1 |
C $Header: /u/gcmpack/models/MITgcmUV/eesupp/src/global_sum.F,v 1.2 1998/04/23 20:37:30 cnh Exp $ |
2 |
|
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 |
I phi, |
13 |
O sumPhi, |
14 |
I myThid ) |
15 |
C /==========================================================\ |
16 |
C | SUBROUTINE GLOBAL_SUM_R4 | |
17 |
C | o Handle sum for real*4 data. | |
18 |
C |==========================================================| |
19 |
C | Perform sum an array of one value per thread and then | |
20 |
C | sum result of all the processes. | |
21 |
C | Notes | |
22 |
C | ===== | |
23 |
C | Within a process only one thread does the sum, each | |
24 |
C | thread is assumed to have already summed its local data. | |
25 |
C | The same thread also does the inter-process sum for | |
26 |
C | example with MPI and then writes the result into a shared| |
27 |
C | location. All threads wait until the sum is avaiailable. | |
28 |
C \==========================================================/ |
29 |
|
30 |
C == Global data == |
31 |
#include "SIZE.h" |
32 |
#include "EEPARAMS.h" |
33 |
#include "EESUPPORT.h" |
34 |
|
35 |
C == Routine arguments == |
36 |
C phi - Array to be summed. |
37 |
C sumPhi - Result of sum. |
38 |
C myThid - My thread id. |
39 |
Real*4 phi(lShare4,MAX_NO_THREADS) |
40 |
Real*4 sumPhi |
41 |
INTEGER myThid |
42 |
CEndOfInterface |
43 |
|
44 |
C == Local variables == |
45 |
C I - Loop counters |
46 |
C mpiRC - MPI return code |
47 |
INTEGER I |
48 |
Real*4 tmp |
49 |
#ifdef ALLOW_USE_MPI |
50 |
INTEGER mpiRC |
51 |
#endif /* ALLOW_USE_MPI */ |
52 |
|
53 |
C-- Can't start until everyone is ready |
54 |
_BARRIER |
55 |
|
56 |
C-- Sum within the process first |
57 |
_BEGIN_MASTER( myThid ) |
58 |
tmp = 0. |
59 |
DO I=1,nThreads |
60 |
tmp = tmp + phi(1,I) |
61 |
ENDDO |
62 |
sumPhi = tmp |
63 |
#ifdef ALLOW_USE_MPI |
64 |
#ifndef ALWAYS_USE_MPI |
65 |
IF ( usingMPI ) THEN |
66 |
#endif |
67 |
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_REAL,MPI_SUM, |
68 |
& MPI_COMM_WORLD,mpiRC) |
69 |
#ifndef ALWAYS_USE_MPI |
70 |
ENDIF |
71 |
#endif |
72 |
#endif /* ALLOW_USE_MPI */ |
73 |
phi(1,1) = sumPhi |
74 |
_END_MASTER( myThid ) |
75 |
C-- |
76 |
_BARRIER |
77 |
C |
78 |
RETURN |
79 |
END |
80 |
|
81 |
|
82 |
CStartOfInterface |
83 |
SUBROUTINE GLOBAL_SUM_R8( |
84 |
I phi, |
85 |
O sumPhi, |
86 |
I myThid ) |
87 |
C /==========================================================\ |
88 |
C | SUBROUTINE GLOBAL_SUM_R8 | |
89 |
C | o Handle sum for real*8 data. | |
90 |
C |==========================================================| |
91 |
C | Perform sum an array of one value per thread and then | |
92 |
C | sum result of all the processes. | |
93 |
C | Notes | |
94 |
C | ===== | |
95 |
C | Within a process only one thread does the sum, each | |
96 |
C | thread is assumed to have already summed its local data. | |
97 |
C | The same thread also does the inter-process sum for | |
98 |
C | example with MPI and then writes the result into a shared| |
99 |
C | location. All threads wait until the sum is avaiailable. | |
100 |
C \==========================================================/ |
101 |
|
102 |
C === Global data === |
103 |
#include "SIZE.h" |
104 |
#include "EEPARAMS.h" |
105 |
#include "EESUPPORT.h" |
106 |
|
107 |
C === Routine arguments === |
108 |
C phi - Array to be summed. |
109 |
C sumPhi - Result of sum. |
110 |
C myThid - My thread id. |
111 |
Real*8 phi(lShare8,MAX_NO_THREADS) |
112 |
Real*8 sumPhi |
113 |
INTEGER myThid |
114 |
CEndOfInterface |
115 |
|
116 |
C === Local variables === |
117 |
C I - Loop counters |
118 |
C mpiRC - MPI return code |
119 |
INTEGER I |
120 |
Real*8 tmp |
121 |
#ifdef ALLOW_USE_MPI |
122 |
INTEGER mpiRC |
123 |
#endif /* ALLOW_USE_MPI */ |
124 |
|
125 |
C-- Can't start until everyone is ready |
126 |
_BARRIER |
127 |
|
128 |
C-- Sum within the process first |
129 |
_BEGIN_MASTER( myThid ) |
130 |
tmp = 0. _d 0 |
131 |
DO I=1,nThreads |
132 |
tmp = tmp + phi(1,I) |
133 |
ENDDO |
134 |
sumPhi = tmp |
135 |
#ifdef ALLOW_USE_MPI |
136 |
#ifndef ALWAYS_USE_MPI |
137 |
IF ( usingMPI ) THEN |
138 |
#endif |
139 |
CALL MPI_Allreduce(tmp,sumPhi,1,MPI_DOUBLE_PRECISION,MPI_SUM, |
140 |
& MPI_COMM_WORLD,mpiRC) |
141 |
#ifndef ALWAYS_USE_MPI |
142 |
ENDIF |
143 |
#endif |
144 |
#endif /* ALLOW_USE_MPI */ |
145 |
C-- Write solution to place where all threads can see it |
146 |
phi(1,1) = sumPhi |
147 |
_END_MASTER( myThid ) |
148 |
|
149 |
C-- Don't leave until we are sure that the sum is done |
150 |
_BARRIER |
151 |
C |
152 |
RETURN |
153 |
END |