Fix real block6 kernel on power8

Makefile.am

@@ -250,13 +250,13 @@ endif
 #  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sparc64_4hv_single_precision.c
 #endif
 #endif
-#
-#if WITH_REAL_VSX_BLOCK4_KERNEL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_4hv_double_precision.c
-#if WANT_SINGLE_PRECISION_REAL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_4hv_single_precision.c
-#endif
-#endif
+
+if WITH_REAL_VSX_BLOCK4_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_4hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_4hv_single_precision.c
+endif
+endif
 
 if WITH_REAL_SSE_BLOCK4_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sse_4hv_double_precision.c
@@ -286,19 +286,19 @@ if WANT_SINGLE_PRECISION_REAL
 endif
 endif
 
-#if WITH_REAL_SPARC64_BLOCK6_KERNEL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sparc64_6hv_double_precision.c
-#if WANT_SINGLE_PRECISION_REAL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sparc64_6hv_single_precision.c
-#endif
-#endif
-#
-#if WITH_REAL_VSX_BLOCK6_KERNEL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_6hv_double_precision.c
-#if WANT_SINGLE_PRECISION_REAL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_6hv_single_precision.c
-#endif
-#endif
+if WITH_REAL_SPARC64_BLOCK6_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sparc64_6hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sparc64_6hv_single_precision.c
+endif
+endif
+
+if WITH_REAL_VSX_BLOCK6_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_6hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_6hv_single_precision.c
+endif
+endif
 
 if WITH_REAL_SSE_BLOCK6_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_sse_6hv_double_precision.c

src/elpa2/compute_hh_trafo.F90

@@ -670,19 +670,19 @@
 #if (!defined(WITH_FIXED_COMPLEX_KERNEL)) || (defined(WITH_FIXED_COMPLEX_KERNEL) && !defined(WITH_COMPLEX_VSX_BLOCK2_KERNEL))
             ttt = mpi_wtime()
             do j = ncols, 1, -1
-#ifdef WITH_OPENMP
-              call single_hh_trafo_&
-                   &MATH_DATATYPE&
-                   &_vsx_1hv_&
-                   &PRECISION&
-                   & (c_loc(a(1,j+off+a_off,istripe,my_thread)), bcast_buffer(1,j+off),nbw,nl,stripe_width)
-#else
-              call single_hh_trafo_&
-                   &MATH_DATATYPE&
-                   &_vsx_1hv_&
-                   &PRECISION&
-                   & (c_loc(a(1,j+off+a_off,istripe)), bcast_buffer(1,j+off),nbw,nl,stripe_width)
-#endif
+!#ifdef WITH_OPENMP
+!              call single_hh_trafo_&
+!                   &MATH_DATATYPE&
+!                   &_vsx_1hv_&
+!                   &PRECISION&
+!                   & (c_loc(a(1,j+off+a_off,istripe,my_thread)), bcast_buffer(1,j+off),nbw,nl,stripe_width)
+!#else
+!              call single_hh_trafo_&
+!                   &MATH_DATATYPE&
+!                   &_vsx_1hv_&
+!                   &PRECISION&
+!                   & (c_loc(a(1,j+off+a_off,istripe)), bcast_buffer(1,j+off),nbw,nl,stripe_width)
+!#endif
             enddo
 #endif /* (!defined(WITH_FIXED_COMPLEX_KERNEL)) || (defined(WITH_FIXED_COMPLEX_KERNEL) && !defined(WITH_COMPLEX_VSX_BLOCK2_KERNEL)) */
 
@@ -973,33 +973,33 @@
              do j = ncols, 2, -2
                w(:,1) = bcast_buffer(1:nbw,j+off)
                w(:,2) = bcast_buffer(1:nbw,j+off-1)
-#ifdef WITH_OPENMP
-               call double_hh_trafo_&
-                    &MATH_DATATYPE&
-                    &_vsx_2hv_&
-                    &PRECISION&
-                    & (c_loc(a(1,j+off+a_off-1,istripe,my_thread)), w, nbw, nl, stripe_width, nbw)
-#else
-               call double_hh_trafo_&
-                    &MATH_DATATYPE&
-                    &_vsx_2hv_&
-                    &PRECISION&
-                    & (c_loc(a(1,j+off+a_off-1,istripe)), w, nbw, nl, stripe_width, nbw)
-#endif
+!#ifdef WITH_OPENMP
+!               call double_hh_trafo_&
+!                    &MATH_DATATYPE&
+!                    &_vsx_2hv_&
+!                    &PRECISION&
+!                    & (c_loc(a(1,j+off+a_off-1,istripe,my_thread)), w, nbw, nl, stripe_width, nbw)
+!#else
+!               call double_hh_trafo_&
+!                    &MATH_DATATYPE&
+!                    &_vsx_2hv_&
+!                    &PRECISION&
+!                    & (c_loc(a(1,j+off+a_off-1,istripe)), w, nbw, nl, stripe_width, nbw)
+!#endif
              enddo
-#ifdef WITH_OPENMP
-             if (j==1) call single_hh_trafo_&
-                 &MATH_DATATYPE&
-                       &_vsx_1hv_&
-                       &PRECISION&
-                       & (c_loc(a(1,1+off+a_off,istripe,my_thread)), bcast_buffer(1,off+1), nbw, nl, stripe_width)
-#else
-             if (j==1) call single_hh_trafo_&
-                 &MATH_DATATYPE&
-                            &_vsx_1hv_&
-                            &PRECISION&
-                            & (c_loc(a(1,1+off+a_off,istripe)), bcast_buffer(1,off+1), nbw, nl, stripe_width)
-#endif
+!#ifdef WITH_OPENMP
+!             if (j==1) call single_hh_trafo_&
+!                 &MATH_DATATYPE&
+!                       &_vsx_1hv_&
+!                       &PRECISION&
+!                       & (c_loc(a(1,1+off+a_off,istripe,my_thread)), bcast_buffer(1,off+1), nbw, nl, stripe_width)
+!#else
+!             if (j==1) call single_hh_trafo_&
+!                 &MATH_DATATYPE&
+!                            &_vsx_1hv_&
+!                            &PRECISION&
+!                            & (c_loc(a(1,1+off+a_off,istripe)), bcast_buffer(1,off+1), nbw, nl, stripe_width)
+!#endif
 
 #ifndef WITH_FIXED_COMPLEX_KERNEL
            endif ! (kernel .eq. ELPA_2STAGE_COMPLEX_VSX_BLOCK2)

src/elpa2/kernels/real_vsx_2hv_template.c

@@ -823,7 +823,7 @@ void double_hh_trafo_real_vsx_2hv_single(float* q, float* hh, int* pnb, int* pnq
 	_SSE_STORE((__vector unsigned int) q2, 0, (unsigned int *)  &q[offset]);
 	q3 = _SSE_LOAD(0, (unsigned long int *)  &q[2*offset]);
 	q3 = _SSE_ADD(q3, y3);
-	_SSE_STORE((__vector unsigned int) q3,0, (unsigned int *)  &q[2*offset]);
+	_SSE_STORE((__vector unsigned int) q3, 0, (unsigned int *)  &q[2*offset]);
 	q4 = _SSE_LOAD(0, (unsigned long int *)  &q[3*offset]);
 	q4 = _SSE_ADD(q4, y4);
 	_SSE_STORE((__vector unsigned int) q4, 0, (unsigned int *)  &q[3*offset]);

src/elpa2/kernels/real_vsx_6hv_double_precision.c

+//    This file is part of ELPA.
+//
+//    The ELPA library was originally created by the ELPA consortium,
+//    consisting of the following organizations:
+//
+//    - Max Planck Computing and Data Facility (MPCDF), formerly known as
+//      Rechenzentrum Garching der Max-Planck-Gesellschaft (RZG),
+//    - Bergische Universität Wuppertal, Lehrstuhl für angewandte
+//      Informatik,
+//    - Technische Universität München, Lehrstuhl für Informatik mit
+//      Schwerpunkt Wissenschaftliches Rechnen ,
+//    - Fritz-Haber-Institut, Berlin, Abt. Theorie,
+//    - Max-Plack-Institut für Mathematik in den Naturwissenschaften,
+//      Leipzig, Abt. Komplexe Strukutren in Biologie und Kognition,
+//      and
+//    - IBM Deutschland GmbH
+//
+//    This particular source code file contains additions, changes and
+//    enhancements authored by Intel Corporation which is not part of
+//    the ELPA consortium.
+//
+//    More information can be found here:
+//    http://elpa.mpcdf.mpg.de/
+//
+//    ELPA is free software: you can redistribute it and/or modify
+//    it under the terms of the version 3 of the license of the
+//    GNU Lesser General Public License as published by the Free
+//    Software Foundation.
+//
+//    ELPA is distributed in the hope that it will be useful,
+//    but WITHOUT ANY WARRANTY; without even the implied warranty of
+//    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+//    GNU Lesser General Public License for more details.
+//
+//    You should have received a copy of the GNU Lesser General Public License
+//    along with ELPA.  If not, see <http://www.gnu.org/licenses/>
+//
+//    ELPA reflects a substantial effort on the part of the original
+//    ELPA consortium, and we ask you to respect the spirit of the
+//    license that we chose: i.e., please contribute any changes you
+//    may have back to the original ELPA library distribution, and keep
+//    any derivatives of ELPA under the same license that we chose for
+//    the original distribution, the GNU Lesser General Public License.
+//
+// Author: Andreas Marek, MPCDF
+
+#include "config-f90.h"
+
+#define REALCASE 1
+#define DOUBLE_PRECISION 1
+#include "../../general/precision_macros.h"
+#include "real_vsx_6hv_template.c"
+#undef REALCASE
+#undef DOUBLE_PRECISION
+

src/elpa2/kernels/real_vsx_6hv_single_precision.c

+//    This file is part of ELPA.
+//
+//    The ELPA library was originally created by the ELPA consortium,
+//    consisting of the following organizations:
+//
+//    - Max Planck Computing and Data Facility (MPCDF), formerly known as
+//      Rechenzentrum Garching der Max-Planck-Gesellschaft (RZG),
+//    - Bergische Universität Wuppertal, Lehrstuhl für angewandte
+//      Informatik,
+//    - Technische Universität München, Lehrstuhl für Informatik mit
+//      Schwerpunkt Wissenschaftliches Rechnen ,
+//    - Fritz-Haber-Institut, Berlin, Abt. Theorie,
+//    - Max-Plack-Institut für Mathematik in den Naturwissenschaften,
+//      Leipzig, Abt. Komplexe Strukutren in Biologie und Kognition,
+//      and
+//    - IBM Deutschland GmbH
+//
+//    This particular source code file contains additions, changes and
+//    enhancements authored by Intel Corporation which is not part of
+//    the ELPA consortium.
+//
+//    More information can be found here:
+//    http://elpa.mpcdf.mpg.de/
+//
+//    ELPA is free software: you can redistribute it and/or modify
+//    it under the terms of the version 3 of the license of the
+//    GNU Lesser General Public License as published by the Free
+//    Software Foundation.
+//
+//    ELPA is distributed in the hope that it will be useful,
+//    but WITHOUT ANY WARRANTY; without even the implied warranty of
+//    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+//    GNU Lesser General Public License for more details.
+//
+//    You should have received a copy of the GNU Lesser General Public License
+//    along with ELPA.  If not, see <http://www.gnu.org/licenses/>
+//
+//    ELPA reflects a substantial effort on the part of the original
+//    ELPA consortium, and we ask you to respect the spirit of the
+//    license that we chose: i.e., please contribute any changes you
+//    may have back to the original ELPA library distribution, and keep
+//    any derivatives of ELPA under the same license that we chose for
+//    the original distribution, the GNU Lesser General Public License.
+//
+// Author: Andreas Marek, MPCDF
+
+#include "config-f90.h"
+
+#define REALCASE 1
+#define SINGLE_PRECISION 1
+#include "../../general/precision_macros.h"
+#include "real_vsx_6hv_template.c"
+#undef REALCASE
+#undef SINGLE_PRECISION
+