QR decomposition for 8x8 matrix

QR decomposition for 8x8 matrix, this model is used for Homography Estimation in RANSAC.
4 point pairs are randomly selected and then formed the matrix in the following way.
A =
ax1, ay1, 1, 0, 0, 0, -ax1*bx1, -ay1*bx1;
ax2, ay2, 1, 0, 0, 0, -ax2*bx2, -ay2*bx2;
ax3, ay3, 1, 0, 0, 0, -ax3*bx3, -ay3*bx3;
0, 0, 0, ax1, ay1, -1, -ax1*by1, -ay1*by1;
0, 0, 0, ax2, ay2, -1, -ax2*by2, -ay2*by2;
0, 0, 0, ax3, ay3, -1, -ax3*by3, -ay3*by3;
ax4, ay4, 1, 0, 0, 0, -ax4*bx4, -ay4*bx4;

One could easily modify it into general purpose 8x8 matrix QR decomposition.

After running the simulink model, the error is calculated by Q*R-Matrix. One could check it.