Wilson Snyder
Wilson Snyder
Can you please attach a self contained test case showing this, ideally in test_regress format (see https://github.com/verilator/verilator/blob/master/docs/internals.rst)? I'm not sure if public inside interfaces has been tried before, so this...
It is much better to not use public, and instead either route all signals directly, or make a wrapper that has the additional outputs you want and looks at what...
See comment in #3590; I suggest see if that works in another pull request then we'll decide which to take.
I suspect V3Tristate is botching this since the nets are non-tristate nets. Perhaps you'd be willing to debug this with assistance?
There's a performance cost to propagating the tristate. I wonder instead should we simply assume any comparison with Z on a non-tristate net is false?
>Maybe it would be better to make sure that each node that may contain z value has user1p()? I think that might work well, as that is basically suggesting the...
BTW please make sure your test has ==, !=, ===, !==.
Please make a separate ticket, thanks. For the accessing of signals please describe how you think that would work with an example as it's not clear to me.
I have made similar specialization experiments overt the years, and they so far have been performance losses. The problem is once not all nodes are the same you have a...
Closing as this isn't something we have to do - but still welcome data on if it helps.