It depends on how you change the set point
The I-PV is better for applications where the setpoint is going to change in jumps instead of ramping smoothly from one value to another another.
Warning, you are now entering the s domain.
Why, because the PID adds two zeros ( lead circuits ) to the CLTF ( closed loop transfer function ) and the I-PV doesn't add any zeros. The zeros extend the bandwidth which means it will respond to high frequency changes in the SP better. This is why motion controllers use PID control However, if the SP is changed in jumps a high bandwidth is not desirable because of the sudden spike in the control output. In this case the lower bandwidth of the I-PV is more desireable.
Why do I call it I-PV? Because the proportional and velocity feedback gains are only in the feedback path and the feedback is negative. PID gains are all in the forward path and applied to the error. Sometimes I-PV is called I-PD because the proportional and derivative gains are in the feedback path.