how to fix time violation during synthesis a scenario

[kirteshmiet]

Hi I have a path in tming report that is giving me setup violation after synthesis , I tried to fix it by giving it more weight in design compiler but still it is not meeting what is the next step i can do . it is violating with 300ps and if ican mange it to 100ps it is ok for me to go as client has given me this margin .

Thanks

 

[rca]

RTL coding modifications.
Synthesis of this sub-block before doing the whole system.
Force the tool to use the force speed grade (rc for example: set_attribute user_speed_grade very_fast [get_attribute subdesign ...])

 

[kirteshmiet]

RTL coding modifications.
Synthesis of this sub-block before doing the whole system.
Force the tool to use the force speed grade (rc for example: set_attribute user_speed_grade very_fast [get_attribute subdesign ...])

thanks rca ..

 

[mr_vasanth]

Try to enable "boundary optimization" during synthesis and see the result. If boundary optimization is enabled combo path across module boundaries will get optimized.

 

[sharath666]

You can also try cell sizing for the cells in the critical path. Use bigger cells instead of existing cells.

 

[kirteshmiet]

You can also try cell sizing for the cells in the critical path. Use bigger cells instead of existing cells.

these all the stpes i tried let say everything is is at it's optimum level then what you will do??? and remember it is at synthesis level.

 

[englishdogg]

my personal opinion - before trying / adding anything to the existing vanilla script that your using is to debug.
I would review the timing report to see if what causes this violation to occur
- is it the cell delays impacting the timing thereby cell selection during synthesis
- has synthesis not optimized the netlist
- has it not done better opto, is there room to optimize logic to reduce level of hierarchy
- are there any preserves
- constraints are well applied - is it a multicycle , delays, pessimism etc

depending on what you diagnose is something that i would add in the scripts else its all trial and error and machine time while we wait for results.

Say you see that it traverses across hierarchies - i would try ungroup, say cell delays are high or it has not used complex cells - i would review the dont_use list etc and so on