As you might already know or be able to sense, convolution can become really slow as images or kernels get larger.

So how can we make it faster?