Filler cell X1 in 28nm node

[ThisIsNotSam]

you are supposed to run 1 or 2 opt rounds after route, but running place_opt on a routed database will kill the routing.

there must be a way to tell opt either not to mess with placement solution or to respect placement violations. look into the set_opt options, there are several.

 

[ranaya]

During postRoute, the Innovus up-sizes and down-size instances to fix timing violations, it will cause x1 gaps. Can we use place_opt to fix these gaps?

Would this help :

 

[stevenv07]

Would this help :
View attachment 164485

As stated, this command does not guarantee all gaps will be larger than 0.14um. Actually, after applying this command, only one single-wide gap is left. Do we fix one remaining gap manually?

 

[ranaya]

As stated, this command does not guarantee all gaps will be larger than 0.14um. Actually, after applying this command, only one single-wide gap is left. Do we fix one remaining gap manually?

That, if you preserved the routing in the design. After the placement, before starting the routing, can you do this and see whether post routing still results larger gaps ?