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
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
|
#define MDIM 3
! 3-d problem */
#if N_DIM == 3
#define NDIM 3
#define NGID 15
#define ik2d 1
#define ik3d 1
! 2-d problem */
#elif N_DIM == 2
#define NDIM 2
#define NGID 9
#define ik2d 1
#define ik3d 0
! 1-d problem */
#else
#define NDIM 1
#define NGID 5
#define ik2d 0
#define ik3d 0
#endif
subroutine write_header_info_sp(nvar_out, ncid, file_creation_time, &
flash_version, total_blocks, time, &
nsteps, nzones_block, unk_labels, &
varid)
use mpi
use pnetcdf
implicit none
integer nvar_out ! num vars to store
integer ncid ! file handle
character(len=*) file_creation_time ! time and date stamp
character(len=*) flash_version ! FLASH version num
integer total_blocks ! total # of blocks
real time ! simulation time
integer nsteps ! total # of timestep
integer nzones_block(3) ! nxb, nyb, nzb
character(len=4) unk_labels(*) ! unknown labels
integer varid(*) ! output: var ids
! local variables
integer i, k, err
integer dim_tot_blocks, dim_nxb, dim_nyb, dim_nzb
integer dim_NGID, dim_NDIM, dim_2
integer dimids(4)
character(len=5) record_label
integer(kind=MPI_OFFSET_KIND) i8NDIM, i8NGID, i8total_blocks
integer(kind=MPI_OFFSET_KIND) i8nzones_block(3), string_size
integer atotal_blocks(1), ansteps(1)
double precision atime(1)
i8NDIM = NDIM
i8NGID = NGID
i8total_blocks = total_blocks
i8nzones_block(:) = nzones_block(:)
atotal_blocks(1) = total_blocks
ansteps(1) = nsteps
! to avoid inconsistent header metadata warning from PnetCDF
atime(1) = time
call MPI_Bcast(atime, 1, MPI_DOUBLE_PRECISION, 0, MPI_COMM_WORLD, err)
err = nfmpi_def_dim(ncid, "dim_tot_blocks", i8total_blocks, dim_tot_blocks)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_tot_blocks")
err = nfmpi_def_dim(ncid, "dim_nxb", i8nzones_block(1), dim_nxb)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_nxb")
err = nfmpi_def_dim(ncid, "dim_nyb", i8nzones_block(2), dim_nyb)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_nyb")
err = nfmpi_def_dim(ncid, "dim_nzb", i8nzones_block(3), dim_nzb)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_nzb")
err = nfmpi_def_dim(ncid, "dim_NGID", i8NGID, dim_NGID)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_NGID")
err = nfmpi_def_dim(ncid, "dim_NDIM", i8NDIM, dim_NDIM)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_NDIM")
err = nfmpi_def_dim(ncid, "dim_2", 2_MPI_OFFSET_KIND, dim_2)
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_dim: dim_2")
dimids(1) = dim_tot_blocks
! define var for refinement level
err = nfmpi_def_var(ncid, "lrefine", NF_INT, 1, dimids, varid(1))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: lrefine")
! define var for nodetype
err = nfmpi_def_var(ncid, "nodetype", NF_INT, 1, dimids, varid(2))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: nodetype")
! define var for global id
dimids(1) = dim_NGID
dimids(2) = dim_tot_blocks
err = nfmpi_def_var(ncid, "gid", NF_INT, 2, dimids, varid(3))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: grid")
! define var for grid coordinates
dimids(1) = dim_NDIM
dimids(2) = dim_tot_blocks
err = nfmpi_def_var(ncid, "coordinates", NF_FLOAT, 2, dimids, varid(4))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: coordinates")
! define var for grid block size
dimids(1) = dim_NDIM
dimids(2) = dim_tot_blocks
err = nfmpi_def_var(ncid, "blocksize", NF_FLOAT, 2, dimids, varid(5))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: blocksize")
! define var for grid bounding box
dimids(1) = dim_2
dimids(2) = dim_NDIM
dimids(3) = dim_tot_blocks
err = nfmpi_def_var(ncid, "bndbox", NF_FLOAT, 3, dimids, varid(6))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: bndbox")
! define var for unknown array
dimids(1) = dim_nxb
dimids(2) = dim_nyb
dimids(3) = dim_nzb
dimids(4) = dim_tot_blocks
do i=1, nvar_out
record_label = unk_labels(i)
do k=1, 4
if (record_label(k:k) .EQ. ' ') record_label(k:k) = '_'
enddo
record_label(5:5) = char(0)
err = nfmpi_def_var(ncid, record_label, NF_FLOAT, 4, dimids, varid(i+6))
if (err .NE. NF_NOERR) call check(err, "nfmpi_def_var: record_label")
enddo
string_size = LEN_TRIM(file_creation_time)
err = nfmpi_put_att_text(ncid, NF_GLOBAL, "file_creation_time", string_size, file_creation_time)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_text: file_creation_time")
string_size = LEN_TRIM(flash_version)
err = nfmpi_put_att_text(ncid, NF_GLOBAL, "flash_version", string_size, flash_version)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_text: flash_version")
err = nfmpi_put_att_int(ncid, NF_GLOBAL, "total_blocks", NF_INT, 1_MPI_OFFSET_KIND, atotal_blocks)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_int: total_blocks")
err = nfmpi_put_att_int(ncid, NF_GLOBAL, "nsteps", NF_INT, 1_MPI_OFFSET_KIND, ansteps)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_int: nsteps")
err = nfmpi_put_att_int(ncid, NF_GLOBAL, "nxb", NF_INT, 1_MPI_OFFSET_KIND, nzones_block(1))
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_int: nxb")
err = nfmpi_put_att_int(ncid, NF_GLOBAL, "nyb", NF_INT, 1_MPI_OFFSET_KIND, nzones_block(2))
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_int: nyb")
err = nfmpi_put_att_int(ncid, NF_GLOBAL, "nzb", NF_INT, 1_MPI_OFFSET_KIND, nzones_block(3))
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_int: nzb")
err = nfmpi_put_att_double(ncid, NF_GLOBAL, "time", NF_DOUBLE, 1_MPI_OFFSET_KIND, atime)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_att_double: time")
err = nfmpi_enddef(ncid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_enddef")
end subroutine write_header_info_sp
!----------------------------------------------------------------------------
! subroutine plotfile
!----------------------------------------------------------------------------
double precision function plotfile_ncmpi_par(filenum, simtime, corners)
!
! plotfile using parallel i/o using PnetCDF
!
! MZ -- 4-29-00
! Jianwei -- 11/15/02
!
! This version of the plotfile routine is based on the PnetCDF
! checkpoint. The IO is done in parallel -- no copying of the data to
! a single processor to do the writing is performed.
!
! This is the SINGLE PRECISION version of the plotfile -- temporary
! storage is used to recast a variable (for every zone/block) into
! single precision before passing it onto the SP version of the C netCDF
! write routines.
!
! The data for all blocks is recast and written together. This makes the
! amount of data that is written very large, which should perform better
! on the parallel filesystems. The overhead for storing an entire
! variable (with corners) is small, <~ 1%.
!
! PnetCDF uses MPI-IO (via ROMIO) to support parallel IO. Each
! processor must open the file, define the dataspaces for each netCDF variable.
!
! A single record for each of the PARAMESH data structures is created. A
! processor only writes to a subset of this record. Each record has a
! dimension with length = tot_blocks. The offset of a processor into this
! dimension is computed by looking at the total number of blocks that are
! below the current processor.
!
! In this version of the plotfile, each variable is given its own
! record -- this makes it easier to change the variable list in the
! future without disturbing the format of the file.
!
! The include file -- ncmpi_flash.h is used for the C routines and mirrors
! the necessary data from physicaldata.fh
!
! written for netCDF 3.0
!
!---------------------------------------------------------------------------
use mpi
use pnetcdf
#include "common.fh"
integer filenum
real simtime
integer block_no
integer i, j, k, ivar, i_store, j_store, k_store
integer ngid
integer err
integer n_to_left(0:NumPEs) ! must extend from 0 to NumPEs-1
! 2-20-00 -- we don't need to allocate more space than necessary
! integer gid(mfaces+1+mchild,maxblocks_tr)
integer gid(nfaces+1+nchild,maxblocks_tr)
integer tot_blocks
save gid
integer nzones_block(3)
! create a character variable to hold the string representation of the block
! number. Note this is set to be 4 characters long (i.e. max = 9999).
character(len=4) :: fnum_string
character(len=512) :: filename
character(len=8) :: str
! set the number of variables we are going to write out
integer, parameter :: num_out = 4
! create a temporary array to hold the 4 character variable names
! this will include those defined in definitions.fh and network_common.fh
character(len=4) :: unklabels(nvar), sunklabels(num_out)
! storage for the current date and time
character(len=40) :: date_string
character(len=4) :: ionam(ionmax), record_label
integer varid(6+num_out)
integer global_offset
! hold pointers to the location in unk of the variables we are writing out
integer iout(num_out)
! allocate storage to hold a single variable information
! this should only be a small memory overhead
integer, parameter :: single = SELECTED_REAL_KIND(p=6)
real (kind=single) :: unkt_crn(1,nxb+1,nyb+k2d,nzb+k3d,maxblocks)
real (kind=single) :: unkt(1,nxb,nyb,nzb,maxblocks)
! allocate storage to hold the coordinate information and bounding box
! information
real (kind=single) :: coord_single(mdim,maxblocks_tr)
real (kind=single) :: blk_sz_single(mdim,maxblocks_tr)
real (kind=single) :: bnd_single(2,mdim,maxblocks_tr)
real (kind=single) :: sp_var1, sp_var2
integer, parameter :: release_len = 40
character (len=release_len) :: flash_release
logical corners
integer ncid, cmode, file_info, reqs(num_out+6), stats(num_out+6)
integer(kind=MPI_OFFSET_KIND) starts(4), counts(4), put_size, buf_size
if (corners) then
corner_t(1) = MPI_Wtime()
else
nocorner_t(1) = MPI_Wtime()
endif
!-----------------------------------------------------------------------------
! set the variables we are going to store
!-----------------------------------------------------------------------------
iout(1) = idens
iout(2) = itemp
iout(3) = ipres
! store the first abundance
iout(4) = inuc_begin
!-----------------------------------------------------------------------------
! compute the total number of blocks left of a given processor number
!-----------------------------------------------------------------------------
! use an allgather routine here to get the number of blocks on each proc.
call MPI_Allgather(lnblocks, 1,MPI_INTEGER, &
n_to_left,1,MPI_INTEGER, &
MPI_COMM_WORLD,err)
! compute the total number of blocks
tot_blocks = 0
do i = 0,NumPEs-1
tot_blocks = tot_blocks + n_to_left(i)
end do
! compute the number of procssors to the left of a processor
do i = NumPEs-1,1,-1
n_to_left(i) = n_to_left(i-1)
end do
n_to_left(0) = 0
do i = 2,NumPEs-1
n_to_left(i) = n_to_left(i) + n_to_left(i-1)
end do
!-----------------------------------------------------------------------------
! compute the global id -- this is a single array which stores the
! neighbor block numbers, the parent, and the children of a given block
!-----------------------------------------------------------------------------
do block_no = 1,lnblocks
ngid = 0
! loop over the faces and store the neighbors
do j = 1,nfaces
ngid = ngid + 1
! if the neighbor exists, then store the block number of the neighbor
! -- take into account the number of blocks below the processor that the
! neighbor is on, so the block number is global
if (neigh(1,j,block_no).gt.0) then
gid(ngid,block_no) = neigh(1,j,block_no) + &
n_to_left(neigh(2,j,block_no))
else
! the neighbor is either a physical boundary or does not exist at that
! level of refinement
gid(ngid,block_no) = neigh(1,j,block_no)
end if
end do
! store the parent of the current block
ngid = ngid + 1
if (parent(1,block_no).gt.0) then
gid(ngid,block_no) = parent(1,block_no) + &
n_to_left(parent(2,block_no))
else
gid(ngid,block_no) = parent(1,block_no)
end if
! store the children of the current block
do j = 1,nchild
ngid = ngid + 1
if (child(1,j,block_no).gt.0) then
gid(ngid,block_no) = child(1,j,block_no) + &
n_to_left(child(2,j,block_no))
else
gid(ngid,block_no) = child(1,j,block_no)
end if
end do
end do
!-----------------------------------------------------------------------------
! open the netCDF file
!-----------------------------------------------------------------------------
write (fnum_string, '(i4.4)') filenum
if (corners) then
filename = trim(basenm)//'ncmpi_plt_crn_'//fnum_string//'.nc'
else
filename = trim(basenm)//'ncmpi_plt_cnt_'//fnum_string//'.nc'
endif
! set up MPI I/O hints for performance enhancement
call MPI_Info_create(file_info, err)
! whether doing independent or collective writes
if (indep_io) then
call MPI_Info_set(file_info, 'romio_no_indep_rw', 'false', err)
else
call MPI_Info_set(file_info, 'romio_no_indep_rw', 'true', err)
endif
! disable data sieving, as FLASH-IO does large continuous writes
call MPI_Info_set(file_info, "romio_ds_write", "disable", err)
! disable file offset alignment for fixed-size variables
call MPI_Info_set(file_info, "nc_var_align_size", "1", err)
cmode = IOR(NF_CLOBBER, NF_64BIT_DATA)
err = nfmpi_create(MPI_COMM_WORLD, trim(filename), cmode, &
file_info, ncid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_create")
call MPI_Info_free(file_info, err)
!-----------------------------------------------------------------------------
! store the scalar information -- # of blocks, simulation time, etc
!-----------------------------------------------------------------------------
date_string = 'now'
! store the number of zones / block in each direction
if (corners) then
nzones_block(1) = nxb+1
nzones_block(2) = nyb+k2d
nzones_block(3) = nzb+k3d
else
nzones_block(1) = nxb
nzones_block(2) = nyb
nzones_block(3) = nzb
endif
! get the names of the fluids being followed
call get_mfluid_property ("short name", ionam)
! merge the two variable lists into one for storage
unklabels(1:nvar-ionmax) = varnam(:)
unklabels(nvar-ionmax+1:nvar) = ionam(:)
! get the subset of the variable labels, corresponding to what we are storing
do i = 1, num_out
sunklabels(i) = unklabels(iout(i))
enddo
sp_var1 = real(simtime, kind = single)
sp_var2 = real(dt, kind = single)
call write_header_info_sp(num_out, &
ncid, &
date_string, &
flash_release(), &
tot_blocks, &
sp_var1, &
nstep, &
nzones_block, &
sunklabels, &
varid)
if (indep_io) then
err = nfmpi_begin_indep_data(ncid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_begin_indep_data")
endif
global_offset = n_to_left(MyPE)
!-----------------------------------------------------------------------------
! store the tree information
!-----------------------------------------------------------------------------
! store the refinement level
starts(1) = global_offset+1
counts(1) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_iput_vara_int(ncid, varid(1), starts, counts, lrefine, reqs(1))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_int: lrefine sp")
else
if (indep_io) then
err = nfmpi_put_vara_int(ncid, varid(1), starts, counts, lrefine)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int: lrefine sp")
else
err = nfmpi_put_vara_int_all(ncid, varid(1), starts, counts, lrefine)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int_all: lrefine sp")
endif
endif
! store the nodetype
if (use_nonblocking_io) then
err = nfmpi_iput_vara_int(ncid, varid(2), starts, counts, nodetype, reqs(2))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_int: nodetype sp")
else
if (indep_io) then
err = nfmpi_put_vara_int(ncid, varid(2), starts, counts, nodetype)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int: nodetype sp")
else
err = nfmpi_put_vara_int_all(ncid, varid(2), starts, counts, nodetype)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int_all: nodetype sp")
endif
endif
! store the global id
starts(1) = 1
starts(2) = global_offset+1
counts(1) = NGID
counts(2) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_iput_vara_int(ncid, varid(3), starts, counts, gid, reqs(3))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_int: gid sp")
else
if (indep_io) then
err = nfmpi_put_vara_int(ncid, varid(3), starts, counts, gid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int: gid sp")
else
err = nfmpi_put_vara_int_all(ncid, varid(3), starts, counts, gid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_int_all: gid sp")
endif
endif
!-----------------------------------------------------------------------------
! store the grid information
!-----------------------------------------------------------------------------
! store the coordinates
do block_no = 1, lnblocks
coord_single(:,block_no) = real(coord(:,block_no), kind = single)
enddo
starts(1) = 1
starts(2) = global_offset+1
counts(1) = NDIM
counts(2) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_iput_vara_real(ncid, varid(4), starts, counts, coord_single, reqs(4))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_real: coord sp")
else
if (indep_io) then
err = nfmpi_put_vara_real(ncid, varid(4), starts, counts, coord_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_read: coord sp")
else
err = nfmpi_put_vara_real_all(ncid, varid(4), starts, counts, coord_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_read_all: coord sp")
endif
endif
! store the block size
do block_no = 1, lnblocks
blk_sz_single(:,block_no) = real(size(:,block_no), kind = single)
enddo
starts(1) = 1
starts(2) = global_offset+1
counts(1) = NDIM
counts(2) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_iput_vara_real(ncid, varid(5), starts, counts, blk_sz_single, reqs(5))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_real: size sp")
else
if (indep_io) then
err = nfmpi_put_vara_real(ncid, varid(5), starts, counts, blk_sz_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real: size sp")
else
err = nfmpi_put_vara_real_all(ncid, varid(5), starts, counts, blk_sz_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real_all: size sp")
endif
endif
! store the bounding box
do block_no = 1, lnblocks
bnd_single(:,:,block_no) = &
real(bnd_box(:,:,block_no), kind = single)
enddo
starts(1) = 1
starts(2) = 1
starts(3) = global_offset+1
counts(1) = 2
counts(2) = NDIM
counts(3) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_iput_vara_real(ncid, varid(6), starts, counts, bnd_single, reqs(6))
if (err .NE. NF_NOERR) call check(err, "nfmpi_iput_vara_real: bnd_box")
else
if (indep_io) then
err = nfmpi_put_vara_real(ncid, varid(6), starts, counts, bnd_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real: bnd_box")
else
err = nfmpi_put_vara_real_all(ncid, varid(6), starts, counts, bnd_single)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real_all: bnd_box")
endif
endif
if (use_nonblocking_io) then
! calculate attach buffer size for using buffered PnetCDF APIs
buf_size = (nxb+1) * (nyb+k2d) * (nzb+k3d) * maxblocks
buf_size = buf_size + nxb * nyb * nzb * maxblocks
buf_size = buf_size * num_out * 4
err = nfmpi_buffer_attach(ncid, buf_size)
endif
!-----------------------------------------------------------------------------
! store the unknowns -- here we will pass the entire unk array on each
! processor. The HDF 5 memory space functionality will pick just the
! interior cells to write to disk.
!-----------------------------------------------------------------------------
if (corners) then
corner_t(2) = MPI_Wtime()
corner_t(1) = corner_t(2) - corner_t(1)
else
nocorner_t(2) = MPI_Wtime()
nocorner_t(1) = nocorner_t(2) - nocorner_t(1)
endif
do ivar = 1, num_out
record_label = unklabels(iout(ivar))
! put the data at the corners if necessary
if (corners) then
! interpolate only the variable we are storing to the corners
! ** Important, the limits of the unkt_crn array do not include
! guard cells, so we need to map the interior of the unk array
! into the unkt_crn array.
do block_no = 1, lnblocks
do k = nguard*k3d+1,nguard*k3d+nzb+k3d
k_store = k - nguard*k3d
do j = nguard*k2d+1,nguard*k2d+nyb+k2d
j_store = j - nguard*k2d
do i = nguard+1,nguard+nxb+1
i_store = i - nguard
#if N_DIM == 2
unkt_crn(1,i_store,j_store,k_store,block_no) = &
real( &
.25*(unk(iout(ivar),i-1,j, k,block_no) + &
unk(iout(ivar),i ,j, k,block_no) + &
unk(iout(ivar),i ,j-1,k,block_no) + &
unk(iout(ivar),i-1,j-1,k,block_no)), &
kind = single)
#endif
#if N_DIM == 3
unkt_crn(1,i_store,j_store,k_store,block_no) = &
real( &
.125*(unk(iout(ivar),i-1,j ,k ,block_no) + &
unk(iout(ivar),i ,j ,k ,block_no) + &
unk(iout(ivar),i ,j-1,k ,block_no) + &
unk(iout(ivar),i-1,j-1,k ,block_no) + &
unk(iout(ivar),i-1,j ,k-1,block_no) + &
unk(iout(ivar),i ,j ,k-1,block_no) + &
unk(iout(ivar),i ,j-1,k-1,block_no) + &
unk(iout(ivar),i-1,j-1,k-1,block_no)), &
kind = single)
#endif
end do
end do
end do
enddo
! we now have the data at the corners, in a 4-byte real array
starts(1) = 1
starts(2) = 1
starts(3) = 1
starts(4) = global_offset+1
counts(1) = nxb+1
counts(2) = nyb+ik2d
counts(3) = nzb+ik3d
counts(4) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_bput_vara_real(ncid, varid(6+ivar), starts, counts, unkt_crn, reqs(ivar+6))
if (err .NE. NF_NOERR) call check(err, "nfmpi_bput_vara_real: unknowns sp")
else
if (indep_io) then
err = nfmpi_put_vara_real(ncid, varid(6+ivar), starts, counts, unkt_crn)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real: unknowns sp")
else
err = nfmpi_put_vara_real_all(ncid, varid(6+ivar), starts, counts, unkt_crn)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real_all: unknowns sp")
endif
endif
else
unkt(1,:,:,:,:) = real(unk(iout(ivar), &
nguard+1:nguard+nxb, &
nguard*k2d+1:nguard*k2d+nyb, &
nguard*k3d+1:nguard*k3d+nzb,:), &
kind = single)
starts(1) = 1
starts(2) = 1
starts(3) = 1
starts(4) = global_offset+1
counts(1) = nxb
counts(2) = nyb
counts(3) = nzb
counts(4) = lnblocks
if (use_nonblocking_io) then
err = nfmpi_bput_vara_real(ncid, varid(6+ivar), starts, counts, unkt, reqs(ivar+6))
if (err .NE. NF_NOERR) call check(err, "nfmpi_bput_vara_real: unknowns sp")
else
if (indep_io) then
err = nfmpi_put_vara_real(ncid, varid(6+ivar), starts, counts, unkt)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real: unknowns sp")
else
err = nfmpi_put_vara_real_all(ncid, varid(6+ivar), starts, counts, unkt)
if (err .NE. NF_NOERR) call check(err, "nfmpi_put_vara_real_all: unknowns sp")
endif
endif
endif
enddo
! wait for all nonblocking requests to complete
if (use_nonblocking_io) then
! wait for the nonblocking I/O to complete
if (indep_io) then
err = nfmpi_wait(ncid, num_out+6, reqs, stats)
if (err .NE. NF_NOERR) call check(err, "(sp) nfmpi_wait: ")
else
err = nfmpi_wait_all(ncid, num_out+6, reqs, stats)
if (err .NE. NF_NOERR) call check(err, "(sp) nfmpi_wait_all: ")
endif
! check the status of each nonblocking request
do i=1, num_out+6
write(str,'(I2)') i
if (stats(i) .NE. NF_NOERR) &
call check(stats(i), '(sp) nfmpi_wait(_all) req '//trim(str))
enddo
! detach the temporary buffer
err = nfmpi_buffer_detach(ncid)
if (err .NE. NF_NOERR) &
call check(err, "(sp) nfmpi_buffer_detach: ")
endif
!-----------------------------------------------------------------------------
! close the file
!-----------------------------------------------------------------------------
if (corners) then
corner_t(3) = MPI_Wtime()
corner_t(2) = corner_t(3) - corner_t(2)
else
nocorner_t(3) = MPI_Wtime()
nocorner_t(2) = nocorner_t(3) - nocorner_t(2)
endif
err = nfmpi_inq_put_size(ncid, put_size)
if (err .NE. NF_NOERR) &
call check(err, "(sp) nfmpi_inq_put_size: ")
err = nfmpi_close(ncid)
if (err .NE. NF_NOERR) call check(err, "nfmpi_close_file sp")
if (corners) then
corner_t(3) = MPI_Wtime() - corner_t(3)
else
nocorner_t(3) = MPI_Wtime() - nocorner_t(3)
endif
plotfile_ncmpi_par = put_size
return
end
|
