In the case of matrix multiplication, it's better programmed in many-core programming (using GPUs) or in case of CPU multi-threading, it would make sense to use per row in very large matrices, or ...

To implement the multiplication 𝐶 = 𝐹 ⊙ A ∗ B using sparse matrices the F and A matrices are stored in CSC format so that the elements of each row are easier to gather, while the B matrix is stored ...