More and more, hardware manufacturers optimize their chips specifically for matrix multiplication workloads. For instance, Nvidia has its Tensor Cores and today most cellphones have AI specialized ...