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
|
REQUIREMENTS:
- ADIOS 1.4.1: http://www.olcf.ornl.gov/center-projects/adios
- Fortran 90 compiler
- MPI
BUILD:
- Install ADIOS
- In Makefile, set ADIOS_DIR to the installation directory
- also set the MPI/F90 compiler name (e.g. mpif90 or ftn)
- run make
1. Run the writer
=================
2x2 processes write 4x5 blocks (8x10 global array), 5 timesteps
$ mpirun -np 4 ./coupling_writer_2D 2 2 4 5 5
Process number : 2 x 2
Array size per process: 4 x 5
rank=0: 2D array pos: 0,0 offset: 0,0
rank=1: 2D array pos: 1,0 offset: 4,0
rank=2: 2D array pos: 0,1 offset: 0,5
rank=3: 2D array pos: 1,1 offset: 4,5
rank=1 set matrix to 1.01
rank=0 set matrix to 0.01
Output file: writer_001.bp
rank=2 set matrix to 2.01
rank=3 set matrix to 3.01
rank=1: ---------------------- write completed ----------------------
rank=2: ---------------------- write completed ----------------------
rank=3: ---------------------- write completed ----------------------
rank=0: ---------------------- write completed ----------------------
...
2. Check the output
===================
$ bpls -l writer.bp
integer /info/nproc 5*scalar = 4 / 4
integer /info/npx 5*scalar = 2 / 2
integer /info/npy 5*scalar = 2 / 2
integer /gdx 5*scalar = 8 / 8
integer /gdy 5*scalar = 10 / 10
integer /aux/ox 5*scalar = 0 / 0
integer /aux/oy 5*scalar = 0 / 0
integer /aux/ldx 5*scalar = 4 / 4
integer /aux/ldy 5*scalar = 5 / 5
double /xy 5*{10, 8} = 0.01 / 3.05 / 1.53 / 1.11812
Note that F90 app writes 8x10 arrays in Fortran column-order,
which is 10x8 in C row-order. bpls sees/shows the C ordering
because it is a C program. The reader below sees 8x10 Fortran
array because it is a Fortan program.
3. Run the reader
=================
$ mpirun -np 3 ./coupling_reader_2D 1 3 MPI 1 1
Process number : 1 x 3
Method for reading: 0
Input file: writer.bp
rank=0: Read in gdx and gdy, step1 from writer.bp
rank=2: Read in gdx and gdy, step1 from writer.bp
rank=1: Read in gdx and gdy, step1 from writer.bp
rank=0: Read in xy(1:2,1:10) from writer.bp
rank=2: Read in xy(5:8,1:10) from writer.bp
rank=1: Read in xy(3:4,1:10) from writer.bp
rank=1: Collectively wrote xy and new xy2 to reader_001.bp
rank=2: Collectively wrote xy and new xy2 to reader_001.bp
rank=0: Collectively wrote xy and new xy2 to reader_001.bp
rank=2: Read in gdx and gdy, step2 from writer.bp
rank=0: Read in gdx and gdy, step2 from writer.bp
rank=1: Read in gdx and gdy, step2 from writer.bp
rank=0: Read in xy(1:2,1:10) from writer.bp
rank=2: Read in xy(5:8,1:10) from writer.bp
rank=1: Read in xy(3:4,1:10) from writer.bp
rank=2: Collectively wrote xy and new xy2 to reader_002.bp
rank=1: Collectively wrote xy and new xy2 to reader_002.bp
rank=0: Collectively wrote xy and new xy2 to reader_002.bp
...
4. Check the output
===================
$ bpls -l reader_001.bp
integer /info/rank scalar = 0
integer /gdx scalar = 8
integer /gdy scalar = 10
integer /aux/ldx scalar = 2
integer /aux/ldy scalar = 10
integer /aux/ox scalar = 0
integer /aux/oy scalar = 0
double /xy {10, 8} = 0.01 / 3.01 / 1.51 / 1.11803
double /xy2 {10, 8} = 0.0001 / 2.0301 / 1.2651 / 0.833742
xy in the two files should match:
$ bpls -l reader_001.bp -d xy -n 8
double /xy {10, 8} = 0.01 / 3.01 / 1.51 / 1.11803
(0,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(1,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(2,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(3,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(4,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(5,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(6,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(7,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(8,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(9,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
$ bpls -l writer.bp -d xy -n 8 -c "1,-1,-1"
double /xy 5*{10, 8} = 0.01 / 3.05 / 1.53 / 1.11812
slice (0:0, 0:9, 0:7)
(0,0,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(0,1,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(0,2,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(0,3,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(0,4,0) 0.01 0.01 0.01 0.01 1.01 1.01 1.01 1.01
(0,5,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(0,6,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(0,7,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(0,8,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
(0,9,0) 2.01 2.01 2.01 2.01 3.01 3.01 3.01 3.01
|
