Skip to content

Commit a895014

Browse files
prathi-windprathi-wind
committed
Undo changes mpi_common as will be refactored in future pull request
1 parent 4f8b48b commit a895014

File tree

1 file changed

+34
-35
lines changed

1 file changed

+34
-35
lines changed

‎src/common/m_mpi_common.fpp‎

Lines changed: 34 additions & 35 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,5 @@
11

22
#:include 'macros.fpp'
3-
#:include 'parallel_macros.fpp'
43

54
!> @brief The module serves as a proxy to the parameters and subroutines
65
!! available in the MPI implementation's MPI module. Specifically,
@@ -26,7 +25,7 @@ module m_mpi_common
2625
implicit none
2726
2827
integer, private :: err_code, ierr, v_size !<
29-
$:GPU_DECLARE(create=["v_size"])
28+
!$acc declare create(v_size)
3029
!! Generic flags used to identify and report MPI errors
3130
3231
real(wp), private, allocatable, dimension(:), target :: buff_send !<
@@ -39,10 +38,10 @@ module m_mpi_common
3938
!! average primitive variables, for a single computational domain boundary
4039
!! at the time, from the relevant neighboring processor.
4140
42-
$:GPU_DECLARE(create=["buff_send","buff_recv"])
41+
!$acc declare create(buff_send, buff_recv)
4342
4443
integer :: halo_size, nVars
45-
$:GPU_DECLARE(create=["halo_size","nVars"])
44+
!$acc declare create(halo_size, nVars)
4645
4746
contains
4847
@@ -639,7 +638,7 @@ contains
639638
#ifdef MFC_MPI
640639
641640
call nvtxStartRange("RHS-COMM-PACKBUF")
642-
$:GPU_UPDATE(device=["v_size"])
641+
!$acc update device(v_size)
643642
644643
#ifdef MFC_SIMULATION
645644
if (qbmm .and. .not. polytropic) then
@@ -693,7 +692,7 @@ contains
693692
#:for mpi_dir in [1, 2, 3]
694693
if (mpi_dir == ${mpi_dir}$) then
695694
#:if mpi_dir == 1
696-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
695+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
697696
do l = 0, p
698697
do k = 0, n
699698
do j = 0, buff_size - 1
@@ -707,7 +706,7 @@ contains
707706
708707
#ifdef MFC_SIMULATION
709708
if (qbmm .and. .not. polytropic) then
710-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
709+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
711710
do l = 0, p
712711
do k = 0, n
713712
do j = 0, buff_size - 1
@@ -722,7 +721,7 @@ contains
722721
end do
723722
end do
724723
725-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
724+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
726725
do l = 0, p
727726
do k = 0, n
728727
do j = 0, buff_size - 1
@@ -739,7 +738,7 @@ contains
739738
end if
740739
#endif
741740
#:elif mpi_dir == 2
742-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
741+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
743742
do i = 1, sys_size
744743
do l = 0, p
745744
do k = 0, buff_size - 1
@@ -755,7 +754,7 @@ contains
755754
756755
#ifdef MFC_SIMULATION
757756
if (qbmm .and. .not. polytropic) then
758-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
757+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
759758
do i = sys_size + 1, sys_size + 4
760759
do l = 0, p
761760
do k = 0, buff_size - 1
@@ -771,7 +770,7 @@ contains
771770
end do
772771
end do
773772
774-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
773+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
775774
do i = sys_size + 1, sys_size + 4
776775
do l = 0, p
777776
do k = 0, buff_size - 1
@@ -789,7 +788,7 @@ contains
789788
end if
790789
#endif
791790
#:else
792-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
791+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
793792
do i = 1, sys_size
794793
do l = 0, buff_size - 1
795794
do k = -buff_size, n + buff_size
@@ -805,7 +804,7 @@ contains
805804
806805
#ifdef MFC_SIMULATION
807806
if (qbmm .and. .not. polytropic) then
808-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
807+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
809808
do i = sys_size + 1, sys_size + 4
810809
do l = 0, buff_size - 1
811810
do k = -buff_size, n + buff_size
@@ -821,7 +820,7 @@ contains
821820
end do
822821
end do
823822
824-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
823+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
825824
do i = sys_size + 1, sys_size + 4
826825
do l = 0, buff_size - 1
827826
do k = -buff_size, n + buff_size
@@ -856,7 +855,7 @@ contains
856855
call nvtxStartRange("RHS-COMM-SENDRECV-RDMA")
857856
#:else
858857
call nvtxStartRange("RHS-COMM-DEV2HOST")
859-
$:GPU_UPDATE(host=["buff_send"])
858+
!$acc update host(buff_send)
860859
call nvtxEndRange
861860
call nvtxStartRange("RHS-COMM-SENDRECV-NO-RMDA")
862861
#:endif
@@ -874,7 +873,7 @@ contains
874873
!$acc wait
875874
#:else
876875
call nvtxStartRange("RHS-COMM-HOST2DEV")
877-
$:GPU_UPDATE(device=["buff_recv"])
876+
!$acc update device(buff_recv)
878877
call nvtxEndRange
879878
#:endif
880879
end if
@@ -891,7 +890,7 @@ contains
891890
#:for mpi_dir in [1, 2, 3]
892891
if (mpi_dir == ${mpi_dir}$) then
893892
#:if mpi_dir == 1
894-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
893+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
895894
do l = 0, p
896895
do k = 0, n
897896
do j = -buff_size, -1
@@ -912,7 +911,7 @@ contains
912911
913912
#ifdef MFC_SIMULATION
914913
if (qbmm .and. .not. polytropic) then
915-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
914+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
916915
do l = 0, p
917916
do k = 0, n
918917
do j = -buff_size, -1
@@ -927,7 +926,7 @@ contains
927926
end do
928927
end do
929928
930-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
929+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
931930
do l = 0, p
932931
do k = 0, n
933932
do j = -buff_size, -1
@@ -944,7 +943,7 @@ contains
944943
end if
945944
#endif
946945
#:elif mpi_dir == 2
947-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
946+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
948947
do i = 1, sys_size
949948
do l = 0, p
950949
do k = -buff_size, -1
@@ -966,7 +965,7 @@ contains
966965
967966
#ifdef MFC_SIMULATION
968967
if (qbmm .and. .not. polytropic) then
969-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
968+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
970969
do i = sys_size + 1, sys_size + 4
971970
do l = 0, p
972971
do k = -buff_size, -1
@@ -982,7 +981,7 @@ contains
982981
end do
983982
end do
984983
985-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
984+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
986985
do i = sys_size + 1, sys_size + 4
987986
do l = 0, p
988987
do k = -buff_size, -1
@@ -1001,7 +1000,7 @@ contains
10011000
#endif
10021001
#:else
10031002
! Unpacking buffer from bc_z%beg
1004-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1003+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
10051004
do i = 1, sys_size
10061005
do l = -buff_size, -1
10071006
do k = -buff_size, n + buff_size
@@ -1024,7 +1023,7 @@ contains
10241023
10251024
#ifdef MFC_SIMULATION
10261025
if (qbmm .and. .not. polytropic) then
1027-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
1026+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
10281027
do i = sys_size + 1, sys_size + 4
10291028
do l = -buff_size, -1
10301029
do k = -buff_size, n + buff_size
@@ -1041,7 +1040,7 @@ contains
10411040
end do
10421041
end do
10431042
1044-
$:GPU_PARALLEL_LOOP(collapse=5,private=["r"])
1043+
!$acc parallel loop collapse(5) gang vector default(present) private(r)
10451044
do i = sys_size + 1, sys_size + 4
10461045
do l = -buff_size, -1
10471046
do k = -buff_size, n + buff_size
@@ -1089,7 +1088,7 @@ contains
10891088
#ifdef MFC_MPI
10901089
10911090
nVars = num_dims + 1
1092-
$:GPU_UPDATE(device=["nVars"])
1091+
!$acc update device(nVars)
10931092
10941093
buffer_counts = (/ &
10951094
buff_size*nVars*(n + 1)*(p + 1), &
@@ -1131,7 +1130,7 @@ contains
11311130
#:for mpi_dir in [1, 2, 3]
11321131
if (mpi_dir == ${mpi_dir}$) then
11331132
#:if mpi_dir == 1
1134-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1133+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
11351134
do l = 0, p
11361135
do k = 0, n
11371136
do j = 0, buff_size - 1
@@ -1144,7 +1143,7 @@ contains
11441143
end do
11451144
11461145
#:elif mpi_dir == 2
1147-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1146+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
11481147
do i = 1, nVars
11491148
do l = 0, p
11501149
do k = 0, buff_size - 1
@@ -1159,7 +1158,7 @@ contains
11591158
end do
11601159
11611160
#:else
1162-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1161+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
11631162
do i = 1, nVars
11641163
do l = 0, buff_size - 1
11651164
do k = -buff_size, n + buff_size
@@ -1187,7 +1186,7 @@ contains
11871186
call nvtxStartRange("RHS-COMM-SENDRECV-RDMA")
11881187
#:else
11891188
call nvtxStartRange("RHS-COMM-DEV2HOST")
1190-
$:GPU_UPDATE(host=["buff_send"])
1189+
!$acc update host(buff_send)
11911190
call nvtxEndRange
11921191
call nvtxStartRange("RHS-COMM-SENDRECV-NO-RMDA")
11931192
#:endif
@@ -1205,7 +1204,7 @@ contains
12051204
!$acc wait
12061205
#:else
12071206
call nvtxStartRange("RHS-COMM-HOST2DEV")
1208-
$:GPU_UPDATE(device=["buff_recv"])
1207+
!$acc update device(buff_recv)
12091208
call nvtxEndRange
12101209
#:endif
12111210
end if
@@ -1221,7 +1220,7 @@ contains
12211220
#:for mpi_dir in [1, 2, 3]
12221221
if (mpi_dir == ${mpi_dir}$) then
12231222
#:if mpi_dir == 1
1224-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1223+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
12251224
do l = 0, p
12261225
do k = 0, n
12271226
do j = -buff_size, -1
@@ -1241,7 +1240,7 @@ contains
12411240
end do
12421241
12431242
#:elif mpi_dir == 2
1244-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1243+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
12451244
do i = 1, nVars
12461245
do l = 0, p
12471246
do k = -buff_size, -1
@@ -1263,7 +1262,7 @@ contains
12631262
12641263
#:else
12651264
! Unpacking buffer from bc_z%beg
1266-
$:GPU_PARALLEL_LOOP(collapse=4,private=["r"])
1265+
!$acc parallel loop collapse(4) gang vector default(present) private(r)
12671266
do i = 1, nVars
12681267
do l = -buff_size, -1
12691268
do k = -buff_size, n + buff_size
@@ -1301,4 +1300,4 @@ contains
13011300
13021301
end subroutine s_finalize_mpi_common_module
13031302
1304-
end module m_mpi_common
1303+
end module m_mpi_common

0 commit comments

Comments
 (0)