Chris@42: Previous: Overview of Fortran interface, Up: Overview of Fortran interface [Contents][Index]
Chris@42:If FFTW is compiled in long double (extended) precision
Chris@42: (see Installation and Customization), you may be able to call the
Chris@42: resulting fftwl_ routines (see Precision) from Fortran if
Chris@42: your compiler supports the C_LONG_DOUBLE_COMPLEX type code.
Chris@42:
Because some Fortran compilers do not support
Chris@42: C_LONG_DOUBLE_COMPLEX, the fftwl_ declarations are
Chris@42: segregated into a separate interface file fftw3l.f03, which you
Chris@42: should include in addition to fftw3.f03 (which declares
Chris@42: precision-independent ‘FFTW_’ constants):
Chris@42:
use, intrinsic :: iso_c_binding Chris@42: include 'fftw3.f03' Chris@42: include 'fftw3l.f03' Chris@42:
We also support using the nonstandard __float128
Chris@42: quadruple-precision type provided by recent versions of gcc on
Chris@42: 32- and 64-bit x86 hardware (see Installation and Customization),
Chris@42: using the corresponding real(16) and complex(16) types
Chris@42: supported by gfortran. The quadruple-precision ‘fftwq_’
Chris@42: functions (see Precision) are declared in a fftw3q.f03
Chris@42: interface file, which should be included in addition to
Chris@42: fftw3l.f03, as above. You should also link with
Chris@42: -lfftw3q -lquadmath -lm as in C.
Chris@42: