7.4 Plan execution in Fortran

In C, in order to use a plan, one normally calls fftw_execute, which executes the plan to perform the transform on the input/output arrays passed when the plan was created (see Using Plans). The corresponding subroutine call in modern Fortran is:

 call fftw_execute(plan)

However, we have had reports that this causes problems with some recent optimizing Fortran compilers. The problem is, because the input/output arrays are not passed as explicit arguments to fftw_execute, the semantics of Fortran (unlike C) allow the compiler to assume that the input/output arrays are not changed by fftw_execute. As a consequence, certain compilers end up repositioning the call to fftw_execute, assuming incorrectly that it does nothing to the arrays.

There are various workarounds to this, but the safest and simplest thing is to not use fftw_execute in Fortran. Instead, use the functions described in New-array Execute Functions, which take the input/output arrays as explicit arguments. For example, if the plan is for a complex-data DFT and was created for the arrays in and out, you would do:

 call fftw_execute_dft(plan, in, out)

There are a few things to be careful of, however:

• You must use the correct type of execute function, matching the way the plan was created. Complex DFT plans should use fftw_execute_dft, Real-input (r2c) DFT plans should use use fftw_execute_dft_r2c, and real-output (c2r) DFT plans should use fftw_execute_dft_c2r. The various r2r plans should use fftw_execute_r2r. Fortunately, if you use the wrong one you will get a compile-time type-mismatch error (unlike legacy Fortran).
• You should normally pass the same input/output arrays that were used when creating the plan. This is always safe.
• If you pass different input/output arrays compared to those used when creating the plan, you must abide by all the restrictions of the new-array execute functions (see New-array Execute Functions). The most tricky of these is the requirement that the new arrays have the same alignment as the original arrays; the best (and possibly only) way to guarantee this is to use the ‘fftw_alloc’ functions to allocate your arrays (see Allocating aligned memory in Fortran). Alternatively, you can use the FFTW_UNALIGNED flag when creating the plan, in which case the plan does not depend on the alignment, but this may sacrifice substantial performance on architectures (like x86) with SIMD instructions (see SIMD alignment and fftw_malloc).