[1] | 1 | SUBROUTINE write_compressed( field, fid_avs, fid_fld, my_id, nxl, nxr, nyn, & |
---|
[667] | 2 | nys, nzb, nz_do3d, prec, nbgp ) |
---|
[1] | 3 | |
---|
| 4 | !------------------------------------------------------------------------------! |
---|
[484] | 5 | ! Current revisions: |
---|
[1] | 6 | ! ----------------- |
---|
| 7 | ! |
---|
| 8 | ! Former revisions: |
---|
| 9 | ! --------------------- |
---|
[3] | 10 | ! $Id: write_compressed.f90 668 2010-12-23 13:22:58Z letzel $ |
---|
[623] | 11 | ! |
---|
[668] | 12 | ! 667 2010-12-23 12:06:00Z suehring/gryschka |
---|
| 13 | ! Array bounds and nx, ny adapted with nbgp |
---|
| 14 | ! |
---|
[623] | 15 | ! 622 2010-12-10 08:08:13Z raasch |
---|
| 16 | ! optional barriers included in order to speed up collective operations |
---|
| 17 | ! |
---|
| 18 | ! Feb. 2007 |
---|
[3] | 19 | ! RCS Log replace by Id keyword, revision history cleaned up |
---|
| 20 | ! |
---|
[1] | 21 | ! Revision 1.4 2006/02/23 13:15:09 raasch |
---|
| 22 | ! nz_plot3d renamed nz_do3d |
---|
| 23 | ! |
---|
| 24 | ! Revision 1.1 1999/03/02 09:25:21 raasch |
---|
| 25 | ! Initial revision |
---|
| 26 | ! |
---|
| 27 | ! |
---|
| 28 | ! Description: |
---|
| 29 | ! ------------ |
---|
| 30 | ! In this routine, 3D-data (to be plotted) are scaled and compressed by |
---|
| 31 | ! the method of bit shifting. It is designed for the use outside of PALM |
---|
| 32 | ! also, which is the reason why most of the data is passed by subroutine |
---|
| 33 | ! arguments. Nevertheless, the module pegrid is needed by MPI calls. |
---|
| 34 | ! |
---|
| 35 | ! Arguments: |
---|
| 36 | ! field = data array to be compressed |
---|
| 37 | ! fid_avs = file-ID of AVS-data-file |
---|
| 38 | ! fid_fld = file-ID of AVS-header-file |
---|
| 39 | ! my_id = ID of the calling PE |
---|
| 40 | ! nxl, nxr = index bounds of the subdomain along x |
---|
| 41 | ! nyn, nys = index bounds of the subdomain along y |
---|
| 42 | ! nzb,nz_do3d = index bounds of the domain along z (can be smaller than |
---|
| 43 | ! the total domain) |
---|
| 44 | ! prec = precision of packed data (number of digits after decimal |
---|
| 45 | ! point) |
---|
| 46 | !------------------------------------------------------------------------------! |
---|
| 47 | |
---|
| 48 | USE pegrid ! needed for MPI_ALLREDUCE |
---|
| 49 | |
---|
| 50 | IMPLICIT NONE |
---|
| 51 | |
---|
| 52 | INTEGER, PARAMETER :: ip4 = SELECTED_INT_KIND ( 9 ) |
---|
| 53 | INTEGER, PARAMETER :: spk = SELECTED_REAL_KIND( 6 ) |
---|
| 54 | |
---|
| 55 | INTEGER :: ampl, dummy1, dummy2, factor, i, ifieldmax, ifieldmax_l, & |
---|
| 56 | ifieldmin, ifieldmin_l, ii, j, k, length, nfree, npack, nsize, & |
---|
| 57 | nx, ny, nz, pos, startpos |
---|
| 58 | INTEGER(ip4) :: imask (32) |
---|
| 59 | INTEGER(ip4), DIMENSION(:), ALLOCATABLE :: ifield, packed_ifield |
---|
| 60 | |
---|
| 61 | INTEGER, INTENT(IN) :: fid_avs, fid_fld, my_id, nxl, nxr, nyn, nys, nzb, & |
---|
[667] | 62 | nz_do3d, prec, nbgp |
---|
[1] | 63 | |
---|
[667] | 64 | REAL(spk), INTENT(IN) :: field(1:((nxr-nxl+1+2*nbgp)*(nyn-nys+1+2*nbgp)*(nz_do3d-nzb+1))) |
---|
[1] | 65 | |
---|
| 66 | ! |
---|
| 67 | !-- Initialise local variables |
---|
| 68 | ampl = 0 |
---|
| 69 | ifieldmax = 0 |
---|
| 70 | ifieldmin = 0 |
---|
| 71 | npack = 0 |
---|
| 72 | nsize = 0 |
---|
| 73 | DO i = 1,32 |
---|
| 74 | imask(i) = (2**i) - 1 |
---|
| 75 | ENDDO |
---|
| 76 | |
---|
[667] | 77 | nx = nxr - nxl + 2*nbgp |
---|
| 78 | ny = nyn - nys + 2*nbgp |
---|
[1] | 79 | nz = nz_do3d - nzb |
---|
| 80 | length = (nx+1) * (ny+1) * (nz+1) |
---|
| 81 | |
---|
| 82 | ! |
---|
| 83 | !-- Allocate memory for integer array |
---|
| 84 | ALLOCATE ( ifield(1:length) ) |
---|
| 85 | |
---|
| 86 | ! |
---|
| 87 | !-- Store data on integer (in desired precision) |
---|
| 88 | factor = 10**prec |
---|
| 89 | DO i = 1, length |
---|
| 90 | ifield(i) = NINT( field(i) * factor ) |
---|
| 91 | ENDDO |
---|
| 92 | |
---|
| 93 | ! |
---|
| 94 | !-- Find minimum and maximum |
---|
| 95 | ifieldmax_l = MAXVAL( ifield ) |
---|
| 96 | ifieldmin_l = MINVAL( ifield ) |
---|
| 97 | |
---|
| 98 | #if defined( __parallel ) |
---|
[622] | 99 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1] | 100 | CALL MPI_ALLREDUCE( ifieldmax_l, ifieldmax, 1, MPI_INTEGER, MPI_MAX, & |
---|
| 101 | comm2d, ierr ) |
---|
[622] | 102 | IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr ) |
---|
[1] | 103 | CALL MPI_ALLREDUCE( ifieldmin_l, ifieldmin, 1, MPI_INTEGER, MPI_MIN, & |
---|
| 104 | comm2d, ierr ) |
---|
| 105 | #else |
---|
| 106 | ifieldmax = ifieldmax_l |
---|
| 107 | ifieldmin = ifieldmin_l |
---|
| 108 | #endif |
---|
| 109 | |
---|
| 110 | ! |
---|
| 111 | !-- Minimum scaling |
---|
| 112 | ifield = ifield - ifieldmin |
---|
| 113 | |
---|
| 114 | ! |
---|
| 115 | !-- Calculate the number of bits needed for each value |
---|
| 116 | ampl = ifieldmax - ifieldmin |
---|
| 117 | nsize = 1 |
---|
| 118 | |
---|
| 119 | DO WHILE ( imask(nsize) < ampl ) |
---|
| 120 | nsize = nsize + 1 |
---|
| 121 | ENDDO |
---|
| 122 | |
---|
| 123 | ! |
---|
| 124 | !-- Calculate size of the packed array |
---|
| 125 | npack = length * nsize |
---|
| 126 | IF ( MOD( npack, 32 ) /= 0 ) npack = npack + 32 |
---|
| 127 | npack = npack / 32 |
---|
| 128 | |
---|
| 129 | ! |
---|
| 130 | !-- Start packing the data |
---|
| 131 | ALLOCATE ( packed_ifield(1:npack) ) |
---|
| 132 | packed_ifield = 0 |
---|
| 133 | |
---|
| 134 | ! |
---|
| 135 | !-- Starting position of a word |
---|
| 136 | startpos = 0 |
---|
| 137 | |
---|
| 138 | ! |
---|
| 139 | !-- Starting position of the word to which data are actually written |
---|
| 140 | ii = 1 |
---|
| 141 | |
---|
| 142 | ! |
---|
| 143 | !-- Compress all data |
---|
| 144 | DO i = 1, length |
---|
| 145 | |
---|
| 146 | ! |
---|
| 147 | !-- Cut the significant bits from the actual grid point value (GPV) |
---|
| 148 | dummy1 = IAND( ifield(i), imask(nsize) ) |
---|
| 149 | |
---|
| 150 | ! |
---|
| 151 | !-- Calculate number of free bits of the actual word after packing the GPV |
---|
| 152 | nfree = 32 - startpos - nsize |
---|
| 153 | |
---|
| 154 | IF ( nfree > 0 ) THEN |
---|
| 155 | ! |
---|
| 156 | !-- GPV fits to the actual word (ii), additional bits are still free. |
---|
| 157 | !-- Shift GPV to the new position |
---|
| 158 | dummy2 = ISHFT( dummy1 ,nfree ) |
---|
| 159 | |
---|
| 160 | ! |
---|
| 161 | !-- Append bits to the already packed data |
---|
| 162 | packed_ifield(ii) = packed_ifield(ii) + dummy2 |
---|
| 163 | |
---|
| 164 | ! |
---|
| 165 | !-- Calculate new starting position |
---|
| 166 | startpos = startpos + nsize |
---|
| 167 | |
---|
| 168 | ELSEIF ( nfree .EQ. 0 ) THEN |
---|
| 169 | ! |
---|
| 170 | !-- GPV fills the actual word (ii) exactly |
---|
| 171 | packed_ifield(ii) = packed_ifield(ii) + dummy1 |
---|
| 172 | |
---|
| 173 | ! |
---|
| 174 | !-- Activate next (new) word |
---|
| 175 | ii = ii + 1 |
---|
| 176 | |
---|
| 177 | ! |
---|
| 178 | !-- Reset starting position of the new word |
---|
| 179 | startpos = 0 |
---|
| 180 | |
---|
| 181 | ELSE |
---|
| 182 | ! |
---|
| 183 | !-- GPV must be split up to the actual (ii) and the following (ii+1) |
---|
| 184 | !-- word. Shift first part of GPV to its position. |
---|
| 185 | dummy2 = ISHFT( dummy1, nfree ) |
---|
| 186 | |
---|
| 187 | ! |
---|
| 188 | !-- Append bits |
---|
| 189 | packed_ifield(ii) = packed_ifield(ii) + dummy2 |
---|
| 190 | |
---|
| 191 | ! |
---|
| 192 | !-- Store rest of GPV on the next word |
---|
| 193 | ii = ii + 1 |
---|
| 194 | packed_ifield(ii) = ISHFT( dummy1, 32+nfree ) |
---|
| 195 | ! |
---|
| 196 | !-- Calculate starting position of the next GPV |
---|
| 197 | startpos = -nfree |
---|
| 198 | |
---|
| 199 | ENDIF |
---|
| 200 | |
---|
| 201 | ENDDO |
---|
| 202 | |
---|
| 203 | ! |
---|
| 204 | !-- Write the compressed 3D array |
---|
| 205 | WRITE ( fid_avs ) packed_ifield |
---|
| 206 | |
---|
| 207 | ! |
---|
| 208 | !-- Write additional informations on FLD-file |
---|
| 209 | IF ( my_id == 0 ) WRITE ( fid_fld, 100 ) prec, ifieldmin, nsize, length |
---|
| 210 | |
---|
| 211 | DEALLOCATE( ifield, packed_ifield ) |
---|
| 212 | |
---|
| 213 | ! |
---|
| 214 | !-- Formats |
---|
| 215 | 100 FORMAT ('# precision = ',I4/ & |
---|
| 216 | '# feldmin = ',I8/ & |
---|
| 217 | '# nbits = ',I2/ & |
---|
| 218 | '# nskip = ',I8) |
---|
| 219 | |
---|
| 220 | END SUBROUTINE write_compressed |
---|