RE: [sv-ac] Statistics counters proposal

Hi Lisa,
 
yes, it would be a 2-step process but sequence coverage is the one less
practical and used. Therefore, this would force the person to think why
sequence coverage (with all matches) is needed.
 
Best regards,
ed
 


________________________________

	From: Lisa Piper [mailto:piper@cadence.com] 
	Sent: Monday, May 07, 2007 7:08 PM
	To: Eduard Cerny; sv-ac@eda.org
	Subject: RE: [sv-ac] Statistics counters proposal
	
	
	Thanks Ed!   
	 
	I agree with your point on the disable iff.  It should be
looking at the "property_spec" only.   I think the existing text misled
me on this.
	 
	I'm not sure I like restricting it to a sequence instance
because then it will always be a two step process to define it (e.g. I
would have to define the sequence and then use the cover property).  
	In case it was not clear, this is important since the coverage
results are stated to be different depending on whether the
property_spec is a sequence (all match) or a property (1 match per
attempt)..
	 
	Lisa


________________________________

		From: Eduard Cerny [mailto:Eduard.Cerny@synopsys.com] 
		Sent: Monday, May 07, 2007 2:09 PM
		To: Lisa Piper; sv-ac@eda.org
		Subject: RE: [sv-ac] Statistics counters proposal
		
		
		hi Lisa,
		 
		I think that diable iff should not change the status
from sequence to property.  
		 
		 Also, sequence expression could be restricted to
sequence instance, to make it clear from the definition that it is a
sequence. 
		 
		Best regards,
		ed
		 


________________________________

			From: owner-sv-ac@eda.org
[mailto:owner-sv-ac@eda.org] On Behalf Of Lisa Piper
			Sent: Monday, May 07, 2007 2:02 PM
			To: sv-ac@eda.org
			Subject: [sv-ac] Statistics counters proposal
			
			
			Hi all,

			 

			I would like to get early feedback on a proposal
that I am writing to clarify the statistic counters that are associated
with assertions. The current chapter 17 LRM specifies that these
counters must exist for cover statements only, yet 29.1.2 text is
written as though all counters are available for all assertions.  I
think there is value in having the statistic counters for all
assertions. For assert and assume statements, it may be useful to know
the number of failures and passes. A failure indicates a bug of some
kind while a pass is confirmation that the stimulus tested the behavior.
For cover properties, only the number of non-vacuous passes may be of
significance. 

			 

			I would like to propose that a minimal set of
pass/fail counts be required for all assertions (immediate and
concurrent), and that others can be optionally provided if desired.  And
only the pass counter is required for cover properties (others are
optional).  Are there any objections to this principal?

			 

			 The other part of the proposal is to clarify
how to distinguish a property_spec from a sequence_expr. (this is Mantis
1768)  A property_spec that is just a single sequence_expr shall be
interpreted as a sequence, whereas a property_spec that is not just a
sequence_expr shall be interpreted as a property.  For example, an
expression that contains an implication operator or a disable iff term,
is considered a property. A property instance that is just a single
sequence_expr is interpreted as a sequence (assuming there is not any
default disable iff clause).  Does this sound ok?

			 

			Lisa


			-- 
			This message has been scanned for viruses and 
			dangerous content by MailScanner
<http://www.mailscanner.info/> , and is 
			believed to be clean. 


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.