The calculations for the indexes into the target matrix, "m" multiply by 3
rather than 4 which causes every n*3 entry to be overwritten and for the last 4 entries
to always be 0.  ie:

 m[0*3+3] = p[3][0]
 m[1*3+0] = p[0][1]

Both assign to entry 3, where: 


 m[0*4+3] = p[3][0]
 m[1*4+0] = p[0][1]

Properly assign to entries 3 and 4.