Skip to content

Commit abf2f5a

Browse files
author
Daniel Vickers
committed
Got errors with the first 9 loops, so now I removed 4 more loops. 5 total loops in this test with new macros
1 parent d5a3f74 commit abf2f5a

File tree

1 file changed

+47
-47
lines changed

1 file changed

+47
-47
lines changed

‎src/common/m_mpi_common.fpp‎

Lines changed: 47 additions & 47 deletions
Original file line numberDiff line numberDiff line change
@@ -837,73 +837,73 @@ contains
837837
end do
838838
$:END_GPU_PARALLEL_LOOP()
839839
840-
$:GPU_PARALLEL_LOOP(collapse=5,private='[r]')
841-
do i = nVar + 1, nVar + 4
842-
do l = 0, p
843-
do k = 0, buff_size - 1
844-
do j = -buff_size, m + buff_size
845-
do q = 1, nb
846-
r = (i - 1) + (q - 1)*4 + nb*4 + v_size* &
847-
((j + buff_size) + (m + 2*buff_size + 1)* &
848-
(k + buff_size*l))
849-
buff_send(r) = real(mv_in(j, k + pack_offset, l, i - nVar, q), kind=wp)
840+
#:call GPU_PARALLEL_LOOP_OLD(collapse=5,private='[r]')
841+
do i = nVar + 1, nVar + 4
842+
do l = 0, p
843+
do k = 0, buff_size - 1
844+
do j = -buff_size, m + buff_size
845+
do q = 1, nb
846+
r = (i - 1) + (q - 1)*4 + nb*4 + v_size* &
847+
((j + buff_size) + (m + 2*buff_size + 1)* &
848+
(k + buff_size*l))
849+
buff_send(r) = real(mv_in(j, k + pack_offset, l, i - nVar, q), kind=wp)
850+
end do
850851
end do
851852
end do
852853
end do
853854
end do
854-
end do
855-
$:END_GPU_PARALLEL_LOOP()
855+
#:endcall GPU_PARALLEL_LOOP_OLD
856856
end if
857857
#:else
858-
$:GPU_PARALLEL_LOOP(collapse=4,private='[r]')
859-
do i = 1, nVar
860-
do l = 0, buff_size - 1
861-
do k = -buff_size, n + buff_size
862-
do j = -buff_size, m + buff_size
863-
r = (i - 1) + v_size* &
864-
((j + buff_size) + (m + 2*buff_size + 1)* &
865-
((k + buff_size) + (n + 2*buff_size + 1)*l))
866-
buff_send(r) = real(q_comm(i)%sf(j, k, l + pack_offset), kind=wp)
858+
#:call GPU_PARALLEL_LOOP_OLD(collapse=4,private='[r]')
859+
do i = 1, nVar
860+
do l = 0, buff_size - 1
861+
do k = -buff_size, n + buff_size
862+
do j = -buff_size, m + buff_size
863+
r = (i - 1) + v_size* &
864+
((j + buff_size) + (m + 2*buff_size + 1)* &
865+
((k + buff_size) + (n + 2*buff_size + 1)*l))
866+
buff_send(r) = real(q_comm(i)%sf(j, k, l + pack_offset), kind=wp)
867+
end do
867868
end do
868869
end do
869870
end do
870-
end do
871-
$:END_GPU_PARALLEL_LOOP()
871+
#:endcall GPU_PARALLEL_LOOP_OLD
872872
873873
if (qbmm_comm) then
874-
$:GPU_PARALLEL_LOOP(collapse=5,private='[r]')
875-
do i = nVar + 1, nVar + 4
876-
do l = 0, buff_size - 1
877-
do k = -buff_size, n + buff_size
878-
do j = -buff_size, m + buff_size
879-
do q = 1, nb
880-
r = (i - 1) + (q - 1)*4 + v_size* &
881-
((j + buff_size) + (m + 2*buff_size + 1)* &
882-
((k + buff_size) + (n + 2*buff_size + 1)*l))
883-
buff_send(r) = real(pb_in(j, k, l + pack_offset, i - nVar, q), kind=wp)
874+
#:call GPU_PARALLEL_LOOP_OLD(collapse=5,private='[r]')
875+
do i = nVar + 1, nVar + 4
876+
do l = 0, buff_size - 1
877+
do k = -buff_size, n + buff_size
878+
do j = -buff_size, m + buff_size
879+
do q = 1, nb
880+
r = (i - 1) + (q - 1)*4 + v_size* &
881+
((j + buff_size) + (m + 2*buff_size + 1)* &
882+
((k + buff_size) + (n + 2*buff_size + 1)*l))
883+
buff_send(r) = real(pb_in(j, k, l + pack_offset, i - nVar, q), kind=wp)
884+
end do
884885
end do
885886
end do
886887
end do
887888
end do
888-
end do
889-
$:END_GPU_PARALLEL_LOOP()
889+
#:endcall GPU_PARALLEL_LOOP_OLD
890890
891-
$:GPU_PARALLEL_LOOP(collapse=5,private='[r]')
892-
do i = nVar + 1, nVar + 4
893-
do l = 0, buff_size - 1
894-
do k = -buff_size, n + buff_size
895-
do j = -buff_size, m + buff_size
896-
do q = 1, nb
897-
r = (i - 1) + (q - 1)*4 + nb*4 + v_size* &
898-
((j + buff_size) + (m + 2*buff_size + 1)* &
899-
((k + buff_size) + (n + 2*buff_size + 1)*l))
900-
buff_send(r) = real(mv_in(j, k, l + pack_offset, i - nVar, q), kind=wp)
891+
#:call GPU_PARALLEL_LOOP_OLD(collapse=5,private='[r]')
892+
do i = nVar + 1, nVar + 4
893+
do l = 0, buff_size - 1
894+
do k = -buff_size, n + buff_size
895+
do j = -buff_size, m + buff_size
896+
do q = 1, nb
897+
r = (i - 1) + (q - 1)*4 + nb*4 + v_size* &
898+
((j + buff_size) + (m + 2*buff_size + 1)* &
899+
((k + buff_size) + (n + 2*buff_size + 1)*l))
900+
buff_send(r) = real(mv_in(j, k, l + pack_offset, i - nVar, q), kind=wp)
901+
end do
901902
end do
902903
end do
903904
end do
904905
end do
905-
end do
906-
$:END_GPU_PARALLEL_LOOP()
906+
#:endcall GPU_PARALLEL_LOOP_OLD
907907
end if
908908
#:endif
909909
end if

0 commit comments

Comments
 (0)