Commit 5eda3ee6 authored by Andreas Marek's avatar Andreas Marek
Browse files

Correct settings for kernel choice at compile time

parent c32bdb9e
...@@ -85,6 +85,12 @@ ...@@ -85,6 +85,12 @@
/* can use complex SSE kernel */ /* can use complex SSE kernel */
#undef WITH_COMPLEX_SSE_KERNEL #undef WITH_COMPLEX_SSE_KERNEL
/* do not use only one specific complex kernel (set at compile time) */
#undef WITH_NO_SPECIFIC_COMPLEX_KERNEL
/* do not use only one specific real kernel (set at compile time) */
#undef WITH_NO_SPECIFIC_REAL_KERNEL
/* use OpenMP threading */ /* use OpenMP threading */
#undef WITH_OPENMP #undef WITH_OPENMP
...@@ -111,6 +117,3 @@ ...@@ -111,6 +117,3 @@
/* can use real SSE kernel */ /* can use real SSE kernel */
#undef WITH_REAL_SSE_KERNEL #undef WITH_REAL_SSE_KERNEL
/* use specific real kernel */
#undef WITH_SPECIFIC_COMPLEX_KERNEL
...@@ -7960,15 +7960,15 @@ $as_echo "#define WITH_REAL_BGQ_KERNEL 1" >>confdefs.h ...@@ -7960,15 +7960,15 @@ $as_echo "#define WITH_REAL_BGQ_KERNEL 1" >>confdefs.h
   
fi fi
   
if test x"${use_specific_complex_kernel}" = x"yes" ; then if test x"${use_specific_complex_kernel}" = x"no" ; then
   
$as_echo "#define WITH_SPECIFIC_COMPLEX_KERNEL 1" >>confdefs.h $as_echo "#define WITH_NO_SPECIFIC_COMPLEX_KERNEL 1" >>confdefs.h
   
fi fi
   
if test x"${use_specific_real_kernel}" = x"yes" ; then if test x"${use_specific_real_kernel}" = x"no" ; then
   
$as_echo "#define WITH_SPECIFIC_COMPLEX_KERNEL 1" >>confdefs.h $as_echo "#define WITH_NO_SPECIFIC_REAL_KERNEL 1" >>confdefs.h
   
fi fi
   
......
...@@ -949,7 +949,7 @@ if test x"${fortran_can_check_environment}" = x"yes" ; then ...@@ -949,7 +949,7 @@ if test x"${fortran_can_check_environment}" = x"yes" ; then
AC_DEFINE([HAVE_ENVIRONMENT_CHECKING],[1],[Fortran can querry environment variables]) AC_DEFINE([HAVE_ENVIRONMENT_CHECKING],[1],[Fortran can querry environment variables])
fi fi
dnl last check whether user wants to compile ony a specific kernel dnl last check whether user wants to compile only a specific kernel
dnl dnl
dnl real kernels dnl real kernels
dnl dnl
...@@ -1437,12 +1437,12 @@ if test x"${install_real_bgq}" = x"yes" ; then ...@@ -1437,12 +1437,12 @@ if test x"${install_real_bgq}" = x"yes" ; then
AC_DEFINE([WITH_REAL_BGQ_KERNEL],[1],[can use real BGQ kernel]) AC_DEFINE([WITH_REAL_BGQ_KERNEL],[1],[can use real BGQ kernel])
fi fi
if test x"${use_specific_complex_kernel}" = x"yes" ; then if test x"${use_specific_complex_kernel}" = x"no" ; then
AC_DEFINE([WITH_SPECIFIC_COMPLEX_KERNEL],[1],[use specific complex kernel]) AC_DEFINE([WITH_NO_SPECIFIC_COMPLEX_KERNEL],[1],[do not use only one specific complex kernel (set at compile time)])
fi fi
if test x"${use_specific_real_kernel}" = x"yes" ; then if test x"${use_specific_real_kernel}" = x"no" ; then
AC_DEFINE([WITH_SPECIFIC_COMPLEX_KERNEL],[1],[use specific real kernel]) AC_DEFINE([WITH_NO_SPECIFIC_REAL_KERNEL],[1],[do not use only one specific real kernel (set at compile time)])
fi fi
LT_INIT LT_INIT
......
...@@ -2735,7 +2735,7 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2735,7 +2735,7 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
endif endif
#endif #endif
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK2 .or. & if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK2 .or. &
THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC .or. & THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC .or. &
THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC_SIMPLE .or. & THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC_SIMPLE .or. &
...@@ -2743,13 +2743,13 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2743,13 +2743,13 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_SSE .or. & THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_SSE .or. &
THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGP .or. & THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGP .or. &
THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGQ) then THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGQ) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
!FORTRAN CODE / X86 INRINISIC CODE / BG ASSEMBLER USING 2 HOUSEHOLDER VECTORS !FORTRAN CODE / X86 INRINISIC CODE / BG ASSEMBLER USING 2 HOUSEHOLDER VECTORS
#if defined(WITH_REAL_GENERIC_KERNEL) #if defined(WITH_REAL_GENERIC_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2761,16 +2761,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2761,16 +2761,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
nbw, nl, stripe_width, nbw) nbw, nl, stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_GENERIC_KERNEL */ #endif /* WITH_REAL_GENERIC_KERNEL */
#if defined(WITH_REAL_GENERIC_SIMPLE_KERNEL) #if defined(WITH_REAL_GENERIC_SIMPLE_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC_SIMPLE) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_GENERIC_SIMPLE) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2782,16 +2782,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2782,16 +2782,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
w, nbw, nl, stripe_width, nbw) w, nbw, nl, stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_GENERIC_SIMPLE_KERNEL */ #endif /* WITH_REAL_GENERIC_SIMPLE_KERNEL */
#if defined(WITH_REAL_SSE_KERNEL) #if defined(WITH_REAL_SSE_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_SSE) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_SSE) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2803,16 +2803,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2803,16 +2803,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
stripe_width, nbw) stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_SSE_KERNEL */ #endif /* WITH_REAL_SSE_KERNEL */
#if defined(WITH_REAL_AVX_BLOCK2_KERNEL) #if defined(WITH_REAL_AVX_BLOCK2_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK2) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK2) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2824,15 +2824,15 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2824,15 +2824,15 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
w, nbw, nl, stripe_width, nbw) w, nbw, nl, stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_AVX_BLOCK2_KERNEL */ #endif /* WITH_REAL_AVX_BLOCK2_KERNEL */
#if defined(WITH_REAL_BGP_KERNEL) #if defined(WITH_REAL_BGP_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGP) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGP) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2844,16 +2844,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2844,16 +2844,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
stripe_width, nbw) stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_BGP_KERNEL */ #endif /* WITH_REAL_BGP_KERNEL */
#if defined(WITH_REAL_BGQ_KERNEL) #if defined(WITH_REAL_BGQ_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGQ) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_BGQ) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
w(:,2) = bcast_buffer(1:nbw,j+off-1) w(:,2) = bcast_buffer(1:nbw,j+off-1)
...@@ -2865,9 +2865,9 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2865,9 +2865,9 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
stripe_width, nbw) stripe_width, nbw)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_BGQ_KERNEL */ #endif /* WITH_REAL_BGQ_KERNEL */
...@@ -2886,16 +2886,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2886,16 +2886,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
#endif #endif
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif ! endif !
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#if defined(WITH_REAL_AVX_BLOCK4_KERNEL) #if defined(WITH_REAL_AVX_BLOCK4_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK4) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK4) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
! X86 INTRINSIC CODE, USING 4 HOUSEHOLDER VECTORS ! X86 INTRINSIC CODE, USING 4 HOUSEHOLDER VECTORS
do j = ncols, 4, -4 do j = ncols, 4, -4
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
...@@ -2928,16 +2928,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2928,16 +2928,16 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
if(jj==1) call single_hh_trafo(a(1,1+off+a_off,istripe), & if(jj==1) call single_hh_trafo(a(1,1+off+a_off,istripe), &
bcast_buffer(1,off+1), nbw, nl, stripe_width) bcast_buffer(1,off+1), nbw, nl, stripe_width)
#endif #endif
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_AVX_BLOCK4_KERNEL */ #endif /* WITH_REAL_AVX_BLOCK4_KERNEL */
#if defined(WITH_REAL_AVX_BLOCK6_KERNEL) #if defined(WITH_REAL_AVX_BLOCK6_KERNEL)
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK6) then if (THIS_REAL_ELPA_KERNEL .eq. REAL_ELPA_KERNEL_AVX_BLOCK6) then
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
! X86 INTRINSIC CODE, USING 6 HOUSEHOLDER VECTORS ! X86 INTRINSIC CODE, USING 6 HOUSEHOLDER VECTORS
do j = ncols, 6, -6 do j = ncols, 6, -6
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
...@@ -2985,9 +2985,9 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, & ...@@ -2985,9 +2985,9 @@ subroutine trans_ev_tridi_to_band_real(na, nev, nblk, nbw, q, ldq, &
if(jjj==1) call single_hh_trafo(a(1,1+off+a_off,istripe), & if(jjj==1) call single_hh_trafo(a(1,1+off+a_off,istripe), &
bcast_buffer(1,off+1), nbw, nl, stripe_width) bcast_buffer(1,off+1), nbw, nl, stripe_width)
#endif #endif
#if defined(WITH_SPECIFIC_REAL_KERNEL) #if defined(WITH_NO_SPECIFIC_REAL_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_REAL_KERNEL */ #endif /* WITH_NO_SPECIFIC_REAL_KERNEL */
#endif /* WITH_REAL_AVX_BLOCK4_KERNEL */ #endif /* WITH_REAL_AVX_BLOCK4_KERNEL */
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
...@@ -5084,9 +5084,9 @@ contains ...@@ -5084,9 +5084,9 @@ contains
#if defined(WITH_COMPLEX_AVX_BLOCK2_KERNEL) #if defined(WITH_COMPLEX_AVX_BLOCK2_KERNEL)
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_AVX_BLOCK2) then if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_AVX_BLOCK2) then
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
ttt = mpi_wtime() ttt = mpi_wtime()
do j = ncols, 2, -2 do j = ncols, 2, -2
w(:,1) = bcast_buffer(1:nbw,j+off) w(:,1) = bcast_buffer(1:nbw,j+off)
...@@ -5106,16 +5106,16 @@ contains ...@@ -5106,16 +5106,16 @@ contains
if(j==1) call single_hh_trafo_complex_sse_avx_1hv(a(1,1+off+a_off,istripe), & if(j==1) call single_hh_trafo_complex_sse_avx_1hv(a(1,1+off+a_off,istripe), &
bcast_buffer(1,off+1), nbw, nl, stripe_width) bcast_buffer(1,off+1), nbw, nl, stripe_width)
#endif #endif
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
#endif /* WITH_COMPLEX_AVX_BLOCK2_KERNEL */ #endif /* WITH_COMPLEX_AVX_BLOCK2_KERNEL */
#if defined(WITH_COMPLEX_GENERIC_SIMPLE_KERNEL) #if defined(WITH_COMPLEX_GENERIC_SIMPLE_KERNEL)
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_GENERIC_SIMPLE) then if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_GENERIC_SIMPLE) then
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
ttt = mpi_wtime() ttt = mpi_wtime()
do j = ncols, 1, -1 do j = ncols, 1, -1
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
...@@ -5126,17 +5126,17 @@ contains ...@@ -5126,17 +5126,17 @@ contains
bcast_buffer(1,j+off),nbw,nl,stripe_width) bcast_buffer(1,j+off),nbw,nl,stripe_width)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
#endif /* WITH_COMPLEX_GENERIC_SIMPLE_KERNEL */ #endif /* WITH_COMPLEX_GENERIC_SIMPLE_KERNEL */
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_GENERIC .or. & if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_GENERIC .or. &
THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_BGP .or. & THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_BGP .or. &
THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_BGQ ) then THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_BGQ ) then
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
ttt = mpi_wtime() ttt = mpi_wtime()
do j = ncols, 1, -1 do j = ncols, 1, -1
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
...@@ -5147,15 +5147,15 @@ contains ...@@ -5147,15 +5147,15 @@ contains
bcast_buffer(1,j+off),nbw,nl,stripe_width) bcast_buffer(1,j+off),nbw,nl,stripe_width)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
#if defined(WITH_COMPLEX_SSE_KERNEL) #if defined(WITH_COMPLEX_SSE_KERNEL)
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_SSE) then if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_SSE) then
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
ttt = mpi_wtime() ttt = mpi_wtime()
do j = ncols, 1, -1 do j = ncols, 1, -1
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
...@@ -5166,9 +5166,9 @@ contains ...@@ -5166,9 +5166,9 @@ contains
bcast_buffer(1,j+off),nbw,nl,stripe_width) bcast_buffer(1,j+off),nbw,nl,stripe_width)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
#endif /* WITH_COMPLEX_SSE_KERNEL */ #endif /* WITH_COMPLEX_SSE_KERNEL */
...@@ -5181,9 +5181,9 @@ contains ...@@ -5181,9 +5181,9 @@ contains
!#endif !#endif
#if defined(WITH_COMPLEX_AVX_BLOCK1_KERNEL) #if defined(WITH_COMPLEX_AVX_BLOCK1_KERNEL)
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_AVX_BLOCK1) then if (THIS_COMPLEX_ELPA_KERNEL .eq. COMPLEX_ELPA_KERNEL_AVX_BLOCK1) then
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
ttt = mpi_wtime() ttt = mpi_wtime()
do j = ncols, 1, -1 do j = ncols, 1, -1
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
...@@ -5194,9 +5194,9 @@ contains ...@@ -5194,9 +5194,9 @@ contains
bcast_buffer(1,j+off),nbw,nl,stripe_width) bcast_buffer(1,j+off),nbw,nl,stripe_width)
#endif #endif
enddo enddo
#if defined(WITH_SPECIFIC_COMPLEX_KERNEL) #if defined(WITH_NO_SPECIFIC_COMPLEX_KERNEL)
endif endif
#endif /* WITH_SPECIFIC_COMPLEX_KERNEL */ #endif /* WITH_NO_SPECIFIC_COMPLEX_KERNEL */
#endif /* WITH_COMPLEX_AVX_BLOCK1_KERNE */ #endif /* WITH_COMPLEX_AVX_BLOCK1_KERNE */
#ifdef WITH_OPENMP #ifdef WITH_OPENMP
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment