Simulations are necessary, but they need to be confirmed by measurements. Quite often simulation and measurement are required together, as they are complementary development tools. In case there is no probe point where it is needed and so the signal can not be measured at the end of the transmission line (e. g. the receiver pad) probing needs to be done somewhere in the middle (e.g. at a serial termination resistor). In such cases simulation are required to complete the picture. As it is difficult to judge on such a measurement result, it is necessary to simulate and compare the signal at the measured point, but evaluate to the (simulated) signal quality at the receiver pad. As alternative high end measurement Equipment can do the simulation realtime during the measurement.
The picture below shows a measurement at 5.3Gb/s. At this data rate it is already insufficient to measure at the back side of the PCB underneath the package.
The upper left picture shows the measured signal at the ball with the original termination settings. The eye is nearly closed. The eye opening could be improved by adjusting the termination settings of the receiver to the line impedance of the PCB. But even such an eye would not be good enough for the receiver to correctly receive the signal. However, that system was working fine, and so a simulation was done to check on the signal quality at the receiver pad.
The center two pictures show the simulated eye at the probing position. The correlation between simulation and measurement is very good. Finally the signal at the pad was checked (right two pictures). Both settings show fairly good signal integrity at the pad. In the end the improvement due to the reduced termination resistance was not really worth to pay the price on power. By these simulations it could be shown, that poor eye quality measured at the ball was caused by the reflection of the receiver and the package of the receiving device.