We have a RTL design that works in multiple operation modes. Currently the DUT’s operation mode is specified thru a parameter that is passed in from the top-level testbench. The mode is realized/elaborated during the optimization phase (vopt) and determines the final DUT architecture. We have one testbench (TB) and I wish to use the same TB to run regressions for all the operation modes. What would be a good approach to randomize the configuration?
In our TB, we do have the UVM configuration object that can be randomized at runtime and pass to the sim environment. The DUT architecture cannot be changed during the simulation phase.
Thanks,
Richard