Effective SIMD Vectorization for Intel Xeon Phi Coprocessors

Algorithm 4

do j = 1, 4

 do k = 1, 4

  sumx = 0.0

  sumy = 0.0

  do i = 1, 4

   sumx = sumx + matrixA(i,k)   matrixB(i,j)

   sumy = sumy + matrixA(i,k)   matrixB(j,i)

  enddo

  matrixC(k,j) = sumx

  matirxD(j,k) = sumy

 enddo

enddo