A tough question:
If a circuit has less than 1000 elements, can BDA's Analog FastSpice be slower than Spectre/Hspice or just as fast as Spectre/Hspice?
As we know, there is not much multi-rate in a tiny analog circuit. So Analog FastSpice's multi-rate technology doesn't help. Furthermore, if the circuit consists of mostly Bsim3 or Bsim4 models, then the device model evaluation time dominates simulation time because a small circuit matrix is very fast to solve. According to your press releases, Analog FastSpice evaluates original device equations. So can we conclude that it has no advantage over Spectre/Hspice under these circumstances? I don't see any. Here I am assuming all the simulators take the same number of timesteps. If a simulator is faster because of less timesteps, that doesn't count. Less timesteps leads to worse accuracy.