Ignoring the cache hierarchy on a CPU for matrix multiplication gets you a 100x performance drop, just like a GPU

Wow! This is remarkable accurate

Isn't that how that always is?

they compare im2col performance with a GPU, while you don't need explicit im2col on a GPU

Isn't this a software problem being solved in hardware? Ideally you would try to avoid going to memory in the first place by fusing the operations, which should be much faster than speeding up memory ops. E.g. you should never do an explicit im2col before a convolution, it should be fused. However it's hard to argue with a 0.019 mm2 area increase.