I know generally speaking FFT and multiplication is usually faster than direct convolve operation, when the array is relatively large. However, I'm convolving a very long signal (say 10 million points) with a very short response (say 1 thousand points). In this case the fftconvolve doesn't seem to make much sense, since it forces a FFT of the second array to the same size of the first array. Is it faster to just do direct convolve in this case?

python scipy fft convolution
Edit
Report