pub fn mma_scaled_template<D: Dialect>(
a_elem: Elem<D>,
b_elem: Elem<D>,
cd_elem: Elem<D>,
k: u32,
n_a_registers: usize,
n_b_registers: usize,
n_c_registers: usize,
n_d_registers: usize,
scales_elem: Elem<D>,
scales_factor: u32,
) -> String