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
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
|
!
! ADIOS is freely available under the terms of the BSD license described
! in the COPYING file in the top level directory of this source distribution.
!
! Copyright (c) 2008 - 2009. UT-BATTELLE, LLC. All rights reserved.
!
!
! COPYARRAY
!
! Write an ADIOS BP file from many processor for test purposes.
!
! nx * ny processes write a 2D array, where each process writes an
! ndx * ndy piece with filling with some values as integer (4 bytes) value
!
! Block written from rank R:
! R*ndx ... R*ndx
! R*ndx+1 ... R*ndx+1
! ... ...
! R*ndx+ndy-1 R*ndx+ndy-1
!
! E.g 2x2 processes writing 2x2 blocks:
! 0 0 | 2 2 0 1 | 2 3
! 1 1 | 3 3 4 5 | 6 7
! ----+---- or -----+----
! 4 4 | 6 6 8 9 | 10 11
! 5 5 | 7 7 12 13| 14 15
!
! (c) Oak Ridge National Laboratory, 2009
! Author: Norbert Podhorszki
!
module copyarray2D_comm
! arguments
character(len=256) :: outputfile, inputfile
integer :: npx, npy ! # of processors in x-y direction
integer :: ndx, ndy ! size of array per processor
integer :: timesteps ! number of timesteps to write
logical :: common_size ! .true. if common local sizes are given as argument
! .false. if we have to read sizes from a file
integer :: gndx, gndy ! size of the global array
integer :: posx, posy ! position index in the array
integer :: offx, offy ! offsets of local array in the global array
integer, dimension(:,:), allocatable :: int_xy
! MPI variables
integer :: group_comm
integer :: rank, nproc
integer :: ierr
! ADIOS variables
character (len=200) :: group
character (len=200) :: filename
!character (len=6) :: nprocstr
integer*8 :: handle, total_size, group_size, adios_totalsize, adios_buf_size
integer :: err
real*8 :: start_time, end_time, total_time,gbs,sz
real*8 :: cache_start_time, cache_end_time, cache_total_time
end module copyarray2D_comm
program genarray
use copyarray2D_comm
use adios_write_mod
implicit none
include 'mpif.h'
call MPI_Init (ierr)
call MPI_Comm_dup (MPI_COMM_WORLD, group_comm, ierr)
call MPI_Comm_rank (MPI_COMM_WORLD, rank, ierr)
call MPI_Comm_size (group_comm, nproc , ierr)
call adios_init ("genarray.xml", group_comm, ierr)
!call MPI_Barrier (group_comm, ierr)
call processArgs()
if (rank == 0) then
print *,"Output file: "//trim(outputfile)
print '(" Process number : ",i0," x ",i0)', npx,npy
if (common_size) then
print '(" Array size per process: ",i0," x ",i0)', ndx,ndy
else
print *," Array sizes per processes taken from file: "//trim(inputfile)
endif
if (nproc .ne. npx*npy) then
print '(" Error: Number of processors ",i0,"does not match ndx*ndy=",i0)', nproc, npx*npy
call exit(1)
endif
endif
!write (*,*) 'rank ', rank, "init completed"
!write (nprocstr,'(i0)') nproc
call determineLocalSize()
call determineGlobalSize()
call determineOffsets()
call generateLocalArray()
call MPI_BARRIER(MPI_COMM_WORLD,err)
start_time = MPI_WTIME()
call readArray()
call MPI_BARRIER(MPI_COMM_WORLD,err)
end_time = MPI_WTIME()
total_time = end_time - start_time
sz = adios_totalsize * nproc/1024.d0/1024.d0/1024.d0 !size in GB
gbs = sz/total_time
!if (rank==0) write(6,*) total_time
if (rank==0) write(6,'(a10,d12.2,2x,d12.2,2x,d12.3)') outputfile,sz,total_time,gbs
call MPI_BARRIER(MPI_COMM_WORLD,err)
start_time = MPI_WTIME()
call writeArray()
call MPI_BARRIER(MPI_COMM_WORLD,err)
end_time = MPI_WTIME()
total_time = end_time - start_time
sz = adios_totalsize * nproc/1024.d0/1024.d0/1024.d0 !size in GB
gbs = sz/total_time
!if (rank==0) write(6,*) total_time
if (rank==0) write(6,'(a10,d12.2,2x,d12.2,2x,d12.3)') outputfile,sz,total_time,gbs
! Terminate
call MPI_Barrier (MPI_COMM_WORLD, ierr)
call adios_finalize (rank, ierr)
call MPI_Finalize (ierr)
end program genarray
!!***************************
subroutine determineLocalSize()
use copyarray2D_comm
implicit none
if (common_size) then
! we are done since we know them from argument
else
! have to read from file
print *, "To be implemented: read sizes from file 1"
call exit(2)
endif
end subroutine determineLocalSize
!!***************************
subroutine determineGlobalSize()
use copyarray2D_comm
implicit none
if (common_size) then
gndx = npx * ndx
gndy = npy * ndy
else
! have to read from file
print *, "To be implemented: read sizes from file 2"
call exit(2)
endif
end subroutine determineGlobalSize
!!***************************
subroutine determineOffsets()
use copyarray2D_comm
implicit none
if (common_size) then
posx = mod(rank, npx) ! 1st dim easy: 0, npx, 2npx... are in the same X position
posy = rank/npx ! 2nd dim: npx processes belong into one dim
offx = posx * ndx
offy = posy * ndy
! print '("rank=",i0," pos: ",i0,",",i0," offset: ",i0,",",i0)', &
! rank, posx, posy, offx, offy
else
! have to read from file
print *, "To be implemented: read sizes from file 2"
call exit(2)
endif
end subroutine determineOffsets
!!***************************
subroutine generateLocalArray()
use copyarray2D_comm
implicit none
integer :: i,j,k, startv
allocate( int_xy(1:ndx, 1:ndy) )
! do j=1,ndy
! startv = (offy+j-1)*gndx + offx
! do i=1,ndx
! !int_xy(i,j) = startv+i-1
! !int_xy(i,j) = rank*ndx+j-1
! int_xy(i,j) = rank
! enddo
! enddo
end subroutine generateLocalArray
!!***************************
subroutine readArray()
use copyarray2D_comm
use adios_write_mod
use adios_read_mod
implicit none
integer :: tstep
character(2) :: mode = "r"
include 'mpif.h'
! Write out data using ADIOS
group = "genarray"
! calculate how much we write from this processor
group_size = 4 * 9 + & ! X,Y, nproc, all size_ and offs_ integers
4 * ndx * ndy + & ! int_xy
4 * ndx * ndy ! int_xyt
adios_totalsize = group_size
call MPI_BARRIER(MPI_COMM_WORLD,ierr)
cache_start_time = MPI_WTIME()
call adios_open (handle, group, inputfile, mode, group_comm, err)
call adios_group_size (handle, group_size, total_size, err)
!print '("rank=",i0," total_size=",i0," err=",i0)', rank, total_size, err
! write dimensions and nproc
adios_buf_size = 4
call adios_read (handle, "X", gndx, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "Y", gndy, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "npx", npx, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "npy", npy, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "nproc", nproc, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "size_x", ndx, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "size_y", ndy, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "offs_x", offx, adios_buf_size, err)
adios_buf_size = 4
call adios_read (handle, "offs_y", offy, adios_buf_size, err)
adios_buf_size = 4 * ndx * ndy
call adios_read (handle, "int_xy", int_xy, adios_buf_size, err)
adios_buf_size = 4 * ndx * ndy
call adios_read (handle, "int_xyt", int_xy, adios_buf_size, err)
call MPI_BARRIER(MPI_COMM_WORLD,ierr)
cache_end_time = MPI_WTIME()
cache_total_time = cache_end_time - cache_start_time
sz = adios_totalsize * nproc/1024.d0/1024.d0/1024.d0 !size in GB
gbs = sz/cache_total_time
!if (rank==0) write(6,*) total_time
if (rank==0) print '("Reading: ",a10,d12.2,2x,d12.2,2x,d12.3)', outputfile,sz,cache_total_time,gbs
! start streaming from buffer to disk
call adios_close (handle, err)
print '("rank=",i0,": write completed")', rank
end subroutine readArray
!!***************************
subroutine writeArray()
use copyarray2D_comm
use adios_write_mod
implicit none
integer :: tstep
character(2) :: mode = "w"
include 'mpif.h'
! Write out data using ADIOS
group = "genarray"
! calculate how much we write from this processor
group_size = 4 * 9 + & ! X,Y, nproc, all size_ and offs_ integers
4 * ndx * ndy + & ! int_xy
4 * ndx * ndy ! int_xyt
adios_totalsize = group_size
do tstep=1,timesteps
if (tstep > 1) mode = "a"
!print '("rank=",i0," group=",A," file=",A," group_size=",i0)', rank, trim(group), &
! trim(outputfile), group_size
call MPI_BARRIER(MPI_COMM_WORLD,ierr)
cache_start_time = MPI_WTIME()
call adios_open (handle, group, outputfile, mode, group_comm, err)
call adios_group_size (handle, group_size, total_size, err)
!print '("rank=",i0," total_size=",i0," err=",i0)', rank, total_size, err
! write dimensions and nproc
call adios_write (handle, "X", gndx, err)
call adios_write (handle, "Y", gndy, err)
call adios_write (handle, "npx", npx, err)
call adios_write (handle, "npy", npy, err)
call adios_write (handle, "nproc", nproc, err)
call adios_write (handle, "size_x", ndx, err)
call adios_write (handle, "size_y", ndy, err)
call adios_write (handle, "offs_x", offx, err)
call adios_write (handle, "offs_y", offy, err)
if (tstep == 1) then
call adios_write (handle, "int_xy", int_xy, err)
endif
call adios_write (handle, "int_xyt", int_xy, err)
call MPI_BARRIER(MPI_COMM_WORLD,ierr)
cache_end_time = MPI_WTIME()
cache_total_time = cache_end_time - cache_start_time
sz = adios_totalsize * nproc/1024.d0/1024.d0/1024.d0 !size in GB
gbs = sz/cache_total_time
!if (rank==0) write(6,*) total_time
if (rank==0) print '("Writing: ",a10,d12.2,2x,d12.2,2x,d12.3)', outputfile,sz,cache_total_time,gbs
! start streaming from buffer to disk
call adios_close (handle, err)
! print '("rank=",i0,": write completed")', rank
enddo
end subroutine writeArray
!!***************************
subroutine usage()
print *, "Usage: genarray input output N M [nx ny | infile] timesteps"
print *, "input: name of input file"
print *, "output: name of output file"
print *, "N: number of processes in X dimension"
print *, "M: number of processes in Y dimension"
print *, "nx: local array size in X dimension per processor"
print *, "ny: local array size in Y dimension per processor"
print *, "timesteps: number of time steps to write"
end subroutine usage
!!***************************
subroutine processArgs()
use copyarray2D_comm
#ifndef __GFORTRAN__
!#ifndef __GNUC__
interface
integer function iargc()
end function iargc
end interface
!#endif
#endif
character(len=256) :: npx_str, npy_str, ndx_str, ndy_str, time_str
integer :: numargs
!! process arguments
numargs = iargc()
!print *,"Number of arguments:",numargs
if ( numargs < 7 ) then
call usage()
call exit(1)
endif
call getarg(1, inputfile)
call getarg(2, outputfile)
call getarg(3, npx_str)
call getarg(4, npy_str)
call getarg(5, ndx_str)
call getarg(6, ndy_str)
call getarg(7, time_str)
read (npx_str,'(i5)') npx
read (npy_str,'(i5)') npy
read (ndx_str,'(i6)') ndx
read (ndy_str,'(i6)') ndy
read (time_str,'(i6)') timesteps
common_size = .true.
end subroutine processArgs
|
