Chris@19: Chris@19: Chris@19: Extended and quadruple precision in Fortran - FFTW 3.3.4 Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19:
Chris@19: Chris@19:

Chris@19: Previous: Overview of Fortran interface, Chris@19: Up: Overview of Fortran interface Chris@19:


Chris@19:
Chris@19: Chris@19:

7.1.1 Extended and quadruple precision in Fortran

Chris@19: Chris@19:

Chris@19: If FFTW is compiled in long double (extended) precision Chris@19: (see Installation and Customization), you may be able to call the Chris@19: resulting fftwl_ routines (see Precision) from Fortran if Chris@19: your compiler supports the C_LONG_DOUBLE_COMPLEX type code. Chris@19: Chris@19:

Because some Fortran compilers do not support Chris@19: C_LONG_DOUBLE_COMPLEX, the fftwl_ declarations are Chris@19: segregated into a separate interface file fftw3l.f03, which you Chris@19: should include in addition to fftw3.f03 (which declares Chris@19: precision-independent ‘FFTW_’ constants): Chris@19: Chris@19:

Chris@19:

       use, intrinsic :: iso_c_binding
Chris@19:        include 'fftw3.f03'
Chris@19:        include 'fftw3l.f03'
Chris@19: 
Chris@19:

We also support using the nonstandard __float128 Chris@19: quadruple-precision type provided by recent versions of gcc on Chris@19: 32- and 64-bit x86 hardware (see Installation and Customization), Chris@19: using the corresponding real(16) and complex(16) types Chris@19: supported by gfortran. The quadruple-precision ‘fftwq_’ Chris@19: functions (see Precision) are declared in a fftw3q.f03 Chris@19: interface file, which should be included in addition to Chris@19: fftw3l.f03, as above. You should also link with Chris@19: -lfftw3q -lquadmath -lm as in C. Chris@19: Chris@19: Chris@19: Chris@19: