transition violations

Status
Not open for further replies.

chanducs24

Member level 2
Joined
Oct 20, 2010
Messages
53
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
1,598
hi

i got 2 real transition violations after routing has been done..how can i fix and TranSlack is -0.072. plz five me some suggetions

Thanks,
chandra.
 

Hi Chandra,

Transition violations are mostly caused by:

1 - Detours caused by obstructions such as macros. This indirectly increases the load seen by the driver, which in turn increase the transition of that net. Using the maximum buffer size may not fix this problem. If this is the case adding an extra buffer should help.

2 - Insufficient drive strenth of the driver buffer (this time the net may be very short but the driver may not be sufficient to drive the load)


To fix transition violations simply use any of the below techniques.

1 - Adding a buffer before the violated pin.
- This time it is important that added buffer can be driven by the previous buffer.
- This is the most aggressive solution and it also helps fixing max net length violations, if they exist.

2 - Upsizing the driver of the net connected to the violated pin.
- Again increasing the driver will increase the load seen by the previous cell. Therefore care should be taken not to create new transition violations.
- This is very useful when fixing small transition violations.

3 - Modify the location of the cell driving the violated pin and reduce the net routing with more direct connections, which in turn will reduce the load seen by the driver.
- Most of the time this may not be efficient, especially when the routing density is high.

I don't know what you use to check maximum transition but if you use Prime Time, using what-if analysis should help (under ECO menu). After applying ECO, update timing and check if violation is fixed or not (by the way load is calculated using previous extraction results and library data, therefore routing data will not be available, but most of the time load caused by routing is less compared to cell input/output capacitances, unless the net is very long).

Otherwise if you are using place-and-route tools checker, then report the max trans violations after applying ECO. Personally I recommend using Prime Time's what-if analysis.

Specifically in your case the violation seems to be small, therefore upsizing the driver should help.

I hope it helps.

Best regards,
Gokhan
---
 
Last edited:
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…