Merge branch 'master' into ELPA_GPU

.gitlab-ci.yml

 before_script:
   - export LANG=C
-  - module load impi/5.1.3 intel/16.0 gcc/4.9 mkl/11.3 autotools
+  - module load impi/5.1.3 intel/16.0 gcc/4.9 mkl/11.3 autotools pkg-config
   - module list
   - export MKL_INTEL_SCALAPACK_MPI_NO_OMP_BASELINE="-L$MKL_HOME/lib/intel64 -lmkl_scalapack_lp64 -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lmkl_blacs_intelmpi_lp64 -lpthread -lm"
   - export MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP="$MKL_INTEL_SCALAPACK_MPI_NO_OMP_BASELINE -I$MKL_HOME/include/intel64/lp64"
@@ -1225,3 +1225,25 @@ distcheck:
     - ./autogen.sh
     - ./configure --with-mpi=no
     - make distcheck DISTCHECK_CONFIGURE_FLAGS="--with-mpi=no" TEST_FLAGS='1500 50 16'
+
+test_project:
+  tags:
+    - cpu
+  script:
+    - mkdir build
+    - pushd build
+    - ../autogen.sh
+    - ../configure SCALAPACK_LDFLAGS="$MKL_GFORTRAN_SCALAPACK_LDFLAGS_MPI_NO_OMP" SCALAPACK_FCFLAGS="$MKL_GFORTRAN_SCALAPACK_FCFLAGS_MPI_NO_OMP" FC=mpif90 --prefix=$PWD/installdest
+    - make -j 8
+    - make install
+    - popd
+    - mkdir test_project/build
+    - pushd test_project/build
+    - ../autogen.sh
+    - ../configure PKG_CONFIG_PATH=../../build/installdest/lib64/pkgconfig
+    - make -j 8
+    - ./test_real
+    - make distclean
+    - popd
+    - make distclean
+    - rm -rf installdest

Makefile.am

@@ -501,15 +501,16 @@ TESTS = $(check_SCRIPTS)
 	echo '$(wrapper)./$^ $$TEST_FLAGS' > $@
 	chmod +x $@
 
-# this one does not want any arguments
-elpa2_test_real_default_kernel_qr_decomposition@SUFFIX@.sh:
-	echo '$(wrapper)./elpa2_test_real_default_kernel_qr_decomposition@SUFFIX@' > $@
-	chmod +x $@
+
+## this one does not want any arguments
+#elpa2_test_real_default_kernel_qr_decomposition@SUFFIX@.sh:
+#	echo '$(wrapper)./elpa2_test_real_default_kernel_qr_decomposition@SUFFIX@' > $@
+#	chmod +x $@
 
 if WANT_SINGLE_PRECISION_REAL
-elpa2_test_real_default_kernel_qr_decomposition_single_precision@SUFFIX@.sh:
-	echo '$(wrapper)./elpa2_test_real_default_kernel_qr_decomposition_single_precision@SUFFIX@' > $@
-	chmod +x $@
+#elpa2_test_real_default_kernel_qr_decomposition_single_precision@SUFFIX@.sh:
+#	echo '$(wrapper)./elpa2_test_real_default_kernel_qr_decomposition_single_precision@SUFFIX@' > $@
+#	chmod +x $@
 endif
 # Preprocessed files (just used for manual inspection)
 mod_precision.i: $(top_srcdir)/src/mod_precision.F90

generated_headers.am

 define extract_interface
 	@echo "Generating $@...";
-	@grep -h "^ *$1" $^ | sed 's/^ *$1//;' > $@ || { rm $@; exit 1; }
+	@grep -h "^ *$1" $^ | sed 's/^ *$1//;' >> $@ || { rm $@; exit 1; }
 endef
 
 elpa test:
@@ -19,8 +19,9 @@ elpa/elpa_generated.h: $(top_srcdir)/src/elpa_c_interface.F90 | elpa
 test/shared_sources/generated.h: $(wildcard $(top_srcdir)/test/shared_sources/*.F90) | test/shared_sources
 	$(call extract_interface,!c>)
 
-elpa/elpa_generated_fortran_interfaces.h: $(wildcard $(top_srcdir)/src/elpa2_kernels/*.c) | elpa
+elpa/elpa_generated_fortran_interfaces.h: $(wildcard $(top_srcdir)/src/elpa2_kernels/*.c) $(wildcard $(top_srcdir)/src/elpa2_kernels/*.s) | elpa
 	$(call extract_interface,!f>)
+	$(call extract_interface,#!f>)
 
 generated_headers= config-f90.h elpa/elpa_generated.h test/shared_sources/generated.h elpa/elpa_generated_fortran_interfaces.h
 generated-headers: $(generated_headers)

src/elpa2.F90

@@ -239,7 +239,7 @@ contains
     endif
 
     if (useQRActual) then
-      if (mod(na,nblk) .ne. 0) then
+      if (mod(na,2) .ne. 0) then
         if (wantDebug) then
           write(error_unit,*) "solve_evp_real_2stage: QR-decomposition: blocksize does not fit with matrixsize"
         endif

src/elpa2_kernels/elpa2_kernels_asm_x86_64_double_precision.s

@@ -54,8 +54,8 @@
 # distributed along with the original code in the file "COPYING".
 #
 # --------------------------------------------------------------------------------------------------
-        .globl double_hh_trafo_double_
-        .globl single_hh_trafo_complex_double_
+        .globl double_hh_trafo_double
+        .globl single_hh_trafo_complex_double
         .text
 
 #-------------------------------------------------------------------------------
@@ -382,8 +382,19 @@
 #   parameter 6: %r9  : ldh
 #
 #-------------------------------------------------------------------------------
+
+#!f>#ifdef WITH_REAL_SSE_ASSEMBLY_KERNEL
+#!f>  interface
+#!f>    subroutine double_hh_trafo_double(q, hh, nb, nq, ldq, ldh) bind(C,name="double_hh_trafo_double")
+#!f>      use, intrinsic :: iso_c_binding
+#!f>      integer(kind=c_int)  :: nb, nq, ldq, ldh
+#!f>      type(c_ptr), value   :: q
+#!f>      real(kind=c_double)  :: hh(nb,6)
+#!f>    end subroutine
+#!f>  end interface
+#!f>#endif
         .align    16,0x90
-double_hh_trafo_double_:
+double_hh_trafo_double:
 
         # Get integer parameters into corresponding registers
 
@@ -698,8 +709,18 @@ return1:
 #   parameter 5: %r8  : ldq
 #
 #-------------------------------------------------------------------------------
+#!f>#ifdef WITH_COMPLEX_SSE_ASSEMBLY_KERNEL
+#!f>  interface
+#!f>    subroutine single_hh_trafo_complex_double(q, hh, nb, nq, ldq) bind(C,name="single_hh_trafo_complex_double")
+#!f>      use, intrinsic :: iso_c_binding
+#!f>      integer(kind=c_int)     :: nb, nq, ldq
+#!f>      complex(kind=c_double)  :: q(*)
+#!f>      complex(kind=c_double)  :: hh(nb,2)
+#!f>    end subroutine
+#!f>  end interface
+#!f>#endif
         .align    16,0x90
-single_hh_trafo_complex_double_:
+single_hh_trafo_complex_double:
 
         # Get integer parameters into corresponding registers
 

src/elpa2_kernels/elpa2_kernels_asm_x86_64_single_precision.s

@@ -41,8 +41,8 @@
 #
 # Author: Andreas Marek, MPCDF
 
-        .globl double_hh_trafo_single_
-        .globl single_hh_trafo_complex_single_
+        .globl double_hh_trafo_single
+        .globl single_hh_trafo_complex_single
 
 	.text
 #-------------------------------------------------------------------------------
@@ -391,8 +391,20 @@
 #   parameter 6: %r9  : ldh
 #
 #-------------------------------------------------------------------------------
+#!f>#ifdef WITH_REAL_SSE_ASSEMBLY_KERNEL
+#!f>#ifdef WANT_SINGLE_PRECISION_REAL
+#!f>  interface
+#!f>    subroutine double_hh_trafo_single(q, hh, nb, nq, ldq, ldh) bind(C,name="double_hh_trafo_single")
+#!f>      use, intrinsic :: iso_c_binding
+#!f>      integer(kind=c_int) :: nb, nq, ldq, ldh
+#!f>      type(c_ptr), value  :: q
+#!f>      real(kind=c_float)  :: hh(nb,6)
+#!f>    end subroutine
+#!f>  end interface
+#!f>#endif
+#!f>#endif
         .align    16,0x90
-double_hh_trafo_single_:
+double_hh_trafo_single:
 
         # Get integer parameters into corresponding registers
 
@@ -716,8 +728,21 @@ return1:
 #   parameter 5: %r8  : ldq
 #
 #-------------------------------------------------------------------------------
+#!f>#ifdef WITH_COMPLEX_SSE_ASSEMBLY_KERNEL
+#!f>#ifdef WANT_SINGLE_PRECISION_COMPLEX
+#!f>  interface
+#!f>    subroutine single_hh_trafo_complex_single(q, hh, nb, nq, ldq) bind(C,name="single_hh_trafo_complex_single")
+#!f>      use, intrinsic :: iso_c_binding
+#!f>      integer(kind=c_int)   :: nb, nq, ldq
+#!f>      complex(kind=c_float) :: q(*)
+#!f>      complex(kind=c_float) :: hh(nb,2)
+#!f>    end subroutine
+#!f>  end interface
+#!f>#endif
+#!f>#endif
+
         .align    16,0x90
-single_hh_trafo_complex_single_:
+single_hh_trafo_complex_single:
 
         # Get integer parameters into corresponding registers
 

src/elpa_qr/elpa_pdlarfb.F90

@@ -748,6 +748,7 @@ subroutine qr_tmerge_pdlarfb_1dcomm_double(m,mb,n,oldk,k,v,ldv,t,ldt,a,lda,basei
         else
             ! do not calculate parts for T merge as there is nothing to merge
 
+            mergeoffset  = 0
             updateoffset = 0
 
             tgenoffset = updateoffset + updatesize

src/mod_compute_hh_trafo_complex.F90

@@ -88,7 +88,7 @@ module compute_hh_trafo_complex
 #endif
            use iso_c_binding
 
-#if defined(HAVE_AVX) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
+#if defined(HAVE_AVX) || defined(HAVE_AVX2) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
          use kernel_interfaces
 #endif
            implicit none
@@ -399,7 +399,7 @@ module compute_hh_trafo_complex
            use timings
 #endif
 
-#if defined(HAVE_AVX) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
+#if defined(HAVE_AVX) || defined(HAVE_AVX2) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
            use kernel_interfaces
 #endif
            use iso_c_binding

src/mod_compute_hh_trafo_real.F90

@@ -103,7 +103,7 @@ module compute_hh_trafo_real
          use timings
 #endif
 
-#if defined(HAVE_AVX) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
+#if defined(HAVE_AVX) || defined(HAVE_AVX2) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
          use kernel_interfaces
 #endif
          implicit none
@@ -282,10 +282,10 @@ module compute_hh_trafo_real
                  w(:,1) = bcast_buffer(1:nbw,j+off)
                  w(:,2) = bcast_buffer(1:nbw,j+off-1)
 #ifdef WITH_OPENMP
-                 call double_hh_trafo_double(a(1,j+off+a_off-1,istripe,my_thread), w, nbw, nl, &
+                 call double_hh_trafo_double(c_loc(a(1,j+off+a_off-1,istripe,my_thread)), w, nbw, nl, &
                                       stripe_width, nbw)
 #else
-                 call double_hh_trafo_double(a(1,j+off+a_off-1,istripe), w, nbw, nl, &
+                 call double_hh_trafo_double(c_loc(a(1,j+off+a_off-1,istripe)), w, nbw, nl, &
                                       stripe_width, nbw)
 #endif
                enddo
@@ -682,7 +682,7 @@ module compute_hh_trafo_real
          use timings
 #endif
 
-#if defined(HAVE_AVX) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
+#if defined(HAVE_AVX) || defined(HAVE_AVX2) || defined(HAVE_SSE_INTRINSICS) || defined(HAVE_SSE_ASSEMBLY)
          use kernel_interfaces
 #endif
          implicit none
@@ -860,10 +860,10 @@ module compute_hh_trafo_real
                  w(:,1) = bcast_buffer(1:nbw,j+off)
                  w(:,2) = bcast_buffer(1:nbw,j+off-1)
 #ifdef WITH_OPENMP
-                 call double_hh_trafo_single(a(1,j+off+a_off-1,istripe,my_thread), w, nbw, nl, &
+                 call double_hh_trafo_single(c_loc(a(1,j+off+a_off-1,istripe,my_thread)), w, nbw, nl, &
                                       stripe_width, nbw)
 #else
-                 call double_hh_trafo_single(a(1,j+off+a_off-1,istripe), w, nbw, nl, &
+                 call double_hh_trafo_single(c_loc(a(1,j+off+a_off-1,istripe)), w, nbw, nl, &
                                       stripe_width, nbw)
 #endif
                enddo

src/mod_mpi.F90

@@ -44,7 +44,6 @@
 #include "config-f90.h"
 
 module elpa_mpi
-  use precision
 #ifndef WITH_MPI
   use elpa_mpi_stubs
 #else

test/fortran_test_programs/test_elpa2_real_qr_default_kernel.F90

@@ -142,15 +142,31 @@ program test_real2_default_kernel_qr_decomposition_double_precision
    successELPA   = .true.
    gpuAvailable  = .false.
 
+   !write_to_file = .false.
+   call read_input_parameters(na, nev, nblk, write_to_file)
 
-   if (COMMAND_ARGUMENT_COUNT() /= 0) then
-     write(error_unit,*) "This program does not support any command-line arguments"
-     stop 1
+!   if (COMMAND_ARGUMENT_COUNT() /= 0) then
+!     write(error_unit,*) "This program does not support any command-line arguments"
+!     stop 1
+!   endif
+
+!   nblk = 2
+!   na   = 4000
+!   nev  = 1500
+
+   ! make sure na, nbl is even
+   if (mod(nblk,2 ) .ne. 0) then
+     nblk = nblk - 1
    endif
 
-   nblk = 2
-   na   = 4000
-   nev  = 1500
+   ! make sure na is even
+   if (mod(na,2) .ne. 0) then
+     na = na - 1
+   endif
+   ! make sure na is at least 34
+   if (na .lt. 34) then
+     na = 34
+   endif
 
    !-------------------------------------------------------------------------------
    !  MPI Initialization

test/fortran_test_programs/test_elpa2_real_qr_default_kernel_single.F90

@@ -143,14 +143,32 @@ program test_real2_default_kernel_qr_decomposition_single_precision
    gpuAvailable  = .false.
 
 
-   if (COMMAND_ARGUMENT_COUNT() /= 0) then
-     write(error_unit,*) "This program does not support any command-line arguments"
-     stop 1
+   !write_to_file = .false.
+   call read_input_parameters(na, nev, nblk, write_to_file)
+
+   !if (COMMAND_ARGUMENT_COUNT() /= 0) then
+   !  write(error_unit,*) "This program does not support any command-line arguments"
+   !  stop 1
+   !endif
+
+   ! override nblk
+   !   nblk = 2
+   !   na   = 4000
+   !   nev  = 1500
+
+   ! make sure na, nbl is even
+   if (mod(nblk,2 ) .ne. 0) then
+     nblk = nblk - 1
    endif
 
-   nblk = 2
-   na   = 4000
-   nev  = 1500
+   ! make sure na is even
+   if (mod(na,2) .ne. 0) then
+     na = na - 1
+   endif
+   ! make sure na is at least 34
+   if (na .lt. 34) then
+     na = 34
+   endif
 
    !-------------------------------------------------------------------------------
    !  MPI Initialization

test_project/Makefile.am

@@ -8,3 +8,6 @@ AM_LDFLAGS = $(ELPA_LIBS)
 #bindir = $(abs_top_builddir)
 bin_PROGRAMS = test_real
 test_real_SOURCES = src/test_real.F90
+
+distclean-local:
+	-rm config-f90.h

test_project/configure.ac

 AC_PREREQ([2.69])
-AC_INIT([elpa_test_project],[2014.06.000], elpa-library@rzg.mpg.de)
+AC_INIT([elpa_test_project],[2016.05.001], elpa-library@rzg.mpg.de)
+elpaversion="2016.05.001"
 AC_CONFIG_SRCDIR([src/test_real.F90])
 
 AM_INIT_AUTOMAKE([foreign -Wall subdir-objects])
@@ -67,9 +68,9 @@ AC_ARG_WITH([openmp],
   if test x"${with_openmp}" = x"yes"; then
 	AC_DEFINE([WITH_OPENMP], [1], [use OpenMP threading])
         AX_ELPA_OPENMP
-	elpa="elpa_openmp-2014.06.001"
+	elpa="elpa_openmp-$elpaversion"
   else
-	elpa="elpa-2014.06.001"
+	elpa="elpa-$elpaversion"
   fi
 
 # Here comes the ELPA specific part

test_project/src/test_real.F90

@@ -3,7 +3,8 @@
 !    The ELPA library was originally created by the ELPA consortium,
 !    consisting of the following organizations:
 !
-!    - Rechenzentrum Garching der Max-Planck-Gesellschaft (RZG),
+!    - 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
@@ -16,7 +17,7 @@
 !
 !
 !    More information can be found here:
-!    http://elpa.rzg.mpg.de/
+!    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
@@ -39,7 +40,6 @@
 !    the original distribution, the GNU Lesser General Public License.
 !
 !
-#include "config-f90.h"
 !>
 !> Fortran test programm to demonstrates the use of
 !> ELPA 1 real case library.
@@ -58,7 +58,7 @@
 !> "output", which specifies that the EV's are written to
 !> an ascii file.
 !>
-program test_real
+program test_real_example
 
 !-------------------------------------------------------------------------------
 ! Standard eigenvalue problem - REAL version
@@ -75,25 +75,9 @@ program test_real
 
    use ELPA1
    use elpa_utilities, only : error_unit
-#ifdef WITH_OPENMP
-   use test_util
-#endif
-
-   use mod_read_input_parameters
-   use mod_check_correctness
-   use mod_setup_mpi
-   use mod_blacs_infrastructure
-   use mod_prepare_matrix
-
-#ifdef HAVE_REDIRECT
-  use redirect
-#endif
-#ifdef HAVE_DETAILED_TIMINGS
- use timings
-#endif
-
+   use elpa_mpi
+   use iso_c_binding
    implicit none
-   include 'mpif.h'
 
    !-------------------------------------------------------------------------------
    ! Please set system size parameters below!
@@ -101,172 +85,113 @@ program test_real
    ! nev:  Number of eigenvectors to be calculated
    ! nblk: Blocking factor in block cyclic distribution
    !-------------------------------------------------------------------------------
-   integer :: nblk
-   integer na, nev
-
-
-   !-------------------------------------------------------------------------------
-   !  Local Variables
-
-   integer             :: np_rows, np_cols, na_rows, na_cols
 
-   integer             :: myid, nprocs, my_prow, my_pcol, mpi_comm_rows, mpi_comm_cols
-   integer             :: i, mpierr, my_blacs_ctxt, sc_desc(9), info, nprow, npcol
+   integer, parameter         :: ik = C_INT32_T
+   integer, parameter         :: rk = C_DOUBLE
 
-   integer, external   :: numroc
+   integer(kind=ik)           :: nblk
+   integer(kind=ik)           :: na, nev
 
-   real*8, allocatable :: a(:,:), z(:,:), tmp1(:,:), tmp2(:,:), as(:,:), ev(:)
+   integer(kind=ik)           :: np_rows, np_cols, na_rows, na_cols
 
-   integer             :: iseed(4096) ! Random seed, size should be sufficient for every generator
+   integer(kind=ik)           :: myid, nprocs, my_prow, my_pcol, mpi_comm_rows, mpi_comm_cols
+   integer(kind=ik)           :: i, mpierr, my_blacs_ctxt, sc_desc(9), info, nprow, npcol
 
+   integer, external          :: numroc
 
-   integer             :: STATUS
-#ifdef WITH_OPENMP
-   integer             :: omp_get_max_threads,  required_mpi_thread_level, &
-                          provided_mpi_thread_level
-#endif
-   logical             :: write_to_file
-   logical             :: success
-   !-------------------------------------------------------------------------------
+   real(kind=rk), allocatable :: a(:,:), z(:,:), ev(:)
 
-   success = .true.
+   integer(kind=ik)           :: iseed(4096) ! Random seed, size should be sufficient for every generator
 
-   call read_input_parameters(na, nev, nblk, write_to_file)
+   integer(kind=ik)           :: STATUS
+   logical                    :: success
+   character(len=8)           :: task_suffix
+   integer(kind=ik)           :: j
 
    !-------------------------------------------------------------------------------
-   !  MPI Initialization
-   call setup_mpi(myid, nprocs)
-
-   if (write_to_file) then
-     if (myid .eq. 0) print *,"Writing output files"
-   endif
-
-#ifdef HAVE_DETAILED_TIMINGS
 
-   ! initialise the timing functionality
-
-#ifdef HAVE_LIBPAPI
-   call timer%measure_flops(.true.)
-#endif
-
-   call timer%measure_allocated_memory(.true.)
-   call timer%measure_virtual_memory(.true.)
-   call timer%measure_max_allocated_memory(.true.)
-
-   call timer%set_print_options(&
-#ifdef HAVE_LIBPAPI
-                print_flop_count=.true., &
-                print_flop_rate=.true., &
-#endif
-                print_allocated_memory = .true. , &
-                print_virtual_memory=.true., &
-                print_max_allocated_memory=.true.)
 
+   success = .true.
 
-  call timer%enable()
+   ! default parameters
+   na = 4000
+   nev = 1500
+   nblk = 16
 
-  call timer%start("program")
-#endif
-   !-------------------------------------------------------------------------------
-   ! Selection of number of processor rows/columns
-   ! We try to set up the grid square-like, i.e. start the search for possible
-   ! divisors of nprocs with a number next to the square root of nprocs
-   ! and decrement it until a divisor is found.
-
-
-   STATUS = 0
-#ifdef WITH_OPENMP
-   if (myid .eq. 0) then
-      print *,"Threaded version of test program"
-      print *,"Using ",omp_get_max_threads()," threads"
-      print *," "
-   endif
-#endif
-    call MPI_BARRIER(MPI_COMM_WORLD, mpierr)
-
-#ifdef HAVE_REDIRECT
-   if (check_redirect_environment_variable()) then
-     if (myid .eq. 0) then
-       print *," "
-       print *,"Redirection of mpi processes is used"
-       print *," "
-       if (create_directories() .ne. 1) then
-         write(error_unit,*) "Unable to create directory for stdout and stderr!"
-         stop
-       endif
-      endif
-      call MPI_BARRIER(MPI_COMM_WORLD, mpierr)
-      call redirect_stdout(myid)
-    endif
-#endif
+   call mpi_init(mpierr)
+   call mpi_comm_rank(mpi_comm_world,myid,mpierr)
+   call mpi_comm_size(mpi_comm_world,nprocs,mpierr)
 
    do np_cols = NINT(SQRT(REAL(nprocs))),2,-1
-      if(mod(nprocs,np_cols) == 0 ) exit
+     if(mod(nprocs,np_cols) == 0 ) exit
    enddo
-
    ! at the end of the above loop, nprocs is always divisible by np_cols
 
    np_rows = nprocs/np_cols
 
-   if(myid==0) then
-      print *
-      print '(a)','Standard eigenvalue problem - REAL version'
-      print *
-      print '(3(a,i0))','Matrix size=',na,', Number of eigenvectors=',nev,', Block size=',nblk
-      print '(3(a,i0))','Number of processor rows=',np_rows,', cols=',np_cols,', total=',nprocs
-      print *
+   if (myid==0) then
+     print *
+     print '(a)','Standard eigenvalue problem - REAL version'
+     print *
+     print '(3(a,i0))','Matrix size=',na,', Number of eigenvectors=',nev,', Block size=',nblk
+     print '(3(a,i0))','Number of processor rows=',np_rows,', cols=',np_cols,', total=',nprocs
+     print *
    endif
 
-   !-------------------------------------------------------------------------------
-   ! Set up BLACS context and MPI communicators
-   !
-   ! The BLACS context is only necessary for using Scalapack.
-   !
-   ! For ELPA, the MPI communicators along rows/cols are sufficient,
-   ! and the grid setup may be done in an arbitrary way as long as it is