some versions of ifort have problems with 'do concurrent'

This commit is contained in:
Martin Diehl 2022-02-14 08:58:40 +01:00
parent 466682e978
commit c66e2336c2
3 changed files with 42 additions and 27 deletions

View File

@ -376,20 +376,32 @@ subroutine utilities_updateGamma(C)
!$OMP PARALLEL DO PRIVATE(l,m,n,o,temp33_cmplx,xiDyad_cmplx,A,A_inv,err)
do k = cells3Offset+1, cells3Offset+cells3; do j = 1, cells(2); do i = 1, cells1Red
if (any([i,j,k] /= 1)) then ! singular point at xi=(0.0,0.0,0.0) i.e. i=j=k=1
do concurrent (l = 1:3, m = 1:3)
#ifndef __INTEL_COMPILER
do concurrent(l = 1:3, m = 1:3)
xiDyad_cmplx(l,m) = conjg(-xi1st(l,i,j,k-cells3Offset))*xi1st(m,i,j,k-cells3Offset)
end do
do concurrent(l = 1:3, m = 1:3)
temp33_cmplx(l,m) = sum(cmplx(C_ref(l,1:3,m,1:3),0.0_pReal)*xiDyad_cmplx)
end do
#else
forall(l = 1:3, m = 1:3) &
xiDyad_cmplx(l,m) = conjg(-xi1st(l,i,j,k-cells3Offset))*xi1st(m,i,j,k-cells3Offset)
forall(l = 1:3, m = 1:3) &
temp33_cmplx(l,m) = sum(cmplx(C_ref(l,1:3,m,1:3),0.0_pReal)*xiDyad_cmplx)
#endif
A(1:3,1:3) = temp33_cmplx%re; A(4:6,4:6) = temp33_cmplx%re
A(1:3,4:6) = temp33_cmplx%im; A(4:6,1:3) = -temp33_cmplx%im
if (abs(math_det33(A(1:3,1:3))) > 1e-16) then
call math_invert(A_inv, err, A)
temp33_cmplx = cmplx(A_inv(1:3,1:3),A_inv(1:3,4:6),pReal)
#ifndef __INTEL_COMPILER
do concurrent(l=1:3, m=1:3, n=1:3, o=1:3)
gamma_hat(l,m,n,o,i,j,k-cells3Offset) = temp33_cmplx(l,n) * xiDyad_cmplx(o,m)
end do
#else
forall(l=1:3, m=1:3, n=1:3, o=1:3) &
gamma_hat(l,m,n,o,i,j,k-cells3Offset) = temp33_cmplx(l,n) * xiDyad_cmplx(o,m)
#endif
end if
end if
end do; end do; end do
@ -497,23 +509,37 @@ subroutine utilities_fourierGammaConvolution(fieldAim)
!$OMP PARALLEL DO PRIVATE(l,m,n,o,temp33_cmplx,xiDyad_cmplx,A,A_inv,err,gamma_hat)
do k = 1, cells3; do j = 1, cells(2); do i = 1, cells1Red
if (any([i,j,k+cells3Offset] /= 1)) then ! singular point at xi=(0.0,0.0,0.0) i.e. i=j=k=1
#ifndef __INTEL_COMPILER
do concurrent(l = 1:3, m = 1:3)
xiDyad_cmplx(l,m) = conjg(-xi1st(l,i,j,k))*xi1st(m,i,j,k)
end do
do concurrent(l = 1:3, m = 1:3)
temp33_cmplx(l,m) = sum(cmplx(C_ref(l,1:3,m,1:3),0.0_pReal)*xiDyad_cmplx)
end do
#else
forall(l = 1:3, m = 1:3) &
xiDyad_cmplx(l,m) = conjg(-xi1st(l,i,j,k))*xi1st(m,i,j,k)
forall(l = 1:3, m = 1:3) &
temp33_cmplx(l,m) = sum(cmplx(C_ref(l,1:3,m,1:3),0.0_pReal)*xiDyad_cmplx)
#endif
A(1:3,1:3) = temp33_cmplx%re; A(4:6,4:6) = temp33_cmplx%re
A(1:3,4:6) = temp33_cmplx%im; A(4:6,1:3) = -temp33_cmplx%im
if (abs(math_det33(A(1:3,1:3))) > 1e-16) then
call math_invert(A_inv, err, A)
temp33_cmplx = cmplx(A_inv(1:3,1:3),A_inv(1:3,4:6),pReal)
#ifndef __INTEL_COMPILER
do concurrent(l=1:3, m=1:3, n=1:3, o=1:3)
gamma_hat(l,m,n,o,1,1,1) = temp33_cmplx(l,n)*xiDyad_cmplx(o,m)
end do
do concurrent(l = 1:3, m = 1:3)
temp33_cmplx(l,m) = sum(gamma_hat(l,m,1:3,1:3,1,1,1)*tensorField_fourier(1:3,1:3,i,j,k))
end do
#else
forall(l=1:3, m=1:3, n=1:3, o=1:3) &
gamma_hat(l,m,n,o,1,1,1) = temp33_cmplx(l,n)*xiDyad_cmplx(o,m)
forall(l = 1:3, m = 1:3) &
temp33_cmplx(l,m) = sum(gamma_hat(l,m,1:3,1:3,1,1,1)*tensorField_fourier(1:3,1:3,i,j,k))
#endif
tensorField_fourier(1:3,1:3,i,j,k) = temp33_cmplx
else
tensorField_fourier(1:3,1:3,i,j,k) = cmplx(0.0_pReal,0.0_pReal,pReal)
@ -524,9 +550,14 @@ subroutine utilities_fourierGammaConvolution(fieldAim)
else memoryEfficient
!$OMP PARALLEL DO PRIVATE(l,m,temp33_cmplx)
do k = 1, cells3; do j = 1, cells(2); do i = 1,cells1Red
#ifndef __INTEL_COMPILER
do concurrent(l = 1:3, m = 1:3)
temp33_cmplx(l,m) = sum(gamma_hat(l,m,1:3,1:3,i,j,k)*tensorField_fourier(1:3,1:3,i,j,k))
end do
#else
forall(l = 1:3, m = 1:3) &
temp33_cmplx(l,m) = sum(gamma_hat(l,m,1:3,1:3,i,j,k)*tensorField_fourier(1:3,1:3,i,j,k))
#endif
tensorField_fourier(1:3,1:3,i,j,k) = temp33_cmplx
end do; end do; end do
!$OMP END PARALLEL DO
@ -758,7 +789,7 @@ subroutine utilities_fourierVectorDivergence()
scalarField_fourier(1:cells1Red,1:cells(2),1:cells3) = sum(vectorField_fourier(1:3,1:cells1Red,1:cells(2),1:cells3) &
*conjg(-xi1st))
*conjg(-xi1st),1)
end subroutine utilities_fourierVectorDivergence

View File

@ -262,9 +262,8 @@ pure function math_identity4th()
math_identity4th(i,j,k,l) = 0.5_pReal*(math_I3(i,k)*math_I3(j,l)+math_I3(i,l)*math_I3(j,k))
enddo
#else
do i=1,3; do j=1,3; do k=1,3; do l=1,3
forall(i=1:3, j=1:3, k=1:3, l=1:3) &
math_identity4th(i,j,k,l) = 0.5_pReal*(math_I3(i,k)*math_I3(j,l)+math_I3(i,l)*math_I3(j,k))
enddo; enddo; enddo; enddo
#endif
end function math_identity4th
@ -338,9 +337,7 @@ pure function math_outer(A,B)
math_outer(i,j) = A(i)*B(j)
enddo
#else
do i=1,size(A,1); do j=1,size(B,1)
math_outer(i,j) = A(i)*B(j)
enddo; enddo
forall(i=1:size(A,1), j=1:size(B,1)) math_outer(i,j) = A(i)*B(j)
#endif
end function math_outer
@ -387,9 +384,7 @@ pure function math_mul3333xx33(A,B)
math_mul3333xx33(i,j) = sum(A(i,j,1:3,1:3)*B(1:3,1:3))
enddo
#else
do i=1,3; do j=1,3
math_mul3333xx33(i,j) = sum(A(i,j,1:3,1:3)*B(1:3,1:3))
enddo; enddo
forall (i=1:3, j=1:3) math_mul3333xx33(i,j) = sum(A(i,j,1:3,1:3)*B(1:3,1:3))
#endif
end function math_mul3333xx33
@ -411,9 +406,7 @@ pure function math_mul3333xx3333(A,B)
math_mul3333xx3333(i,j,k,l) = sum(A(i,j,1:3,1:3)*B(1:3,1:3,k,l))
enddo
#else
do i=1,3; do j=1,3; do k=1,3; do l=1,3
math_mul3333xx3333(i,j,k,l) = sum(A(i,j,1:3,1:3)*B(1:3,1:3,k,l))
enddo; enddo; enddo; enddo
forall(i=1:3, j=1:3, k=1:3, l=1:3) math_mul3333xx3333(i,j,k,l) = sum(A(i,j,1:3,1:3)*B(1:3,1:3,k,l))
#endif
end function math_mul3333xx3333
@ -752,9 +745,7 @@ pure function math_3333to99(m3333)
math_3333to99(i,j) = m3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j))
enddo
#else
do i=1,9; do j=1,9
math_3333to99(i,j) = m3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j))
enddo; enddo
forall(i=1:9, j=1:9) math_3333to99(i,j) = m3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j))
#endif
end function math_3333to99
@ -775,9 +766,7 @@ pure function math_99to3333(m99)
math_99to3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j)) = m99(i,j)
enddo
#else
do i=1,9; do j=1,9
math_99to3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j)) = m99(i,j)
enddo; enddo
forall(i=1:9, j=1:9) math_99to3333(MAPPLAIN(1,i),MAPPLAIN(2,i),MAPPLAIN(1,j),MAPPLAIN(2,j)) = m99(i,j)
#endif
end function math_99to3333
@ -810,9 +799,7 @@ pure function math_sym3333to66(m3333,weighted)
math_sym3333to66(i,j) = w(i)*w(j)*m3333(MAPNYE(1,i),MAPNYE(2,i),MAPNYE(1,j),MAPNYE(2,j))
enddo
#else
do i=1,6; do j=1,6
math_sym3333to66(i,j) = w(i)*w(j)*m3333(MAPNYE(1,i),MAPNYE(2,i),MAPNYE(1,j),MAPNYE(2,j))
enddo; enddo
forall(i=1:6, j=1:6) math_sym3333to66(i,j) = w(i)*w(j)*m3333(MAPNYE(1,i),MAPNYE(2,i),MAPNYE(1,j),MAPNYE(2,j))
#endif
end function math_sym3333to66
@ -950,9 +937,7 @@ pure function math_3333toVoigt66_stiffness(C) result(C_tilde)
C_tilde(i,j) = C(MAPVOIGT(1,i),MAPVOIGT(2,i),MAPVOIGT(1,j),MAPVOIGT(2,j))
end do
#else
do i=1,6; do j=1,6
C_tilde(i,j) = C(MAPVOIGT(1,i),MAPVOIGT(2,i),MAPVOIGT(1,j),MAPVOIGT(2,j))
end do; end do
forall(i=1:6, j=1:6) C_tilde(i,j) = C(MAPVOIGT(1,i),MAPVOIGT(2,i),MAPVOIGT(1,j),MAPVOIGT(2,j))
#endif
end function math_3333toVoigt66_stiffness

View File

@ -379,10 +379,9 @@ module function plastic_deltaState(ph, en) result(broken)
en
logical :: broken
real(pReal), dimension(3,3) :: &
real(pReal), dimension(3,3) :: &
Mp
integer :: &
myOffset, &
mySize