All,
I think we can all agree to the following:
* Should clock tics (in *clock_times) be required? I think a model
that does not return clock_times is not a "good quality" model. Requiring
the Rx model to return clock tics should not be an issue for the model maker
unless it does not have a DFE or ADC.
* Should Rx_Decision_Time be required? I think a model that does not
return Rx_Decision_Time is not a "good quality" model. Requiring the Rx
model to return Rx_Decision_Time should not be an issue for the model maker
unless it does not have a DFE or ADC.
Should we require clock tics and Rx_Decision_Time? Personally, I have no
objection. The EDA tool can ignore them if it chooses. The only problem I
have is that if the model maker does not do a good job of outputting
clock_times or Rx_Decision_Time, It may be better to rely on the EDA tool
algorithms over the model deciding it.
Be careful about going to far in what is required vs. what is a "good
quality" model. I think that a "good quality" model is a Dual Model that
supports both statistical and time domain (Init_Returns_Impulse=True and
GetWave_Exists=True). I expect a BIRD requiring Init_Returns_Impulse=True
and GetWave_Exists=True Would not get very far.
Walter
Walter Katz
<mailto:wkatz@xxxxxxxxxx> wkatz@xxxxxxxxxx
Office 978.461-0449 x 133
Mobile 720.417-3762