Minutes: Analog System Verilog Assertions: Oct 21st, 2008

From: Anand Himyanshu <HIMYANSHU.ANAND_at_.....>
Date: Wed Oct 22 2008 - 12:27:00 PDT
Attendees: Listed at the end of the summary
 
Documents: All documents related to Analog assertion released to the
committee can be accessed at
http://www.eda-stds.org/verilog-ams/htmlpages/public-docs/AMS_Assertions
/
<http://www.eda-stds.org/verilog-ams/htmlpages/public-docs/AMS_Assertion
s/> 
 
Summary: (Please correct if I misquoted someone or missed some important
point)
------------------------------------------------------------------------
------------------------------------------------
 
Discussion on scope of the working group was discussed. There is still
not full agreement on the scope, though it was felt by a majority that
the scope should be more focused so that problem is practical enough to
solve while at the same time interesting enough for not only analog
mixed signal designers but also for analog designers. Mike would like to
see going beyond the waveform property verification and including more
categories from his taxonomy in the assertion language. Ken Bakalar
wanted to separate the implementation details from the language
definition, as to whether the assertion would be evaluated by analog
engine or digital engine. 
 
Question regarding designing a universal analog assertion language (like
PSL with flavors for Verilog-AMS, VHDL-AMS, System-C) was discussed. It
was felt that this working group should focus its efforts on extending
SVA instead of trying to define a universal language.
 
Questions were raised (by Ken Bakalar?) about the potential
problems/issues for extending SVA to dense time domain and whether we
could leverage some of the work done for extending PSL. Dejan mentioned
that fundamentally there does not seem to be a big hole to extend SVA
(regular expressions) that includes dense time. The challenge might be
in the integration of the LTL temporal operators + dense time with
regular expressions. There has been some work in this area which can be
looked into. (Refer: http://www-verimag.imag.fr/~maler/Papers/timed.pdf
<http://www-verimag.imag.fr/~maler/Papers/timed.pdf> ,
http://www-verimag.imag.fr/PEOPLE/Oded.Maler/Papers/kleene.pdf
<http://www-verimag.imag.fr/PEOPLE/Oded.Maler/Papers/kleene.pdf> ). 
 
John Havlicek mentioned that LTL operators have been added to the P1800
and that all the basic operators were there. He felt that the local
variables will probably play an important part in the analog extension
to SVA. Dejan mentioned that local variables should be allowed to be
used at any time. John said that they should be restricted when they can
be assigned but they could be referenced after they were assigned at any
point.
 
The work done at IIT Kharagpur was also discussed briefly. The working
group did not reach a consensus on whether monitors should be in analog
domain and produce digital signals to be used within the digital
assertions. Scott Little felt that assertions should not be split.
Himyanshu Anand wanted the language to be implementation agnostic and
leave it to the simulators to decide to split the work for evaluating
the assertions.
 
The question whether analog assertions should control the accuracy of
the simulator was discussed. It was felt by a majority and a consensus
was reached that accuracy needs to be defined in assertions to have a
local affect on the simulation rather than a global affect. The basic
things to extend SVA to analog properties would be to add real
(continuous) domain signals to SVA and allow inequalities to be defined
inside assertions involving reals along with the notion of asynchronous
time and handling intervals. The work will involve combining temporal
logic with regular expressions and defining them over dense time (or
asynchronous time)
 
It was decided that Kevin's proposal did not belong to this working
committee. Frequency analysis properties along with parametric sweeps
were also discussed.
 
Prabal asked whether PSL was looked at and why SVA was chosen for the
analog property extension work. Himyanshu mentioned that SVA was chosen
because there was a parallel work going to start soon that will
integrate System Verilog into Verilog-AMS, thus extending SVA would
facilitate that integration. Ken Bakalar mentioned that extending SVA
with small things (reals and inequalities involving reals) would allow
rapid prototyping of analog properties that would require changes to
only to digital engine. 
 
David Sharrit had raised the question whether the analog property
language will replace the MDL and proprietary .measure statements.
Himyanshu felt that it would be complimentary rather than a super set of
such features. Scott mentioned that there are features in .measure like
languages which do not sit well the notion of assertion or property.
Himyanshu mentioned that there might be some features which will require
to be added to the assertion language but probably not all, like the
frequency calculation. Scott added some other features which probably
should be accessible from the assertion language like slew rate,
settling time, etc.
 
 
Name	 Company	 18-Jun-08	 18-Aug-08	 23-Sep-08
7-Oct-08	 21-Oct-08	
							
Erik Seligman	 Intel	 1	 1	 1	 1	 1	
Thanapoom Lertpanyavit 	 Intel	 1	 1	 1	 1	 1	
David Smith	 Synopsys	 					
Mike Demler	 Synopsys	 1	 1	 1	 1	 1	
Ed Cerny	 Synopsys	 1	 1	 1	 1	 	
Venkataramanan Srinivasan	 Synopsys	 		1
1	 	
Dave Cronauer	 Synopsys	 1	 1	 1	 1	 1	
Prabal Bhattacharya	 Cadence	 		1
1	
Scott Cranston	 Cadence	 1	 				
Adam Sherer	 Cadence	 1	 				
Walter Hartong	 Cadence	 					
Kishore Karnane	 Cadence	 				1	
Don O'Riordan	 Cadence	 					
Ned Utzig	 Cadence	 					
Martin O'Leary	 Cadence	 1	 1	 			
Dennis Brophy	 Mentor Graphics	 1	 1

Kenneth Bakalar	 Mentor Graphics	 	1	 1	 1
1	
Kevin Jones	 Individual Contributor	 	1	 1	 1
1	
Patrick O'Halloran	 Tiburon Design Automation
1	 1	 1	
David Sharrit	 Tiburon Design Automation
1	 	
Marek Mierzwinski	 Tiburon Design Automation	 	1
1	 1	
Dejan Nickovic	 Verimag	 	1	 	1	 1	
Pallab Dasgupta	 IIT-KGP	 					
Sri Chandra	 Freescale	 1	 1	 	1	 1	
Dave Miller	 Freescale	 1	 		1	 	
John Havlicek	 Freescale	 1	 1	 1	 1	 1	
Hillel Miller	 Freescale	 1	 	1	 1	 1	
Scott Little	 Freescale	 	1	 1	 1	 1	
Himyanshu Anand	 Freescale	 1	 1	 1	 1	 1	
Brian Mulvaney	 Freescale	 1	 	1	 		
Phillip LaPlace	 Freescale	 		1	 		
Daniel Christopher	 Freescale	 	1

Linda Lin	 Freescale	 		1	 		
Saurabh Garg	 Freescale	 	1	 1	 1	 	
Benjamin Ehlers	 Freescale	 	1	 1	 		
Zhang Yonggang	 Freescale	 					
Bill Read	 Freescale	 1	 				
Fais Yaniv	 Freescale	 			1	 	
Mark Kole	 NXP	 					
Jasmine	 NSC	 		1	 		
David Tamura	 NSC	 					
Kevin Cameron	 True Circuits	 		1	 1	 1	
David Jonathan	 Qualcomm	 1	 				
Laisne Michael	 Qualcomm	 1	 				
Li Xiang	 Qualcomm	 1	 				
							
	Attendance 	 19	 18	 21	 21	 17	
 
Himyanshu Anand
EDA Strategy, Vendor Relation and Customer Collaboration,
Design Technology Organization,
Freescale Semiconductor
 
7700 W. Parmer Lane,
Loc/MD: TX32/PL34,
Austin, TX - 78729
 
Ph    : +1-512-996-5623
Fax   : +1-512-996-7432

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Wed Oct 22 12:29:04 2008

This archive was generated by hypermail 2.1.8 : Wed Oct 22 2008 - 12:29:25 PDT