• Implementation Models - The models are developed to balance simplicity and accuracy. 
  Linear approximations are used to simplify the implementation.
   
  
• Basic Modeling Fitting - The models are calibrated against a limited set of measured data points, and those datapoints are chosen to align to key data rates and operating modes.
   
• Model Parameters - The models do not include all parameters, such as PRE/POST emphasis.


• Adaptation - The models do not  include adaptive logic, such as DFE and CTLE, which dynamically adjusts to match the system loss.
   
• Process/Temperature - The process and temperature have little impact on the models

For information on what impacts the results for static power estimation from report_power see (Xilinx Answer 61291)