Re: [sv-ec] non-blocking event trigger

From: Steven Sharp <sharp_at_.....>
Date: Fri Sep 22 2006 - 13:43:42 PDT
>From: "Rich, Dave" <Dave_Rich@mentor.com>

>We have the ->> non-blocking event trigger which schedules its event for
>the NBA region.. Presumably this is a testbench feature designed for use
>in program blocks, but since we don't have a re-NBA region, this
>requires that you go back to the non-program block regions (the design)
>before unblocking the events waiting in the program block.

I believe these were suggested long before program blocks.  The idea was
just to give triggering of events the same options as assigning to
registers.  If you view an event trigger as an update of an abstract
variable that has the concept of a change but not an actual value, then
it is reasonable to allow a nonblocking update.  These were not proposed
for any specific use in program blocks, and the people proposing program
blocks may not have paid them much attention.


>And why is it we need NB triggers to program event variables, but can't
>have NBAs to other program variables?

If NBAs to program variables are not allowed, then I agree that NB triggers
to program events should not be allowed either.  I think this is just a
something that was not noticed.  The two features were added independently,
and nobody noticed that this should be specified.

I think that our implementation has assumed that all restrictions that
apply to NBAs also apply to NB triggers (if they make sense), even if
the LRM did not spell that out.

Steven Sharp
sharp@cadence.com
Received on Fri Sep 22 13:43:46 2006

This archive was generated by hypermail 2.1.8 : Fri Sep 22 2006 - 13:44:11 PDT