discrepancy in parasitic capacitance determined from Cadence and calc from BSIM param

i wanted to calculate the parasitic capacitance of an nfet_rf mos in Candence softwate, for which i employed the attched circuit in which the transistor is in the triode region; with drain, body and source connected to ground, while a sinosoidal input of 1.5V at gate and an inductor attached at gate to resonate with the gate capacitance (Cgs and Cgd in this case). ac analysis was employed and the capacitnce found to be 871.57f. next we gave this circuit the resonant frequency as input and at that value the input current became negligible indicating that the parallel combination of inductor and the gate capacitance is resonating. similar techniuqe was done to determine capacitances when transistor is in cutoff and saturation. HOWEVER, when the values of capacitances were calculated using the BSIM paramters, the capacitances were found to be around three times higher (2.357p in the above case). has anyone ever faced this sort of a discrepancy?

)