1 |
adcroft |
1.5 |
C $Header: /u/gcmpack/models/MITgcmUV/eesupp/src/global_sum.F,v 1.4 1998/10/28 03:11:35 cnh Exp $ |
2 |
cnh |
1.1 |
|
3 |
|
|
C-- File global_sum.F: Routines that perform global sum on an array |
4 |
|
|
C of thread values. |
5 |
|
|
C Contents |
6 |
|
|
C o global_sum_r4 |
7 |
|
|
C o global_sum_r8 |
8 |
|
|
#include "CPP_EEOPTIONS.h" |
9 |
|
|
|
10 |
|
|
CStartOfInterface |
11 |
|
|
SUBROUTINE GLOBAL_SUM_R4( |
12 |
adcroft |
1.6 |
U sumPhi, |
13 |
cnh |
1.1 |
I myThid ) |
14 |
|
|
C /==========================================================\ |
15 |
|
|
C | SUBROUTINE GLOBAL_SUM_R4 | |
16 |
|
|
C | o Handle sum for real*4 data. | |
17 |
|
|
C |==========================================================| |
18 |
|
|
C | Perform sum an array of one value per thread and then | |
19 |
|
|
C | sum result of all the processes. | |
20 |
|
|
C | Notes | |
21 |
|
|
C | ===== | |
22 |
|
|
C | Within a process only one thread does the sum, each | |
23 |
|
|
C | thread is assumed to have already summed its local data. | |
24 |
|
|
C | The same thread also does the inter-process sum for | |
25 |
|
|
C | example with MPI and then writes the result into a shared| |
26 |
|
|
C | location. All threads wait until the sum is avaiailable. | |
27 |
|
|
C \==========================================================/ |
28 |
adcroft |
1.6 |
IMPLICIT NONE |
29 |
cnh |
1.1 |
|
30 |
|
|
C == Global data == |
31 |
|
|
#include "SIZE.h" |
32 |
|
|
#include "EEPARAMS.h" |
33 |
|
|
#include "EESUPPORT.h" |
34 |
|
|
|
35 |
|
|
C == Routine arguments == |
36 |
|
|
C sumPhi - Result of sum. |
37 |
|
|
C myThid - My thread id. |
38 |
|
|
Real*4 sumPhi |
39 |
|
|
INTEGER myThid |
40 |
|
|
CEndOfInterface |
41 |
|
|
|
42 |
|
|
C == Local variables == |
43 |
adcroft |
1.6 |
C phi - Array to be summed. |
44 |
cnh |
1.1 |
C I - Loop counters |
45 |
|
|
C mpiRC - MPI return code |
46 |
adcroft |
1.6 |
Real*4 phi(lShare4,MAX_NO_THREADS) |
47 |
cnh |
1.1 |
INTEGER I |
48 |
|
|
Real*4 tmp |
49 |
|
|
#ifdef ALLOW_USE_MPI |
50 |
|
|
INTEGER mpiRC |
51 |
|
|
#endif /* ALLOW_USE_MPI */ |
52 |
|
|
|
53 |
adcroft |
1.6 |
C-- write local sum into array |
54 |
|
|
phi(1,myThid) = sumPhi |
55 |
|
|
|
56 |
cnh |
1.4 |
C-- Can not start until everyone is ready |
57 |
cnh |
1.1 |
_BARRIER |
58 |
|
|
|
59 |
|
|
C-- Sum within the process first |
60 |
|
|
_BEGIN_MASTER( myThid ) |
61 |
|
|
tmp = 0. |
62 |
|
|
DO I=1,nThreads |
63 |
|
|
tmp = tmp + phi(1,I) |
64 |
|
|
ENDDO |
65 |
|
|
sumPhi = tmp |
66 |
|
|
#ifdef ALLOW_USE_MPI |
67 |
|
|
#ifndef ALWAYS_USE_MPI |
68 |
|
|
IF ( usingMPI ) THEN |
69 |
|
|
#endif |
70 |
|
|
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_REAL,MPI_SUM, |
71 |
adcroft |
1.6 |
& MPI_COMM_WORLD,mpiRC) |
72 |
cnh |
1.1 |
#ifndef ALWAYS_USE_MPI |
73 |
|
|
ENDIF |
74 |
|
|
#endif |
75 |
|
|
#endif /* ALLOW_USE_MPI */ |
76 |
adcroft |
1.6 |
C-- Write solution to place where all threads can see it |
77 |
cnh |
1.1 |
phi(1,1) = sumPhi |
78 |
adcroft |
1.6 |
|
79 |
cnh |
1.1 |
_END_MASTER( myThid ) |
80 |
|
|
C-- |
81 |
|
|
_BARRIER |
82 |
adcroft |
1.6 |
|
83 |
|
|
C-- set result for every process |
84 |
|
|
sumPhi = phi(1,1) |
85 |
|
|
|
86 |
cnh |
1.1 |
RETURN |
87 |
|
|
END |
88 |
|
|
|
89 |
|
|
|
90 |
|
|
CStartOfInterface |
91 |
|
|
SUBROUTINE GLOBAL_SUM_R8( |
92 |
adcroft |
1.6 |
U sumPhi, |
93 |
cnh |
1.1 |
I myThid ) |
94 |
|
|
C /==========================================================\ |
95 |
|
|
C | SUBROUTINE GLOBAL_SUM_R8 | |
96 |
|
|
C | o Handle sum for real*8 data. | |
97 |
|
|
C |==========================================================| |
98 |
|
|
C | Perform sum an array of one value per thread and then | |
99 |
|
|
C | sum result of all the processes. | |
100 |
|
|
C | Notes | |
101 |
|
|
C | ===== | |
102 |
|
|
C | Within a process only one thread does the sum, each | |
103 |
|
|
C | thread is assumed to have already summed its local data. | |
104 |
|
|
C | The same thread also does the inter-process sum for | |
105 |
|
|
C | example with MPI and then writes the result into a shared| |
106 |
|
|
C | location. All threads wait until the sum is avaiailable. | |
107 |
|
|
C \==========================================================/ |
108 |
adcroft |
1.6 |
IMPLICIT NONE |
109 |
cnh |
1.1 |
|
110 |
|
|
C === Global data === |
111 |
|
|
#include "SIZE.h" |
112 |
|
|
#include "EEPARAMS.h" |
113 |
|
|
#include "EESUPPORT.h" |
114 |
|
|
|
115 |
|
|
C === Routine arguments === |
116 |
|
|
C sumPhi - Result of sum. |
117 |
|
|
C myThid - My thread id. |
118 |
|
|
Real*8 sumPhi |
119 |
|
|
INTEGER myThid |
120 |
|
|
CEndOfInterface |
121 |
|
|
|
122 |
|
|
C === Local variables === |
123 |
adcroft |
1.6 |
C phi - Array to be summed. |
124 |
cnh |
1.1 |
C I - Loop counters |
125 |
|
|
C mpiRC - MPI return code |
126 |
adcroft |
1.6 |
Real*8 phi(lShare8,MAX_NO_THREADS) |
127 |
cnh |
1.1 |
INTEGER I |
128 |
|
|
Real*8 tmp |
129 |
|
|
#ifdef ALLOW_USE_MPI |
130 |
|
|
INTEGER mpiRC |
131 |
|
|
#endif /* ALLOW_USE_MPI */ |
132 |
|
|
|
133 |
adcroft |
1.6 |
C-- write local sum into array |
134 |
|
|
phi(1,myThid) = sumPhi |
135 |
|
|
|
136 |
cnh |
1.4 |
C-- Can not start until everyone is ready |
137 |
cnh |
1.1 |
_BARRIER |
138 |
|
|
|
139 |
|
|
C-- Sum within the process first |
140 |
|
|
_BEGIN_MASTER( myThid ) |
141 |
|
|
tmp = 0. _d 0 |
142 |
|
|
DO I=1,nThreads |
143 |
|
|
tmp = tmp + phi(1,I) |
144 |
|
|
ENDDO |
145 |
|
|
sumPhi = tmp |
146 |
|
|
#ifdef ALLOW_USE_MPI |
147 |
|
|
#ifndef ALWAYS_USE_MPI |
148 |
|
|
IF ( usingMPI ) THEN |
149 |
|
|
#endif |
150 |
|
|
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_DOUBLE_PRECISION,MPI_SUM, |
151 |
adcroft |
1.6 |
& MPI_COMM_WORLD,mpiRC) |
152 |
cnh |
1.1 |
#ifndef ALWAYS_USE_MPI |
153 |
|
|
ENDIF |
154 |
|
|
#endif |
155 |
|
|
#endif /* ALLOW_USE_MPI */ |
156 |
|
|
C-- Write solution to place where all threads can see it |
157 |
|
|
phi(1,1) = sumPhi |
158 |
|
|
_END_MASTER( myThid ) |
159 |
|
|
|
160 |
cnh |
1.4 |
C-- Do not leave until we are sure that the sum is done |
161 |
cnh |
1.1 |
_BARRIER |
162 |
adcroft |
1.6 |
|
163 |
|
|
C-- set result for every process |
164 |
|
|
sumPhi = phi(1,1) |
165 |
|
|
|
166 |
cnh |
1.1 |
RETURN |
167 |
|
|
END |