_win);
MPI_Get(loc_buf, 1, loc_type, peer, 0, 1, rem_type, buf_win);
// Use the datatype only on the remote side (must have SUB_XDIM == XDIM)
// MPI_Get(loc_buf, SUB_XDIM*SUB_YDIM, MPI_DOUBLE, peer, 0, 1, rem_type, buf_win);
MPI_Win_unlock(peer, buf_win);
MPI_Type_free(&loc_type);
MPI_Type_free(&rem_type);
MPI_Barrier(MPI_COMM_WORLD);
// Verify that the results are correct
errors = 0;
for (i = 0; i < SUB_XDIM; i++) {
for (j = 0; j < SUB_YDIM; j++) {
const double actual = *(loc_buf + i + j*XDIM);
const double expected = (1.0 + peer);
if (actual - expected > 1e-10) {
printf("%d: Data validation failed at [%d, %d] expected=%f actual=%f\n",
rank, j, i, expected, actual);
errors++;
fflush(stdout);
}
}
}
for (i = SUB_XDIM; i < XDIM; i++) {
for (j = 0; j < SUB_YDIM; j++) {
const double actual = *(loc_buf + i + j*XDIM);
const double expected = 1.0 + rank;
if (actual - expected > 1e-10) {
printf("%d: Data validation failed at [%d, %d] expected=%f actual=%f\n",
rank, j, i, expected, actual);
errors++;
fflush(stdout);
}
}
}
for (i = 0; i < XDIM; i++) {
for (j = SUB_YDIM; j < YDIM; j++) {
const double actual = *(loc_buf + i + j*XDIM);
const double expected = 1.0 + rank;
if (actual - expected > 1e-10) {
printf("%d: Data validation failed at [%d, %d] expected=%f actual=%f\n",
rank, j, i, expected, actual);
errors++;
fflush(stdout);
}
}
}
MPI_Win_free(&buf_win);
MPI_Free_mem(win_buf);
MPI_Free_mem(loc_buf);
MPI_Finalize();
if (errors == 0) {
printf("%d: Success\n", rank);
return 0;
} else {
printf("%d: Fail\n", rank);
return 1;
}
}
|
