DAMASK_EICMD/code/prec.f90

206 lines
9.9 KiB
Fortran

!--------------------------------------------------------------------------------------------------
!> @author Franz Roters, Max-Planck-Institut für Eisenforschung GmbH
!> @author Philip Eisenlohr, Max-Planck-Institut für Eisenforschung GmbH
!> @author Christoph Kords, Max-Planck-Institut für Eisenforschung GmbH
!> @author Martin Diehl, Max-Planck-Institut für Eisenforschung GmbH
!> @author Luv Sharma, Max-Planck-Institut für Eisenforschung GmbH
!> @brief setting precision for real and int type
!> @details setting precision for real and int type and for DAMASK_NaN. Definition is made
!! depending on makro "INT" defined during compilation
!! for details on NaN see https://software.intel.com/en-us/forums/topic/294680
!--------------------------------------------------------------------------------------------------
module prec
#if !(defined(__GFORTRAN__) && __GNUC__ < 5)
use, intrinsic :: & ! unfortunately not avialable in gfortran <= 5
IEEE_arithmetic
#endif
implicit none
private
#if (FLOAT==8)
integer, parameter, public :: pReal = 8 !< floating point double precision (was selected_real_kind(15,300), number with 15 significant digits, up to 1e+-300)
#ifdef __INTEL_COMPILER
real(pReal), parameter, public :: DAMASK_NaN = Z'7FF8000000000000' !< quiet NaN for double precision (from http://www.hpc.unimelb.edu.au/doc/f90lrm/dfum_035.html, copy can be found in documentation/Code/Fortran)
#endif
#ifdef __GFORTRAN__
real(pReal), parameter, public :: DAMASK_NaN = real(Z'7FF8000000000000',pReal) !< quiet NaN for double precision (from http://www.hpc.unimelb.edu.au/doc/f90lrm/dfum_035.html, copy can be found in documentation/Code/Fortran)
#endif
#else
NO SUITABLE PRECISION FOR REAL SELECTED, STOPPING COMPILATION
#endif
#if (INT==4)
integer, parameter, public :: pInt = 4 !< integer representation 32 bit (was selected_int_kind(9), number with at least up to +- 1e9)
#elif (INT==8)
integer, parameter, public :: pInt = 8 !< integer representation 64 bit (was selected_int_kind(12), number with at least up to +- 1e12)
#else
NO SUITABLE PRECISION FOR INTEGER SELECTED, STOPPING COMPILATION
#endif
integer, parameter, public :: pLongInt = 8 !< integer representation 64 bit (was selected_int_kind(12), number with at least up to +- 1e12)
real(pReal), parameter, public :: tol_math_check = 1.0e-8_pReal !< tolerance for internal math self-checks (rotation)
integer(pInt), allocatable, dimension(:) :: realloc_lhs_test
type, public :: p_vec !< variable length datatype used for storage of state
real(pReal), dimension(:), pointer :: p
end type p_vec
type, public :: p_intvec
integer(pInt), dimension(:), pointer :: p
end type p_intvec
!http://stackoverflow.com/questions/3948210/can-i-have-a-pointer-to-an-item-in-an-allocatable-array
type, public :: tState
integer(pInt) :: &
sizeState = 0_pInt , & !< size of state
sizeDotState = 0_pInt, & !< size of dot state, i.e. parts of the state that are integrated
sizeDeltaState = 0_pInt, & !< size of delta state, i.e. parts of the state that have discontinuous rates
sizePostResults = 0_pInt !< size of output data
real(pReal), pointer, dimension(:), contiguous :: &
atolState
real(pReal), pointer, dimension(:,:), contiguous :: & ! a pointer is needed here because we might point to state/doState. However, they will never point to something, but are rather allocated and, hence, contiguous
state, & !< state
dotState, & !< state rate
state0
real(pReal), allocatable, dimension(:,:) :: &
partionedState0, &
subState0, &
state_backup, &
deltaState, &
previousDotState, & !< state rate of previous xxxx
previousDotState2, & !< state rate two xxxx ago
dotState_backup, & !< backup of state rate
RK4dotState
real(pReal), allocatable, dimension(:,:,:) :: &
RKCK45dotState
end type
type, extends(tState), public :: tPlasticState
integer(pInt) :: &
nSlip = 0_pInt , &
nTwin = 0_pInt, &
nTrans = 0_pInt
logical :: &
nonlocal = .false. !< absolute tolerance for state integration
real(pReal), pointer, dimension(:,:), contiguous :: &
slipRate, & !< slip rate
accumulatedSlip !< accumulated plastic slip
end type
type, public :: tSourceState
type(tState), dimension(:), allocatable :: p !< tState for each active source mechanism in a phase
end type
type, public :: tHomogMapping
integer(pInt), pointer, dimension(:,:) :: p
end type
type, public :: tPhaseMapping
integer(pInt), pointer, dimension(:,:,:) :: p
end type
#ifdef FEM
type, public :: tOutputData
integer(pInt) :: &
sizeIpCells = 0_pInt , &
sizeResults = 0_pInt
real(pReal), allocatable, dimension(:,:) :: &
output !< output data
end type
#endif
public :: &
prec_init, &
prec_isNaN
contains
!--------------------------------------------------------------------------------------------------
!> @brief reporting precision and checking if DAMASK_NaN is set correctly
!--------------------------------------------------------------------------------------------------
subroutine prec_init
use, intrinsic :: &
iso_fortran_env ! to get compiler_version and compiler_options (at least for gfortran 4.6 at the moment)
implicit none
integer(pInt) :: worldrank = 0_pInt
#ifdef PETSc
#include <petsc/finclude/petscsys.h>
PetscErrorCode :: ierr
#endif
external :: &
quit, &
MPI_Comm_rank, &
MPI_Abort
#ifdef PETSc
call MPI_Comm_rank(PETSC_COMM_WORLD,worldrank,ierr);CHKERRQ(ierr)
#endif
mainProcess: if (worldrank == 0) then
write(6,'(/,a)') ' <<<+- prec init -+>>>'
#include "compilation_info.f90"
write(6,'(a,i3)') ' Bytes for pReal: ',pReal
write(6,'(a,i3)') ' Bytes for pInt: ',pInt
write(6,'(a,i3)') ' Bytes for pLongInt: ',pLongInt
write(6,'(a,e10.3)') ' NaN: ', DAMASK_NaN
write(6,'(a,l3)') ' NaN != NaN: ',DAMASK_NaN /= DAMASK_NaN
write(6,'(a,l3,/)') ' NaN check passed ',prec_isNAN(DAMASK_NaN)
endif mainProcess
if ((.not. prec_isNaN(DAMASK_NaN)) .or. (DAMASK_NaN == DAMASK_NaN)) call quit(9000)
realloc_lhs_test = [1_pInt,2_pInt]
if (realloc_lhs_test(2)/=2_pInt) call quit(9000)
end subroutine prec_init
!--------------------------------------------------------------------------------------------------
!> @brief figures out if a floating point number is NaN
! basically just a small wrapper, because gfortran < 5.0 does not have the IEEE module
!--------------------------------------------------------------------------------------------------
logical elemental pure function prec_isNaN(a)
implicit none
real(pReal), intent(in) :: a
#if (defined(__GFORTRAN__) && __GNUC__ < 5)
intrinsic :: isNaN
prec_isNaN = isNaN(a)
#else
prec_isNaN = IEEE_is_NaN(a)
#endif
end function prec_isNaN
!--------------------------------------------------------------------------------------------------
!> @brief equality comparison for double precision
! replaces "==" but for certain (relative) tolerance. Counterpart to dNeq
! http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm
!--------------------------------------------------------------------------------------------------
logical elemental pure function dEq(a,b,tol)
real(pReal), intent(in) :: a,b
real(pReal), intent(in), optional :: tol
real(pReal), parameter :: eps = 2.2204460492503131E-16 ! DBL_EPSILON in C
dEq = merge(.True., .False.,abs(a-b) <= merge(tol,eps,present(tol))*maxval(abs([a,b])))
end function dEq
!--------------------------------------------------------------------------------------------------
!> @brief inequality comparison for double precision
! replaces "!=" but for certain (relative) tolerance. Counterpart to dEq
! http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm
!--------------------------------------------------------------------------------------------------
logical elemental pure function dNeq(a,b,tol)
real(pReal), intent(in) :: a,b
real(pReal), intent(in), optional :: tol
real(pReal), parameter :: eps = 2.2204460492503131E-16 ! DBL_EPSILON in C
dNeq = merge(.False., .True.,abs(a-b) <= merge(tol,eps,present(tol))*maxval(abs([a,b])))
end function dNeq
end module prec