[sv-ec] covergroup sample method

From: John Havlicek <john.havlicek_at_.....>
Date: Wed Oct 17 2007 - 10:25:23 PDT
Hi Jonathan:

I would like to find a better way of moving data from assertions to
covergroups.

It is a user requirement to be able to use assertions to define the
correlated data fields to go into a covergroup.  The assertion may
collect various data fields at different simulation times and store
them in local variables.  The structure of the assertion defines how
the data are correlated.

When the assertion has collected all required fields, it wants to post
the data to a particular covergroup instance.

In general, multiple evaluation attempts or sub-attempts of the 
assertion may want to post in the same timestep.  The order in 
which they post is non-deterministic, but all the posts should get
into the covergroup instance.  In general, there is no bound on 
the number of posts in a timestep.

Does the class wrapper method you are suggesting accommodate such
usage?

Even if it does, I guess that I am looking for something cleaner.

J.H.


> Michael,
> 
> > The problem is that we want to trigger coverage from
> > within assertions using data in assertion local variables.
> 
> I too have observed exactly this requirement in several
> applications.  Thus far, I've worked around it in the 
> obvious-but-inelegant way.
> 
> However, I think it probably makes more sense to use the
> existing class machinery to build a suitable wrapper
> class, with a sample() method (written as a task so 
> that it's legitimate to call it from within an assertion)
> that does all the necessary business of copying its 
> arguments to data members of the coverage wrapper class
> and then calling the real coverage group's sample() method.
> 
> In other words, I'd be reluctant to add a new language 
> feature to support something that can rather easily
> be done in the existing language, especially at this
> late stage in the development of 1800-2008.
> 
> On the other hand, building a class that's specifically
> designed to be a wrapper for a covergroup most emphatically
> *does* raise interesting questions about extensibility of 
> covergroups and coverpoints.
> -- 
> Jonathan Bromley, Consultant

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Wed Oct 17 10:26:00 2007

This archive was generated by hypermail 2.1.8 : Wed Oct 17 2007 - 10:26:08 PDT