To make sure I'm understanding this, the issue here is that it's not vectorizing correctly, so that the outputs for the different entries are not independent? And this shows up if you have NaNs, causing the NaNs to propagate everywhere? And this is true even with the axis argument?

Yes it would seem so, which axis argument sorry?

Is this what you are using: https://docs.scipy.org/doc/scipy-1.14.1/reference/generated/scipy.signal.fftconvolve.html?

ah yes sorry I though you meant something in dspeed, I'll see if this fixes, I'm going to take the processor apart tomorrow to see where the problem is

ok so if any nans are in input it will give nan output even with axis specified