some question about SI fix.

[Kaisia]

when fixing SI and timing in signoff mode, i get some warnings as below:

***WARNSI-2239) The victim net XX has a receiver, XXX:A(cell XXXX), with a non-monotonic output waveform, the bumpy transition can cause incorrect delay uncertainty measurements to be reported. Perform ECO on this net to reduce the coupling capacitance.

I want to know the reason and how to do this ECO to work efficiently.

 

[songzm2008]

You can insert a buffer in the net connected with XXX:A

 

[srideepa]

The warning can be interpreted in this way : The net is connected to cell XXX:A (Receiving End). It says that the waveform is non monotonic(It switches its direction from rising to falling or vise versa). So the problem is Xtalk noise is causing a glitch there.

To avoid the aforementioned warning, you can add a buffer at the driving side (The net will be driven by some YYY:Y which needs to be buffered). Or, you can increase the spacing for this net. Otherwise, simply reroute this net in some other direction to avoid it from being victimized (Depending upon the congestion, you may get good as well as bad results).

 

[Kaisia]

how to specify the space for the select nets by the tool?

 

[srideepa]

Hi,
There is something called "NON DEFAULT RULE" in which we can specify the custom spacing, shield and width for net names. Enforce NDR for the net that is having crosstalk and then do eco route. But you can buffer the sourcepin which will be easy.

Regards,
R.Srideepa

 

[Kaisia]

if adding some buffer near the 'driving' cell, does it needed to point the added buffer output to the very 'receiving' input only?(the 'driving' cell will go to other cells' inputs.)