Re: IBIS 9/15/95 MINUTES

Again I want to emphasize some simple faces about non monotonicity

1. IBIS does not preclude non monotonic data

2. Simulation of non monotonic is as much of a black art as any non linear simulation. We have succeessfully simulated them.

- kumar
 
> NON-MONOTONOCITY
> Ron Christopher lead the discussion by indicating for a frame of reference
> that handling non-monotonic drivers seems to be a barrier to a major
> vendor (IBM) for adopting IBIS. The goal of IBIS was for a transfer of
> useful information while preserving proprietary process information. IBIS
> does allow the full characterization of non-monotonic VI data (with warning
> messages), but the real problem is that the actual internal mechanisms are
> guarded. Furthermore, performance information is not available. Both
> are needed for simulator vendors to actually process the data properly.
>
> Ron suggested that some independent vendor such as Zeelan could provide
> measurement based data to give characterization information. That would
> be a start, but not necessarily the total picture. Also, there may
> be problems of vendors suppling such sensitive chips for measurement
> for public distribution and the willingness of whoever does the measurements
> to distribute the information. Ron will pursue this path.
>
> The technical discussions regarding non-monotonic impact related to possible
> delay artifacts. Jon Powell indicated that attempts to reverse engineer
> the non-monotonic effects to provide a simulation model were to date
> unsuccessful. The simulation data did not match the actual data. So the
> internal mechanisms are probably needed.
>
> Other companies may have similar technology. However, TI's reference to
> GTL structures are probably not similar.
>
> Most likely the non-monotonic architecture will be more fully understood
> and could be implemented in IBIS. Sources of information may emerge
> through patents and pending patents, papers and other vendors. Vendors
> interested in using IBIS for these effects need to participate in the IBIS
> activities to supply the essential information so that EDA vendors can
> correctly architect the simulators. Ron, through his contacts could urge
> vendors to become active in IBIS.
>
Received on Wed Sep 20 05:27:10 1995