Denti:
They are very precise as long as the input is correct. However they do suffer a lot from input derived errors. If you have due to pressure change within the time between ATIS-cycles an input error of just one hPa it might make landing on a marginal runway impossible if you follow the GP for too long. Which is the reason why Boeing changed our SOP to only continue visually below any MDA/DA on IAN approaches, the GP is not longer deemed safe below that.
I don't believe we are speaking of quite the same thing. I am making reference to a Baro VNAV path that is locked in the nav database as part of the instrument approach procedure, not a path constructed by the flight crew.