Re: 1076.1 LRM Interpretation: Initialization

Hi Deepika,

To answer your question, you have to look at the simulation cycle, as you 
suggest, and the definition of s'ramp. First, I assume that when you ask for the 
value at time 0, you mean after all delta cycles at 0 ns have been completed and 
time is about to advance. This state is defined in the LRM as the quiescent 
point..

If you look at the definition of s'ramp, you will see that the process is 
suspended in the "wait on domain" statement until the domain signal has an 
event, which is right after the quiescent point has been reached. Since the 
initial value of slope is 0.0, the value of s'ramp before reaching the quiescent 
point is in essence given by s'delayed, which means that since in your example s 
is assigned the value 10.0 at time 0 ns, the value of val at the quiescent point 
will also be 10.0. This corresponds to your scenario b). 

The rationale for this behavior (or more precisely the definition of s'ramp 
producing this behavior) is the understanding that the quiescent point is 
supposed to reflect a state of the model where all transients have settled. The 
definition considers all activity at time 0 ns as being startup transients that 
must settle before the quiescent point will be reached.

Thanks.
Ernst Christen

 On Mon, 23 May 2005 16:04:19 -0400, Deepika Devarajan wrote:
> All,
> I have a question regarding the interpretation of the 1076.1 LRM for a 
mixed-signal model. 
> Consider the following model:
> ENTITY ramp_test IS
> END ENTITY ramp_test;
> ARCHITECTURE behav OF ramp_test IS
> SIGNAL pulse_signal : REAL := 2.0;
> QUANTITY val : REAL;
> BEGIN
> pulse_signal <= 10.0;
> val == pulse_signal'ramp(1.0,1.0);
> END ARCHITECTURE behav;
> What should val be at time = 0? Should (a) val rise from 2.0 to 10.0 over the 
first second or (b) the initialization cycle force val to be 10.0 from t=0?
> While (a) seems to be more intuitive, (b) seems to be consistent with the 
simulation cycle (refer Sections 12.6.4 and 12.6.6 and pp. 211-212).
> Any comments/feedback is appreciated.
> Thanks,
> -Deepika