| 1 |
jmc |
1.15 |
C $Header: /u/gcmpack/MITgcm/eesupp/src/global_sum.F,v 1.14 2009/06/10 03:47:03 jmc Exp $ |
| 2 |
adcroft |
1.9 |
C $Name: $ |
| 3 |
cnh |
1.1 |
|
| 4 |
jmc |
1.14 |
#include "CPP_EEOPTIONS.h" |
| 5 |
|
|
|
| 6 |
cnh |
1.1 |
C-- File global_sum.F: Routines that perform global sum on an array |
| 7 |
|
|
C of thread values. |
| 8 |
|
|
C Contents |
| 9 |
jmc |
1.13 |
C o GLOBAL_SUM_R4 |
| 10 |
|
|
C o GLOBAL_SUM_R8 |
| 11 |
|
|
C o GLOBAL_SUM_INT |
| 12 |
|
|
|
| 13 |
jmc |
1.14 |
C---+----1----+----2----+----3----+----4----+----5----+----6----+----7-|--+----| |
| 14 |
cnh |
1.10 |
CBOP |
| 15 |
|
|
C !ROUTINE: GLOBAL_SUM_R4 |
| 16 |
|
|
|
| 17 |
|
|
C !INTERFACE: |
| 18 |
jmc |
1.13 |
SUBROUTINE GLOBAL_SUM_R4( |
| 19 |
adcroft |
1.6 |
U sumPhi, |
| 20 |
cnh |
1.1 |
I myThid ) |
| 21 |
jmc |
1.13 |
|
| 22 |
cnh |
1.10 |
C !DESCRIPTION: |
| 23 |
|
|
C *==========================================================* |
| 24 |
jmc |
1.13 |
C | SUBROUTINE GLOBAL\_SUM\_R4 |
| 25 |
|
|
C | o Handle sum for real*4 data. |
| 26 |
cnh |
1.10 |
C *==========================================================* |
| 27 |
jmc |
1.13 |
C | Perform sum an array of one value per thread and then |
| 28 |
|
|
C | sum result of all the processes. |
| 29 |
|
|
C | Notes: |
| 30 |
|
|
C | Within a process only one thread does the sum. |
| 31 |
|
|
C | The same thread also does the inter-process sum for |
| 32 |
|
|
C | example with MPI and then writes the result into a shared |
| 33 |
|
|
C | location. All threads wait until the sum is avaiailable. |
| 34 |
cnh |
1.10 |
C *==========================================================* |
| 35 |
cnh |
1.1 |
|
| 36 |
cnh |
1.10 |
C !USES: |
| 37 |
jmc |
1.13 |
IMPLICIT NONE |
| 38 |
|
|
|
| 39 |
cnh |
1.1 |
C == Global data == |
| 40 |
|
|
#include "SIZE.h" |
| 41 |
|
|
#include "EEPARAMS.h" |
| 42 |
|
|
#include "EESUPPORT.h" |
| 43 |
adcroft |
1.9 |
#include "GLOBAL_SUM.h" |
| 44 |
cnh |
1.1 |
|
| 45 |
cnh |
1.10 |
C !INPUT/OUTPUT PARAMETERS: |
| 46 |
cnh |
1.1 |
C == Routine arguments == |
| 47 |
jmc |
1.14 |
C sumPhi :: Result of sum. |
| 48 |
|
|
C myThid :: My thread id. |
| 49 |
cnh |
1.1 |
Real*4 sumPhi |
| 50 |
|
|
INTEGER myThid |
| 51 |
|
|
|
| 52 |
cnh |
1.10 |
C !LOCAL VARIABLES: |
| 53 |
cnh |
1.1 |
C == Local variables == |
| 54 |
jmc |
1.14 |
C I :: Loop counters |
| 55 |
|
|
C mpiRC :: MPI return code |
| 56 |
cnh |
1.1 |
INTEGER I |
| 57 |
|
|
Real*4 tmp |
| 58 |
|
|
#ifdef ALLOW_USE_MPI |
| 59 |
|
|
INTEGER mpiRC |
| 60 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 61 |
cnh |
1.10 |
CEOP |
| 62 |
cnh |
1.1 |
|
| 63 |
adcroft |
1.6 |
C-- write local sum into array |
| 64 |
jmc |
1.14 |
phiGSR4(1,myThid) = sumPhi |
| 65 |
adcroft |
1.6 |
|
| 66 |
cnh |
1.4 |
C-- Can not start until everyone is ready |
| 67 |
adcroft |
1.9 |
CALL BAR2( myThid ) |
| 68 |
cnh |
1.1 |
|
| 69 |
|
|
C-- Sum within the process first |
| 70 |
|
|
_BEGIN_MASTER( myThid ) |
| 71 |
|
|
tmp = 0. |
| 72 |
|
|
DO I=1,nThreads |
| 73 |
jmc |
1.14 |
tmp = tmp + phiGSR4(1,I) |
| 74 |
cnh |
1.1 |
ENDDO |
| 75 |
|
|
sumPhi = tmp |
| 76 |
|
|
#ifdef ALLOW_USE_MPI |
| 77 |
|
|
#ifndef ALWAYS_USE_MPI |
| 78 |
|
|
IF ( usingMPI ) THEN |
| 79 |
|
|
#endif |
| 80 |
|
|
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_REAL,MPI_SUM, |
| 81 |
jmc |
1.11 |
& MPI_COMM_MODEL,mpiRC) |
| 82 |
cnh |
1.1 |
#ifndef ALWAYS_USE_MPI |
| 83 |
|
|
ENDIF |
| 84 |
|
|
#endif |
| 85 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 86 |
adcroft |
1.6 |
C-- Write solution to place where all threads can see it |
| 87 |
jmc |
1.14 |
phiGSR4(1,0) = sumPhi |
| 88 |
adcroft |
1.6 |
|
| 89 |
cnh |
1.1 |
_END_MASTER( myThid ) |
| 90 |
|
|
C-- |
| 91 |
adcroft |
1.9 |
CALL BAR2( myThid ) |
| 92 |
adcroft |
1.6 |
|
| 93 |
|
|
C-- set result for every process |
| 94 |
jmc |
1.15 |
sumPhi = phiGSR4(1,0) |
| 95 |
adcroft |
1.6 |
|
| 96 |
cnh |
1.1 |
RETURN |
| 97 |
|
|
END |
| 98 |
jmc |
1.13 |
|
| 99 |
cnh |
1.1 |
|
| 100 |
jmc |
1.14 |
C---+----1----+----2----+----3----+----4----+----5----+----6----+----7-|--+----| |
| 101 |
cnh |
1.10 |
CBOP |
| 102 |
|
|
C !ROUTINE: GLOBAL_SUM_R8 |
| 103 |
|
|
|
| 104 |
|
|
C !INTERFACE: |
| 105 |
jmc |
1.13 |
SUBROUTINE GLOBAL_SUM_R8( |
| 106 |
adcroft |
1.6 |
U sumPhi, |
| 107 |
cnh |
1.1 |
I myThid ) |
| 108 |
jmc |
1.14 |
|
| 109 |
cnh |
1.10 |
C !DESCRIPTION: |
| 110 |
|
|
C *==========================================================* |
| 111 |
jmc |
1.13 |
C | SUBROUTINE GLOBAL\_SUM\_R8 |
| 112 |
|
|
C | o Handle sum for real*8 data. |
| 113 |
cnh |
1.10 |
C *==========================================================* |
| 114 |
jmc |
1.13 |
C | Perform sum an array of one value per thread and then |
| 115 |
|
|
C | sum result of all the processes. |
| 116 |
|
|
C | Notes: |
| 117 |
|
|
C | Within a process only one thread does the sum. |
| 118 |
|
|
C | The same thread also does the inter-process sum for |
| 119 |
|
|
C | example with MPI and then writes the result into a shared |
| 120 |
|
|
C | location. All threads wait until the sum is avaiailable. |
| 121 |
cnh |
1.10 |
C *==========================================================* |
| 122 |
cnh |
1.1 |
|
| 123 |
cnh |
1.10 |
C !USES: |
| 124 |
jmc |
1.14 |
IMPLICIT NONE |
| 125 |
|
|
|
| 126 |
cnh |
1.1 |
C === Global data === |
| 127 |
|
|
#include "SIZE.h" |
| 128 |
|
|
#include "EEPARAMS.h" |
| 129 |
|
|
#include "EESUPPORT.h" |
| 130 |
adcroft |
1.9 |
#include "GLOBAL_SUM.h" |
| 131 |
cnh |
1.1 |
|
| 132 |
cnh |
1.10 |
C !INPUT/OUTPUT PARAMETERS: |
| 133 |
cnh |
1.1 |
C === Routine arguments === |
| 134 |
jmc |
1.14 |
C sumPhi :: Result of sum. |
| 135 |
|
|
C myThid :: My thread id. |
| 136 |
cnh |
1.1 |
Real*8 sumPhi |
| 137 |
|
|
INTEGER myThid |
| 138 |
|
|
|
| 139 |
cnh |
1.10 |
C !LOCAL VARIABLES: |
| 140 |
cnh |
1.1 |
C === Local variables === |
| 141 |
jmc |
1.14 |
C I :: Loop counters |
| 142 |
|
|
C mpiRC :: MPI return code |
| 143 |
cnh |
1.1 |
INTEGER I |
| 144 |
|
|
Real*8 tmp |
| 145 |
|
|
#ifdef ALLOW_USE_MPI |
| 146 |
|
|
INTEGER mpiRC |
| 147 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 148 |
cnh |
1.10 |
CEOP |
| 149 |
cnh |
1.1 |
|
| 150 |
adcroft |
1.6 |
C-- write local sum into array |
| 151 |
jmc |
1.14 |
phiGSR8(1,myThid) = sumPhi |
| 152 |
adcroft |
1.6 |
|
| 153 |
cnh |
1.4 |
C-- Can not start until everyone is ready |
| 154 |
jmc |
1.14 |
C CALL FOOL_THE_COMPILER( phiGSR8 ) |
| 155 |
adcroft |
1.9 |
C CALL MS |
| 156 |
|
|
CALL BAR2( myThid ) |
| 157 |
|
|
C _BARRIER |
| 158 |
|
|
C _BARRIER |
| 159 |
jmc |
1.14 |
C CALL FOOL_THE_COMPILER( phiGSR8 ) |
| 160 |
cnh |
1.1 |
|
| 161 |
|
|
C-- Sum within the process first |
| 162 |
|
|
_BEGIN_MASTER( myThid ) |
| 163 |
|
|
tmp = 0. _d 0 |
| 164 |
|
|
DO I=1,nThreads |
| 165 |
jmc |
1.14 |
tmp = tmp + phiGSR8(1,I) |
| 166 |
cnh |
1.1 |
ENDDO |
| 167 |
|
|
sumPhi = tmp |
| 168 |
|
|
#ifdef ALLOW_USE_MPI |
| 169 |
|
|
#ifndef ALWAYS_USE_MPI |
| 170 |
|
|
IF ( usingMPI ) THEN |
| 171 |
|
|
#endif |
| 172 |
|
|
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_DOUBLE_PRECISION,MPI_SUM, |
| 173 |
jmc |
1.11 |
& MPI_COMM_MODEL,mpiRC) |
| 174 |
heimbach |
1.7 |
#ifndef ALWAYS_USE_MPI |
| 175 |
|
|
ENDIF |
| 176 |
|
|
#endif |
| 177 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 178 |
|
|
C-- Write solution to place where all threads can see it |
| 179 |
jmc |
1.14 |
phiGSR8(1,0) = sumPhi |
| 180 |
heimbach |
1.7 |
_END_MASTER( myThid ) |
| 181 |
|
|
|
| 182 |
|
|
C-- Do not leave until we are sure that the sum is done |
| 183 |
jmc |
1.14 |
C CALL FOOL_THE_COMPILER( phiGSR8 ) |
| 184 |
adcroft |
1.9 |
C CALL MS |
| 185 |
|
|
C _BARRIER |
| 186 |
|
|
C _BARRIER |
| 187 |
|
|
CALL BAR2( myThid ) |
| 188 |
jmc |
1.14 |
C CALL FOOL_THE_COMPILER( phiGSR8 ) |
| 189 |
heimbach |
1.7 |
|
| 190 |
|
|
C-- set result for every process |
| 191 |
jmc |
1.14 |
sumPhi = phiGSR8(1,0) |
| 192 |
heimbach |
1.7 |
|
| 193 |
|
|
RETURN |
| 194 |
|
|
END |
| 195 |
cnh |
1.10 |
|
| 196 |
jmc |
1.14 |
C---+----1----+----2----+----3----+----4----+----5----+----6----+----7-|--+----| |
| 197 |
cnh |
1.10 |
CBOP |
| 198 |
|
|
C !ROUTINE: GLOBAL_SUM_INT |
| 199 |
|
|
C !INTERFACE: |
| 200 |
jmc |
1.13 |
SUBROUTINE GLOBAL_SUM_INT( |
| 201 |
heimbach |
1.7 |
U sumPhi, |
| 202 |
|
|
I myThid ) |
| 203 |
jmc |
1.14 |
|
| 204 |
cnh |
1.10 |
C !DESCRIPTION: |
| 205 |
|
|
C *==========================================================* |
| 206 |
jmc |
1.13 |
C | SUBROUTINE GLOBAL\_SUM\_INT |
| 207 |
|
|
C | o Handle sum for integer data. |
| 208 |
cnh |
1.10 |
C *==========================================================* |
| 209 |
jmc |
1.13 |
C | Perform sum an array of one value per thread and then |
| 210 |
|
|
C | sum result of all the processes. |
| 211 |
|
|
C | Notes: |
| 212 |
|
|
C | Within a process only one thread does the sum. |
| 213 |
|
|
C | The same thread also does the inter-process sum for |
| 214 |
|
|
C | example with MPI and then writes the result into a shared |
| 215 |
|
|
C | location. All threads wait until the sum is avaiailable. |
| 216 |
cnh |
1.10 |
C *==========================================================* |
| 217 |
heimbach |
1.7 |
|
| 218 |
cnh |
1.10 |
C !USES: |
| 219 |
jmc |
1.14 |
IMPLICIT NONE |
| 220 |
|
|
|
| 221 |
heimbach |
1.7 |
C === Global data === |
| 222 |
|
|
#include "SIZE.h" |
| 223 |
|
|
#include "EEPARAMS.h" |
| 224 |
|
|
#include "EESUPPORT.h" |
| 225 |
adcroft |
1.9 |
#include "GLOBAL_SUM.h" |
| 226 |
heimbach |
1.7 |
|
| 227 |
cnh |
1.10 |
C !INPUT/OUTPUT PARAMETERS: |
| 228 |
heimbach |
1.7 |
C === Routine arguments === |
| 229 |
jmc |
1.14 |
C sumPhi :: Result of sum. |
| 230 |
|
|
C myThid :: My thread id. |
| 231 |
heimbach |
1.7 |
INTEGER sumPhi |
| 232 |
|
|
INTEGER myThid |
| 233 |
|
|
|
| 234 |
cnh |
1.10 |
C !LOCAL VARIABLES: |
| 235 |
heimbach |
1.7 |
C === Local variables === |
| 236 |
jmc |
1.14 |
C I :: Loop counters |
| 237 |
|
|
C mpiRC :: MPI return code |
| 238 |
heimbach |
1.7 |
INTEGER I |
| 239 |
|
|
INTEGER tmp |
| 240 |
|
|
#ifdef ALLOW_USE_MPI |
| 241 |
|
|
INTEGER mpiRC |
| 242 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 243 |
cnh |
1.10 |
CEOP |
| 244 |
heimbach |
1.7 |
|
| 245 |
|
|
C-- write local sum into array |
| 246 |
adcroft |
1.9 |
phiGSI(1,myThid) = sumPhi |
| 247 |
heimbach |
1.7 |
|
| 248 |
|
|
C-- Can not start until everyone is ready |
| 249 |
|
|
_BARRIER |
| 250 |
|
|
|
| 251 |
|
|
C-- Sum within the process first |
| 252 |
|
|
_BEGIN_MASTER( myThid ) |
| 253 |
jmc |
1.13 |
tmp = 0 |
| 254 |
heimbach |
1.7 |
DO I=1,nThreads |
| 255 |
adcroft |
1.9 |
tmp = tmp + phiGSI(1,I) |
| 256 |
heimbach |
1.7 |
ENDDO |
| 257 |
|
|
sumPhi = tmp |
| 258 |
|
|
#ifdef ALLOW_USE_MPI |
| 259 |
|
|
#ifndef ALWAYS_USE_MPI |
| 260 |
|
|
IF ( usingMPI ) THEN |
| 261 |
|
|
#endif |
| 262 |
|
|
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_INTEGER,MPI_SUM, |
| 263 |
jmc |
1.11 |
& MPI_COMM_MODEL,mpiRC) |
| 264 |
cnh |
1.1 |
#ifndef ALWAYS_USE_MPI |
| 265 |
|
|
ENDIF |
| 266 |
|
|
#endif |
| 267 |
|
|
#endif /* ALLOW_USE_MPI */ |
| 268 |
|
|
C-- Write solution to place where all threads can see it |
| 269 |
jmc |
1.14 |
phiGSI(1,0) = sumPhi |
| 270 |
cnh |
1.1 |
_END_MASTER( myThid ) |
| 271 |
|
|
|
| 272 |
cnh |
1.4 |
C-- Do not leave until we are sure that the sum is done |
| 273 |
cnh |
1.1 |
_BARRIER |
| 274 |
adcroft |
1.6 |
|
| 275 |
|
|
C-- set result for every process |
| 276 |
jmc |
1.14 |
sumPhi = phiGSI(1,0) |
| 277 |
adcroft |
1.6 |
|
| 278 |
cnh |
1.1 |
RETURN |
| 279 |
|
|
END |
Parent Directory
| 
Revision Log
| 
Revision Graph