对lopsided和kodis进行合并,减少symmetry_bd开销,有0.01~0.02s单步效果

This commit is contained in:
2026-02-08 23:21:54 +08:00
parent f345b0e520
commit c5f8a18ba4
6 changed files with 789 additions and 1240 deletions

View File

@@ -1000,86 +1000,7 @@
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
#if 0
! x direction
if(i+2 <= imax .and. i-2 >= imin)then
!
! f(i-2) - 8 f(i-1) + 8 f(i+1) - f(i+2)
! fx(i) = ---------------------------------------------
! 12 dx
fx(i,j,k)=d12dx*(fh(i-2,j,k)-EIT*fh(i-1,j,k)+EIT*fh(i+1,j,k)-fh(i+2,j,k))
elseif(i+1 <= imax .and. i-1 >= imin)then
!
! - f(i-1) + f(i+1)
! fx(i) = --------------------------------
! 2 dx
fx(i,j,k)=d2dx*(-fh(i-1,j,k)+fh(i+1,j,k))
! set imax and imin 0
endif
! y direction
if(j+2 <= jmax .and. j-2 >= jmin)then
fy(i,j,k)=d12dy*(fh(i,j-2,k)-EIT*fh(i,j-1,k)+EIT*fh(i,j+1,k)-fh(i,j+2,k))
elseif(j+1 <= jmax .and. j-1 >= jmin)then
fy(i,j,k)=d2dy*(-fh(i,j-1,k)+fh(i,j+1,k))
! set jmax and jmin 0
endif
! z direction
if(k+2 <= kmax .and. k-2 >= kmin)then
fz(i,j,k)=d12dz*(fh(i,j,k-2)-EIT*fh(i,j,k-1)+EIT*fh(i,j,k+1)-fh(i,j,k+2))
elseif(k+1 <= kmax .and. k-1 >= kmin)then
fz(i,j,k)=d2dz*(-fh(i,j,k-1)+fh(i,j,k+1))
! set kmax and kmin 0
endif
#elif 0
! x direction
if(i+2 <= imax .and. i-2 >= imin)then
!
! f(i-2) - 8 f(i-1) + 8 f(i+1) - f(i+2)
! fx(i) = ---------------------------------------------
! 12 dx
fx(i,j,k)=d12dx*(fh(i-2,j,k)-EIT*fh(i-1,j,k)+EIT*fh(i+1,j,k)-fh(i+2,j,k))
elseif(i+3 <= imax .and. i-1 >= imin)then
fx(i,j,k)=d12dx*(-3.d0*fh(i-1,j,k)-1.d1*fh(i,j,k)+1.8d1*fh(i+1,j,k)-6.d0*fh(i+2,j,k)+fh(i+3,j,k))
elseif(i+1 <= imax .and. i-3 >= imin)then
fx(i,j,k)=d12dx*( 3.d0*fh(i+1,j,k)+1.d1*fh(i,j,k)-1.8d1*fh(i-1,j,k)+6.d0*fh(i-2,j,k)-fh(i-3,j,k))
! set imax and imin 0
endif
! y direction
if(j+2 <= jmax .and. j-2 >= jmin)then
fy(i,j,k)=d12dy*(fh(i,j-2,k)-EIT*fh(i,j-1,k)+EIT*fh(i,j+1,k)-fh(i,j+2,k))
elseif(j+3 <= jmax .and. j-1 >= jmin)then
fy(i,j,k)=d12dy*(-3.d0*fh(i,j-1,k)-1.d1*fh(i,j,k)+1.8d1*fh(i,j+1,k)-6.d0*fh(i,j+2,k)+fh(i,j+3,k))
elseif(j+1 <= jmax .and. j-3 >= jmin)then
fy(i,j,k)=d12dy*( 3.d0*fh(i,j+1,k)+1.d1*fh(i,j,k)-1.8d1*fh(i,j-1,k)+6.d0*fh(i,j-2,k)-fh(i,j-3,k))
! set jmax and jmin 0
endif
! z direction
if(k+2 <= kmax .and. k-2 >= kmin)then
fz(i,j,k)=d12dz*(fh(i,j,k-2)-EIT*fh(i,j,k-1)+EIT*fh(i,j,k+1)-fh(i,j,k+2))
elseif(k+3 <= kmax .and. k-1 >= kmin)then
fz(i,j,k)=d12dz*(-3.d0*fh(i,j,k-1)-1.d1*fh(i,j,k)+1.8d1*fh(i,j,k+1)-6.d0*fh(i,j,k+2)+fh(i,j,k+3))
elseif(k+1 <= kmax .and. k-3 >= kmin)then
fz(i,j,k)=d12dz*( 3.d0*fh(i,j,k+1)+1.d1*fh(i,j,k)-1.8d1*fh(i,j,k-1)+6.d0*fh(i,j,k-2)-fh(i,j,k-3))
! set kmax and kmin 0
endif
#else
! for bam comparison
if(i+2 <= imax .and. i-2 >= imin .and. &
j+2 <= jmax .and. j-2 >= jmin .and. &
@@ -1094,7 +1015,7 @@
fy(i,j,k)=d2dy*(-fh(i,j-1,k)+fh(i,j+1,k))
fz(i,j,k)=d2dz*(-fh(i,j,k-1)+fh(i,j,k+1))
endif
#endif
enddo
enddo
enddo
@@ -1404,85 +1325,7 @@
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
#if 0
!~~~~~~ fxx
if(i+2 <= imax .and. i-2 >= imin)then
!
! - f(i-2) + 16 f(i-1) - 30 f(i) + 16 f(i+1) - f(i+2)
! fxx(i) = ----------------------------------------------------------
! 12 dx^2
fxx(i,j,k) = Fdxdx*(-fh(i-2,j,k)+F16*fh(i-1,j,k)-F30*fh(i,j,k) &
-fh(i+2,j,k)+F16*fh(i+1,j,k) )
elseif(i+1 <= imax .and. i-1 >= imin)then
!
! f(i-1) - 2 f(i) + f(i+1)
! fxx(i) = --------------------------------
! dx^2
fxx(i,j,k) = Sdxdx*(fh(i-1,j,k)-TWO*fh(i,j,k) &
+fh(i+1,j,k) )
endif
!~~~~~~ fyy
if(j+2 <= jmax .and. j-2 >= jmin)then
fyy(i,j,k) = Fdydy*(-fh(i,j-2,k)+F16*fh(i,j-1,k)-F30*fh(i,j,k) &
-fh(i,j+2,k)+F16*fh(i,j+1,k) )
elseif(j+1 <= jmax .and. j-1 >= jmin)then
fyy(i,j,k) = Sdydy*(fh(i,j-1,k)-TWO*fh(i,j,k) &
+fh(i,j+1,k) )
endif
!~~~~~~ fzz
if(k+2 <= kmax .and. k-2 >= kmin)then
fzz(i,j,k) = Fdzdz*(-fh(i,j,k-2)+F16*fh(i,j,k-1)-F30*fh(i,j,k) &
-fh(i,j,k+2)+F16*fh(i,j,k+1) )
elseif(k+1 <= kmax .and. k-1 >= kmin)then
fzz(i,j,k) = Sdzdz*(fh(i,j,k-1)-TWO*fh(i,j,k) &
+fh(i,j,k+1) )
endif
!~~~~~~ fxy
if(i+2 <= imax .and. i-2 >= imin .and. j+2 <= jmax .and. j-2 >= jmin)then
!
! ( f(i-2,j-2) - 8 f(i-1,j-2) + 8 f(i+1,j-2) - f(i+2,j-2) )
! - 8 ( f(i-2,j-1) - 8 f(i-1,j-1) + 8 f(i+1,j-1) - f(i+2,j-1) )
! + 8 ( f(i-2,j+1) - 8 f(i-1,j+1) + 8 f(i+1,j+1) - f(i+2,j+1) )
! - ( f(i-2,j+2) - 8 f(i-1,j+2) + 8 f(i+1,j+2) - f(i+2,j+2) )
! fxy(i,j) = ----------------------------------------------------------------
! 144 dx dy
fxy(i,j,k) = Fdxdy*( (fh(i-2,j-2,k)-F8*fh(i-1,j-2,k)+F8*fh(i+1,j-2,k)-fh(i+2,j-2,k)) &
-F8 *(fh(i-2,j-1,k)-F8*fh(i-1,j-1,k)+F8*fh(i+1,j-1,k)-fh(i+2,j-1,k)) &
+F8 *(fh(i-2,j+1,k)-F8*fh(i-1,j+1,k)+F8*fh(i+1,j+1,k)-fh(i+2,j+1,k)) &
- (fh(i-2,j+2,k)-F8*fh(i-1,j+2,k)+F8*fh(i+1,j+2,k)-fh(i+2,j+2,k)))
elseif(i+1 <= imax .and. i-1 >= imin .and. j+1 <= jmax .and. j-1 >= jmin)then
! f(i-1,j-1) - f(i+1,j-1) - f(i-1,j+1) + f(i+1,j+1)
! fxy(i,j) = -----------------------------------------------------------
! 4 dx dy
fxy(i,j,k) = Sdxdy*(fh(i-1,j-1,k)-fh(i+1,j-1,k)-fh(i-1,j+1,k)+fh(i+1,j+1,k))
endif
!~~~~~~ fxz
if(i+2 <= imax .and. i-2 >= imin .and. k+2 <= kmax .and. k-2 >= kmin)then
fxz(i,j,k) = Fdxdz*( (fh(i-2,j,k-2)-F8*fh(i-1,j,k-2)+F8*fh(i+1,j,k-2)-fh(i+2,j,k-2)) &
-F8 *(fh(i-2,j,k-1)-F8*fh(i-1,j,k-1)+F8*fh(i+1,j,k-1)-fh(i+2,j,k-1)) &
+F8 *(fh(i-2,j,k+1)-F8*fh(i-1,j,k+1)+F8*fh(i+1,j,k+1)-fh(i+2,j,k+1)) &
- (fh(i-2,j,k+2)-F8*fh(i-1,j,k+2)+F8*fh(i+1,j,k+2)-fh(i+2,j,k+2)))
elseif(i+1 <= imax .and. i-1 >= imin .and. k+1 <= kmax .and. k-1 >= kmin)then
fxz(i,j,k) = Sdxdz*(fh(i-1,j,k-1)-fh(i+1,j,k-1)-fh(i-1,j,k+1)+fh(i+1,j,k+1))
endif
!~~~~~~ fyz
if(j+2 <= jmax .and. j-2 >= jmin .and. k+2 <= kmax .and. k-2 >= kmin)then
fyz(i,j,k) = Fdydz*( (fh(i,j-2,k-2)-F8*fh(i,j-1,k-2)+F8*fh(i,j+1,k-2)-fh(i,j+2,k-2)) &
-F8 *(fh(i,j-2,k-1)-F8*fh(i,j-1,k-1)+F8*fh(i,j+1,k-1)-fh(i,j+2,k-1)) &
+F8 *(fh(i,j-2,k+1)-F8*fh(i,j-1,k+1)+F8*fh(i,j+1,k+1)-fh(i,j+2,k+1)) &
- (fh(i,j-2,k+2)-F8*fh(i,j-1,k+2)+F8*fh(i,j+1,k+2)-fh(i,j+2,k+2)))
elseif(j+1 <= jmax .and. j-1 >= jmin .and. k+1 <= kmax .and. k-1 >= kmin)then
fyz(i,j,k) = Sdydz*(fh(i,j-1,k-1)-fh(i,j+1,k-1)-fh(i,j-1,k+1)+fh(i,j+1,k+1))
endif
#else
! for bam comparison
if(i+2 <= imax .and. i-2 >= imin .and. &
j+2 <= jmax .and. j-2 >= jmin .and. &
@@ -1518,7 +1361,7 @@
fxz(i,j,k) = Sdxdz*(fh(i-1,j,k-1)-fh(i+1,j,k-1)-fh(i-1,j,k+1)+fh(i+1,j,k+1))
fyz(i,j,k) = Sdydz*(fh(i,j-1,k-1)-fh(i,j+1,k-1)-fh(i,j-1,k+1)+fh(i,j+1,k+1))
endif
#endif
enddo
enddo
enddo