Honored Contributor
Honored ContributorHello Axel,
I can verify the one LE difference you reported and see, that postfit netlists are different. But I don't see a timing difference, if I would, as long as timing isn't a critical application parameter, I would regard it a minor implementation detail. Usually other parts of a design will be more an object to speed optimization. I think, the test doesn't show much more than arbitrary results of the ambiguous "one state hot" coding. In contrast, if you define "sequential" coding style, the two test cases result in identical post mapping netlists (as far as I see) because the unique coding has no degree of freedom (at least with 2^n states). Thus I I don't see clear conclusions for FSM coding from the test, e. g. if the redundant "holding" transitions are meaningful somehow. I'll continue to omit them simply for writing economy, the same reason why I generally prefer what you call "one segment" form. Thank you for the interesting discussion, Frank