If some Arrays (or whatever) are defined using the Float128 type from Quadmath.jl, will operations from FFTW.jl be done in quad precision? The FFTW library itself supports quad-precision (Precision (FFTW 3.3.10)), but I couldn’t find any mention of precision in the FFTW.jl README or issues/PRs.
Looks like no, FFTW and AbstractFFTs only work out of the box with BLAS floats.
Thanks for the info @cgeoga
