Similarly, we have taken 3p2 steps for division whereas division can be reduced to multiplication upto a constant factor and hence is again accomplished in a constant multiple of p log(p) steps.
It will probably suffice for cryptological applications to make the asymmetric assumption that the cryptographer (encryption/decryption process) uses multiplication that takes p2 steps while the cryptanalyst (the ``code-breaker'') uses multiplication that takes p log(p) steps. If the cryptosystems that we design are secure under such asymmetric assumptions they will also be secure under more symmetric assumptions!