Activate simple block4 block6 kernels

Makefile.am

@@ -110,6 +110,7 @@ EXTRA_libelpa@SUFFIX@_private_la_DEPENDENCIES = \
   src/elpa2/kernels/complex_template.F90 \
   src/elpa2/kernels/simple_template.F90 \
   src/elpa2/kernels/simple_block4_template.F90 \
+  src/elpa2/kernels/simple_block6_template.F90 \
   src/elpa2/pack_unpack_cpu.F90 \
   src/elpa2/pack_unpack_gpu.F90 \
   src/elpa2/compute_hh_trafo.F90 \
@@ -194,9 +195,10 @@ if WITH_REAL_GENERIC_SIMPLE_BLOCK4_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block4.F90
 endif
 
-#if WITH_REAL_GENERIC_SIMPLE_BLOCK6_KERNEL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block6.F90
-#endif
+if WITH_REAL_GENERIC_SIMPLE_BLOCK6_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block6.F90
+endif
+
 if WITH_REAL_BGP_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_bgp.f90
 endif
@@ -802,6 +804,7 @@ EXTRA_DIST = \
   src/elpa2/kernels/real_template.F90 \
   src/elpa2/kernels/simple_template.F90 \
   src/elpa2/kernels/simple_block4_template.F90 \
+  src/elpa2/kernels/simple_block6_template.F90 \
   src/elpa2/pack_unpack_cpu.F90 \
   src/elpa2/pack_unpack_gpu.F90 \
   src/elpa2/qr/elpa_pdgeqrf_template.F90 \

ci_test_scripts/run_ci_tests.sh

@@ -150,10 +150,10 @@ then
     echo " "
     echo "Exit Code of sbatch: $exitCode"
     echo " "
-    if (( $exitCode > 0 ))
-    then
+    #if (( $exitCode > 0 ))
+    #then
       cat ./ELPA_CI_2gpu.err.*
-    fi
+    #fi
     
   fi
 
@@ -181,17 +181,17 @@ then
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
-    fi
+    #fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
-  then
+  #if [ $exitCode -ne 0 ]
+  #then
     cat ./test-suite.log
-  fi
+  #fi
 
   exit $exitCode
 

ci_test_scripts/run_distcheck_tests.sh

@@ -126,17 +126,17 @@ then
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
-    fi
+    #fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
-  then
+  #if [ $exitCode -ne 0 ]
+  #then
     cat ./test-suite.log
-  fi
+  #fi
 
   exit $exitCode
 

ci_test_scripts/run_project_tests.sh

@@ -129,18 +129,18 @@ then
     echo "mkdir -p build" >> ./run_${CLUSTER}_1node.sh
     echo "pushd build"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running autogen " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running autogen " >> ./run_${CLUSTER}_1node.sh
     echo "../autogen.sh"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running configure " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running configure " >> ./run_${CLUSTER}_1node.sh
     echo "../configure " "$configureArgs" >> ./run_${CLUSTER}_1node.sh
     echo " " >> ./run_${CLUSTER}_1node.sh
     echo "export TASKS=$mpiTasks" >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make " >> ./run_${CLUSTER}_1node.sh
     echo "make -j 8"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make install" >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make install" >> ./run_${CLUSTER}_1node.sh
     echo "make install" >> ./run_${CLUSTER}_1node.sh
     echo "popd" >> ./run_${CLUSTER}_1node.sh
     echo "mkdir -p $projectName/build" >> ./run_${CLUSTER}_1node.sh
@@ -149,19 +149,19 @@ then
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo " Testting project "  >> ./run_${CLUSTER}_1node.sh
+    echo " #Testting project "  >> ./run_${CLUSTER}_1node.sh
 
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
 
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running autogen " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running autogen " >> ./run_${CLUSTER}_1node.sh
     echo "../autogen.sh" >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running configure " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running configure " >> ./run_${CLUSTER}_1node.sh
     echo "../configure " "$projectConfigureArgs " >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make " >> ./run_${CLUSTER}_1node.sh
     echo "make -j 8" >> ./run_${CLUSTER}_1node.sh
     echo "export LD_LIBRARY_PATH=$MKL_HOME/lib/intel64:\$LD_LIBRARY_PATH" >> ./run_${CLUSTER}_1node.sh
     echo "./$projectExecutable" >> ./run_${CLUSTER}_1node.sh
@@ -184,18 +184,13 @@ then
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
-    fi
+    #fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
-  then
-    cat ./test-suite.log
-  fi
-
   exit $exitCode
 
 fi

configure.ac

@@ -50,13 +50,15 @@ if test x$_cv_gnu_make_command = x ; then
         AC_MSG_ERROR([Need GNU Make])
 fi
 
+enable_legacy=no
+
 AC_MSG_CHECKING(whether legacy interface should be provided)
 AC_ARG_ENABLE([legacy-interface],
               AS_HELP_STRING([--enable-legacy-interface],
                              [build legacy API, default no]),
               [
 	       if test x"$enableval" = x"yes"; then
-	         enable_legayc=yes
+	         enable_legacy=yes
 	       else
 	         enable_legacy=no
 	       fi
@@ -613,6 +615,7 @@ m4_define(elpa_m4_generic_kernels, [
         real_generic
         real_generic_simple
         real_generic_simple_block4
+        real_generic_simple_block6
         complex_generic
         complex_generic_simple
 ])

elpa/elpa_constants.h.in

@@ -50,7 +50,8 @@ enum ELPA_SOLVERS {
         X(ELPA_2STAGE_REAL_VSX_BLOCK2, 25, @ELPA_2STAGE_REAL_VSX_BLOCK2_COMPILED@, __VA_ARGS__) \
         X(ELPA_2STAGE_REAL_VSX_BLOCK4, 26, @ELPA_2STAGE_REAL_VSX_BLOCK4_COMPILED@, __VA_ARGS__) \
         X(ELPA_2STAGE_REAL_VSX_BLOCK6, 27, @ELPA_2STAGE_REAL_VSX_BLOCK6_COMPILED@, __VA_ARGS__) \
-        X(ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK4, 28, @ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK4_COMPILED@, __VA_ARGS__)
+        X(ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK4, 28, @ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK4_COMPILED@, __VA_ARGS__) \
+        X(ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK6, 29, @ELPA_2STAGE_REAL_GENERIC_SIMPLE_BLOCK6_COMPILED@, __VA_ARGS__)
 
 #define ELPA_FOR_ALL_2STAGE_REAL_KERNELS_AND_DEFAULT(X) \
         ELPA_FOR_ALL_2STAGE_REAL_KERNELS(X) \

src/elpa2/legacy_interface/elpa2_utilities.F90

@@ -56,7 +56,7 @@ module elpa2_utilities
   implicit none
   public
 
-  integer(kind=c_int), parameter :: number_of_real_kernels = ELPA_2STAGE_NUMBER_OF_REAL_KERNELS - 7
+  integer(kind=c_int), parameter :: number_of_real_kernels = ELPA_2STAGE_NUMBER_OF_REAL_KERNELS - 11
   integer(kind=c_int), parameter :: number_of_complex_kernels = ELPA_2STAGE_NUMBER_OF_COMPLEX_KERNELS
 
 #ifdef WITH_REAL_GENERIC_KERNEL

src/elpa_abstract_impl.F90

@@ -50,7 +50,6 @@
 module elpa_abstract_impl
   use elpa_api
   use elpa_generated_fortran_interfaces
-  use elpa_utilities, only : error_unit
 
 #ifdef HAVE_DETAILED_TIMINGS
   use ftimings
@@ -98,6 +97,7 @@ module elpa_abstract_impl
     !> \result  error      integer, the error code
     subroutine elpa_set_integer(self, name, value, error)
       use iso_c_binding
+      use elpa_utilities, only : error_unit
       class(elpa_abstract_impl_t)     :: self
       character(*), intent(in)        :: name
       integer(kind=c_int), intent(in) :: value
@@ -131,6 +131,7 @@ module elpa_abstract_impl
     !> \param   error      integer, optional, to store an error code
     subroutine elpa_get_integer(self, name, value, error)
       use iso_c_binding
+      use elpa_utilities, only : error_unit
       class(elpa_abstract_impl_t)    :: self
       character(*), intent(in)       :: name
       integer(kind=c_int)            :: value
@@ -163,6 +164,7 @@ module elpa_abstract_impl
     !> \result  error      integer, the error code
     subroutine elpa_set_double(self, name, value, error)
       use iso_c_binding
+      use elpa_utilities, only : error_unit
       class(elpa_abstract_impl_t)     :: self
       character(*), intent(in)        :: name
       real(kind=c_double), intent(in) :: value
@@ -195,6 +197,7 @@ module elpa_abstract_impl
     !> \param   error      integer, optional, to store an error code
     subroutine elpa_get_double(self, name, value, error)
       use iso_c_binding
+      use elpa_utilities, only : error_unit
       class(elpa_abstract_impl_t)    :: self
       character(*), intent(in)       :: name
       real(kind=c_double)            :: value

test_project_1stage/Makefile.am

@@ -2,7 +2,7 @@
 
 ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} -I m4
 
-AM_FCFLAGS = @FC_MODINC@modules @FC_MODOUT@modules $(ELPA_FCFLAGS)
+AM_FCFLAGS = @FC_MODINC@modules $(ELPA_FCFLAGS)
 AM_LDFLAGS = $(ELPA_LIBS)
 
 #bindir = $(abs_top_builddir)

test_project_1stage/src/test_real.F90

@@ -111,7 +111,7 @@ program test_real_example
    character(len=8)                 :: task_suffix
    integer                          :: j
 
-   integer, parameter               :: error_unit = 0
+   integer, parameter               :: error_units = 0
 
    class(elpa_t), pointer           :: e
    !-------------------------------------------------------------------------------
@@ -157,12 +157,12 @@ program test_real_example
    call descinit(sc_desc, na, na, nblk, nblk, 0, 0, my_blacs_ctxt, na_rows, info)
 
    if (info .ne. 0) then
-     write(error_unit,*) 'Error in BLACS descinit! info=',info
-     write(error_unit,*) 'Most likely this happend since you want to use'
-     write(error_unit,*) 'more MPI tasks than are possible for your'
-     write(error_unit,*) 'problem size (matrix size and blocksize)!'
-     write(error_unit,*) 'The blacsgrid can not be set up properly'
-     write(error_unit,*) 'Try reducing the number of MPI tasks...'
+     write(error_units,*) 'Error in BLACS descinit! info=',info
+     write(error_units,*) 'Most likely this happend since you want to use'
+     write(error_units,*) 'more MPI tasks than are possible for your'
+     write(error_units,*) 'problem size (matrix size and blocksize)!'
+     write(error_units,*) 'The blacsgrid can not be set up properly'
+     write(error_units,*) 'Try reducing the number of MPI tasks...'
      call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 

test_project_1stage_legacy_api/Makefile.am

@@ -2,7 +2,7 @@
 
 ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} -I m4
 
-AM_FCFLAGS = @FC_MODINC@modules @FC_MODOUT@modules $(ELPA_FCFLAGS)
+AM_FCFLAGS = @FC_MODINC@modules $(ELPA_FCFLAGS)
 AM_LDFLAGS = $(ELPA_LIBS)
 
 #bindir = $(abs_top_builddir)

test_project_1stage_legacy_api/src/test_real.F90

@@ -110,7 +110,7 @@ program test_real_example
    character(len=8)           :: task_suffix
    integer           :: j
 
-   integer, parameter :: error_unit = 0
+   integer, parameter :: error_units = 0
    !-------------------------------------------------------------------------------
 
 
@@ -170,12 +170,12 @@ program test_real_example
    call descinit(sc_desc, na, na, nblk, nblk, 0, 0, my_blacs_ctxt, na_rows, info)
 
    if (info .ne. 0) then
-     write(error_unit,*) 'Error in BLACS descinit! info=',info
-     write(error_unit,*) 'Most likely this happend since you want to use'
-     write(error_unit,*) 'more MPI tasks than are possible for your'
-     write(error_unit,*) 'problem size (matrix size and blocksize)!'
-     write(error_unit,*) 'The blacsgrid can not be set up properly'
-     write(error_unit,*) 'Try reducing the number of MPI tasks...'
+     write(error_units,*) 'Error in BLACS descinit! info=',info
+     write(error_units,*) 'Most likely this happend since you want to use'
+     write(error_units,*) 'more MPI tasks than are possible for your'
+     write(error_units,*) 'problem size (matrix size and blocksize)!'
+     write(error_units,*) 'The blacsgrid can not be set up properly'
+     write(error_units,*) 'Try reducing the number of MPI tasks...'
      call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 
@@ -215,7 +215,7 @@ program test_real_example
                             na_cols, mpi_comm_rows, mpi_comm_cols, mpi_comm_world)
 
    if (.not.(success)) then
-      write(error_unit,*) "elpa_solve_evp_real_1stage_double produced an error! Aborting..."
+      write(error_units,*) "elpa_solve_evp_real_1stage_double produced an error! Aborting..."
       call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 

test_project_2stage/Makefile.am

@@ -2,7 +2,7 @@
 
 ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} -I m4
 
-AM_FCFLAGS = @FC_MODINC@modules @FC_MODOUT@modules $(ELPA_FCFLAGS)
+AM_FCFLAGS = @FC_MODINC@modules $(ELPA_FCFLAGS)
 AM_LDFLAGS = $(ELPA_LIBS)
 
 #bindir = $(abs_top_builddir)

test_project_2stage/src/test_real2.F90

@@ -111,7 +111,7 @@ program test_real_example
    character(len=8)                 :: task_suffix
    integer                          :: j
 
-   integer, parameter               :: error_unit = 0
+   integer, parameter               :: error_units = 0
 
    class(elpa_t), pointer           :: e
    !-------------------------------------------------------------------------------
@@ -157,12 +157,12 @@ program test_real_example
    call descinit(sc_desc, na, na, nblk, nblk, 0, 0, my_blacs_ctxt, na_rows, info)
 
    if (info .ne. 0) then
-     write(error_unit,*) 'Error in BLACS descinit! info=',info
-     write(error_unit,*) 'Most likely this happend since you want to use'
-     write(error_unit,*) 'more MPI tasks than are possible for your'
-     write(error_unit,*) 'problem size (matrix size and blocksize)!'
-     write(error_unit,*) 'The blacsgrid can not be set up properly'
-     write(error_unit,*) 'Try reducing the number of MPI tasks...'
+     write(error_units,*) 'Error in BLACS descinit! info=',info
+     write(error_units,*) 'Most likely this happend since you want to use'
+     write(error_units,*) 'more MPI tasks than are possible for your'
+     write(error_units,*) 'problem size (matrix size and blocksize)!'
+     write(error_units,*) 'The blacsgrid can not be set up properly'
+     write(error_units,*) 'Try reducing the number of MPI tasks...'
      call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 

test_project_2stage_legacy_api/Makefile.am

@@ -2,7 +2,7 @@
 
 ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} -I m4
 
-AM_FCFLAGS = @FC_MODINC@modules @FC_MODOUT@modules $(ELPA_FCFLAGS)
+AM_FCFLAGS = @FC_MODINC@modules $(ELPA_FCFLAGS)
 AM_LDFLAGS = $(ELPA_LIBS)
 
 #bindir = $(abs_top_builddir)

test_project_2stage_legacy_api/src/test_real2.F90

@@ -76,7 +76,7 @@ program test_real_example
    use iso_c_binding
    use elpa1
    use elpa2
-   !use elpa_utilities, only : error_unit
+   !use elpa_utilities, only : error_units
 #ifdef HAVE_MPI_MODULE
    use mpi
    implicit none
@@ -111,7 +111,7 @@ program test_real_example
    character(len=8)           :: task_suffix
    integer           :: j
 
-   integer, parameter :: error_unit = 0
+   integer, parameter :: error_units = 0
    !-------------------------------------------------------------------------------
 
 
@@ -171,12 +171,12 @@ program test_real_example
    call descinit(sc_desc, na, na, nblk, nblk, 0, 0, my_blacs_ctxt, na_rows, info)
 
    if (info .ne. 0) then
-     write(error_unit,*) 'Error in BLACS descinit! info=',info
-     write(error_unit,*) 'Most likely this happend since you want to use'
-     write(error_unit,*) 'more MPI tasks than are possible for your'
-     write(error_unit,*) 'problem size (matrix size and blocksize)!'
-     write(error_unit,*) 'The blacsgrid can not be set up properly'
-     write(error_unit,*) 'Try reducing the number of MPI tasks...'
+     write(error_units,*) 'Error in BLACS descinit! info=',info
+     write(error_units,*) 'Most likely this happend since you want to use'
+     write(error_units,*) 'more MPI tasks than are possible for your'
+     write(error_units,*) 'problem size (matrix size and blocksize)!'
+     write(error_units,*) 'The blacsgrid can not be set up properly'
+     write(error_units,*) 'Try reducing the number of MPI tasks...'
      call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 
@@ -216,7 +216,7 @@ program test_real_example
                             na_cols, mpi_comm_rows, mpi_comm_cols, mpi_comm_world)
 
    if (.not.(success)) then
-      write(error_unit,*) "elpa_solve_evp_real_2stage_double produced an error! Aborting..."
+      write(error_units,*) "elpa_solve_evp_real_2stage_double produced an error! Aborting..."
       call MPI_ABORT(mpi_comm_world, 1, mpierr)
    endif
 

test_project_C/Makefile.am

@@ -2,7 +2,7 @@
 
 ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} -I m4
 
-AM_FCFLAGS = @FC_MODINC@modules @FC_MODOUT@modules $(ELPA_FCFLAGS)
+AM_FCFLAGS = @FC_MODINC@modules $(ELPA_FCFLAGS)
 AM_CFLAGS = $(ELPA_CFLAGS)
 AM_LDFLAGS = $(ELPA_LIBS)
 

test_project_C/src/test_blacs_infrastructure.F90

@@ -92,7 +92,7 @@ module test_blacs_infrastructure
                                        np_rows, np_cols, na_rows,  &