harpv
Member level 4
Hi,
I understand that libraries define max_transition setting to indicate the limit of their characterization.
So anything beyond this will be flagged by static timing analysis as violations. In design sometimes if the violations are below a certain threshold they are waived.
There could be many reasons for a waiver like max transition on a cell which is not on a timing critical path, or a static signal with a transition violation etc.
For e.g. say in a slow corner, a buffer input sees a transition of 4.399 when the max_tran is 4.
This violation of .399 is below 10% of the required time and hence it's waived from fixing in the design.
Is it purely library vendor dependent as to whatever this threshold is or is there any general understanding in the industry regarding this so that even if we switch libraries these thresholds will remain?
Thanks
I understand that libraries define max_transition setting to indicate the limit of their characterization.
So anything beyond this will be flagged by static timing analysis as violations. In design sometimes if the violations are below a certain threshold they are waived.
There could be many reasons for a waiver like max transition on a cell which is not on a timing critical path, or a static signal with a transition violation etc.
For e.g. say in a slow corner, a buffer input sees a transition of 4.399 when the max_tran is 4.
This violation of .399 is below 10% of the required time and hence it's waived from fixing in the design.
Is it purely library vendor dependent as to whatever this threshold is or is there any general understanding in the industry regarding this so that even if we switch libraries these thresholds will remain?
Thanks