it makes a lot of difference !
EIRP should be computed independantly for each cell, depending on:
– trx power
– number of combining stages and bts losses
– feeder length
and of course, you must add the actual antenna gain on top of all that. So each cell should be associated with its actual antenna model.
A difference of 4dB (which is the typical loss from one combining stage) makes A LOT of difference in the radio coverage.