Subject: Re: questions about CBV
From: Adam Krolnik (krolnik@lsil.com)
Date: Fri Apr 12 2002 - 14:03:49 PDT
Hi John;
Thanks for the explanation of weak vs. strong clocking.
I guess I've used weak clocking without knowing it...
You write:
>Yes. In the statement layer of CBV, the local variables capture
>sampled values that are local to the thread and its descendents. The
That's good, but...
>I think it is accurate to say that the CBV analogue of your assertion
>would result in both processes terminating with the first "done". In
That's bad IMHO.
>CBV, the user needs to provide the code that will ensure the correct
>correspondences in cases like these. Often, one needs some auxiliary
>FSM. For example, ...
This is a simple protocol example, that a temporal regular expression
should be capable of handling. When threads sync to the same
set of events, then what good is having threaded assertions? It
would seem that they are self fulfilling - the first one passes,
the second one follows it...
>Much more complicated protocol correspondences are routinely coded
>in CBV.
I expect that they are. And for complex protocols, one expects to
provide more coding. For basic pipelines, one should not have to
create FSM's, delays, etc. for checking them. Erich talked about
this relating to first match vs. all match. I don't understand
what exactly what this implies. But I do understand what a user
would have a hard time coding to check the above example if threads
match on the same event (which I would guess is all-match.)
Thanks John, for the helpful comments about CBV.
Adam Krolnik
Verification Mgr.
LSI Logic Corp.
Plano TX. 75074
This archive was generated by hypermail 2b28 : Fri Apr 12 2002 - 14:06:14 PDT