Why do corners (ss, ff, tt) require different simulation times?

[diarmuid]

Hello,

One thing I noticed is that when simulating over ss, ff and tt corners there is a definite difference in simulation times.

These are as follows:

- ff corners are slowest
- ss corners are fastest
- tt corners are in between

In general I am seeing ss corners to be ~ 10% faster than tt corners, and ~ 20/30% faster that ff corners.

First of all:

1.) Has anyone else noticed this?
2.) Are the models not the same over corners (vary only be parameter values)?

... anyone have any ideas why this is occuring?

Thanks,

Diarmuid

 

[dick_freebird]

The faster the circuit activity, the more the simulator has
to downrange the timestep, meaning more timesteps to
solve.

 

[diarmuid]

I see, so ff results in more activity i.e. smaller timesteps and so takes longer.

Makes sense now.

Thanks.