Yes that seems to be the case. The apparant problem now is of course that we cannot calibrate the library since we do not know the file sizes. This is however normally not a problem since the FFTW library actually will measure powers of two. So if we would set up a program in which we simply plan for sizes 1024 2048 4096 8192 up to something around 60'000'000 or so and then save the wisdom file then the library would work quite fast in all situations.
(In reply to comment #0) > Something we discussed a hile ago was the possibility to use an fftw > calibration program that will calibrate the library before using it. This will > solve the bug that the library appears to hang. > is the calibration occurring in bpmCounter::createFFTWPlans?
Something we discussed a hile ago was the possibility to use an fftw calibration program that will calibrate the library before using it. This will solve the bug that the library appears to hang.