@Dinesh A: We know more or less the reason why, which boils down to the fact that hierarchical timing analysis doesn't work in OpenROAD. We are working on a method which involves flattening hierarchical DEF and running flat timing. The MPW-2 caravel is https://github.com/efabless/caravel but you need to check out by tag

mpw-2b

, which is the version that was taped out.

@Tim Edwards Sorry, I am not clear on your response. Caravel Makefile already has option to run top-level STA in typical corner, which uses Standard cell library + block level netlist/Spef + Top-level interconnecting netlist/Spef + SDC. My understanding this is the one of the industry standard approach to do top-level timing analysis https://github.com/efabless/caravel/blob/9949306c42ded3cad03e1f4566d7cd1f8215b0bb/Makefile#L629 Are we saying Hierarchically SPEF generated based on sub-macro's LEF is not accurate, we need flat DEF based on STA ?

"_Caravel Makefile already has option to run top-level STA in typical corner, which uses Standard cell library + block level netlist/Spef + Top-level interconnecting netlist/Spef + SDC_" --- Yes, it does. And according to what I currently understand, that doesn't work in OpenROAD.