So it has nothing to do with it being a SNN and more to do with...

So it has nothing to do with it being a SNN and more to do with the datatype used... This doesn't contradict my point? These are all parallelised processes, so it will take the same amount of time to propagate a whole image through a layer of CNN vs propagating one event through an SNN. Maybe there is a tiny difference due to datatype but I doubt it is significant and am I correct in saying you are just speculating? Or do you know that it contributes a significant delay?

In any case, doesn't change that the fact that it is more the number of layers in the network rather than the type of network. Yes a CNN will use more power, will it be slower? I haven't heard a compelling reason why. This was the main point behind Simon Thorpes argument that the NNs of today are bloated. The time it takes for a human to recognise an image is too fast for it involve several "layers" of neurons.