Add timer calls which corresponds to the old primitive timing of the legacy routines

src/elpa1/elpa1_template.X90

@@ -212,18 +212,16 @@ function elpa_solve_evp_&
      &" // ": error when allocating e, tau "//errorMessage
      stop 1
    endif
-!   ttt0 = MPI_Wtime()
 
+   call obj%timer%start("forward")
    call tridiag_&
    &MATH_DATATYPE&
    &_&
    &PRECISION&
    & (obj, na, a, lda, nblk, matrixCols, mpi_comm_rows, mpi_comm_cols, ev, e, tau, do_useGPU)
-   !ttt1 = MPI_Wtime()
-   !if(my_prow==0 .and. my_pcol==0 .and. summary_timings) write(error_unit,*) 'Time tridiag_real :',ttt1-ttt0
-   !time_evp_fwd = ttt1-ttt0
+   call obj%timer%stop("forward")
 
-   !ttt0 = MPI_Wtime()
+   call obj%timer%start("solve")
    call solve_tridi_&
    &PRECISION&
    & (obj, na, nev, ev, e,  &
@@ -234,19 +232,20 @@ function elpa_solve_evp_&
       q_real, l_rows,  &
 #endif
       nblk, matrixCols, mpi_comm_rows, mpi_comm_cols, wantDebug, success)
+   call obj%timer%stop("solve")
    if (.not.(success)) return
 
 #if COMPLEXCASE == 1
    q(1:l_rows,1:l_cols_nev) = q_real(1:l_rows,1:l_cols_nev)
 #endif
+
+   call obj%timer%start("back")
    call trans_ev_&
    &MATH_DATATYPE&
    &_&
    &PRECISION&
    & (obj, na, nev, a, lda, tau, q, ldq, nblk, matrixCols, mpi_comm_rows, mpi_comm_cols, do_useGPU)
-   !ttt1 = MPI_Wtime()
-   !if(my_prow==0 .and. my_pcol==0 .and. summary_timings) write(error_unit,*) 'Time trans_ev_real:',ttt1-ttt0
-   !time_evp_back = ttt1-ttt0
+   call obj%timer%stop("back")
 #if COMPLEXCASE == 1
    deallocate(q_real, stat=istat, errmsg=errorMessage)
    if (istat .ne. 0) then

src/elpa2/elpa2_template.X90

@@ -286,6 +286,7 @@
       endif
 
       ! Reduction full -> band
+      call obj%timer%start("bandred")
       call bandred_&
       &MATH_DATATYPE&
       &_&
@@ -297,6 +298,7 @@
        , useQRActual &
 #endif
        )
+      call obj%timer%stop("bandred")
       if (.not.(success)) return
     end if  ! matrix not already banded on input
 
@@ -311,6 +313,7 @@
        stop 1
      endif
 
+     call obj%timer%start("tridiag")
      call tridiag_band_&
      &MATH_DATATYPE&
      &_&
@@ -323,6 +326,7 @@
      call mpi_bcast(e, na, MPI_REAL_PRECISION, 0, mpi_comm_all, mpierr)
      call obj%timer%stop("mpi_communication")
 #endif /* WITH_MPI */
+     call obj%timer%stop("tridiag")
 
 #if COMPLEXCASE == 1
      l_rows = local_index(na, my_prow, np_rows, nblk, -1) ! Local rows of a and q
@@ -339,6 +343,7 @@
 #endif
 
      ! Solve tridiagonal system
+     call obj%timer%start("solve")
      call solve_tridi_&
      &PRECISION &
      (obj, na, nev, ev, e, &
@@ -349,7 +354,7 @@
      q_real, ubound(q_real,dim=1), &
 #endif
      nblk, matrixCols, mpi_comm_rows, mpi_comm_cols, wantDebug, success)
-
+     call obj%timer%stop("solve")
      if (.not.(success)) return
 
      deallocate(e, stat=istat, errmsg=errorMessage)
@@ -372,6 +377,8 @@
      endif
 #endif
      ! Backtransform stage 1
+     call obj%timer%start("trans_ev_to_band")
+
      call trans_ev_tridi_to_band_&
      &MATH_DATATYPE&
      &_&
@@ -380,6 +387,7 @@
      q_dev, &
      ldq, matrixCols, hh_trans, mpi_comm_rows, mpi_comm_cols, wantDebug, do_useGPU_trans_ev_tridi, &
      success=success, kernel=kernel)
+     call obj%timer%stop("trans_ev_to_band")
 
      if (.not.(success)) return
 
@@ -402,6 +410,8 @@
        endif
 
        ! Backtransform stage 2
+       call obj%timer%start("trans_ev_to_full")
+
        call trans_ev_band_to_full_&
        &MATH_DATATYPE&
        &_&
@@ -414,6 +424,7 @@
         , useQRActual  &
 #endif
         )
+       call obj%timer%stop("trans_ev_to_full")
 
        deallocate(tmat, stat=istat, errmsg=errorMessage)
        if (istat .ne. 0) then