Re: AMS Designer & nlAction

Andrew:
I hand edit the ams.env file.  It seems that if I don't have the exact
whitespace that it is expecting it deletes the line when design prep
is run.  That is if I add include files in the HED GUI and generate an
ams.env file it works.  If I create the ams.env file manually and
don't make the whitespace the same as the way the HED makes it then it
gets deleted.  Has anyone else seen the case where lines added to the
ams.env file get deleted when running design prep?
---
Erik


Andrew Beckett <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> Hi Erik,
> 
> I'm quite surprised by this, because I wouldn't have thought that the LDV bit
> actually reads or touches the ams.env file. I was using ldv 50.s010 and this
> was fine - can't easily check an older version anytime soon because I'll be out
> of the UK office for a week or so (and I'd have to specifically get hold of
> an old version which I don't have installed). 
> 
> If this bothers you, I'd take it up with support (perhaps via the SR you have
> already for the netlisting stuff).
> 
> Regards,
> 
> Andrew.
> 
> On 7 Nov 2003 11:48:47 -0800, [EMAIL PROTECTED] (Erik Wanta) wrote:
> 
> >Andrew:
> >I noticed that some versions of ldv delete my amsDirect initFile
> >option in the ams.env file when I run design prep.  ldv 50.s007
> >removes my initFile option and doesn't work but ldv 51 doesn't and
> >works.
> >---
> >Erik
> >
> >
> >Andrew Beckett <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> >> Erik,
> >> 
> >> I've been testing this, and it all works fine, as you've seen.
> >> 
> >> It's documented (Chapter 13 of AMS Environment User Guide in IC5033).
> >> It wasn't on sourcelink a few days ago, but is now.
> >> 
> >> I've also just written up a solution for sourcelink which should (hopefully) get
> >> published on sourcelink after review.
> >> 
> >> Andrew.
> >> 
> >> On Mon, 03 Nov 2003 18:48:10 +0000, Andrew Beckett <[EMAIL PROTECTED]> wrote:
> >> 
> >> >Didn't get to this today, sorry.
> >> >
> >> >But that looks like what I'd have expected to work.
> >> >
> >> >netlisting procedure support for AMS Designer was only added for IC5033 (it was 
> >> >quite
> >> >a big change, so wasn't added in an IC5032 ISR). So you can't do it in IC5032, 
> >> >sorry.
> >> >
> >> >Andrew.
> >> >
> >> >On 3 Nov 2003 02:13:35 -0800, [EMAIL PROTECTED] (Erik Wanta) wrote:
> >> >
> >> >>Andrew:
> >> >>I got this to work by adding the following to my ams.env:
> >> >>amsDirect        initFile        string  "/path/mynetlistprocedure.il"
> >> >>
> >> >>Note I don't see amsDirect initFile documented on sourcelink anywhere.
> >> >>
> >> >>So, one can add netlist procedures to 50 33 for amsd.  It is too bad
> >> >>that assura crashes with 50 33 so we can't move to it.  Any ideas on
> >> >>how to get the netlist procedure working with 50 32?
> >> >>---
> >> >>Erik 
> >> >>
> >> >>
> >> >>Andrew Beckett <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> >> >>> Erik,
> >> >>> 
> >> >>> OK, I'll look at this when I get back into the office tomorrow. I think 
> >> >>> another
> >> >>> customer has hit the same problem in my absence and one of my colleagues is
> >> >>> dealing with it.
> >> >>> 
> >> >>> Anyway, I'll answer some of the questions below:
> >> >>> 
> >> >>> On 2 Nov 2003 07:10:20 -0800, [EMAIL PROTECTED] (Erik Wanta) wrote:
> >> >>> 
> >> >>> >Andrew:
> >> >>> >I have taken this up with customer support and am told that you are
> >> >>> >the initiator of PCR 601739.  Support can't get this to work.  They
> >> >>> >are telling me that they are waiting to contact you.  I figured I
> >> >>> >would try to go right to the source and see if there is an easy
> >> >>> >solution to this problem that you were aware of. 
> >> >>> 
> >> >>> I'll have to dig out my notes - I can't remember if I actually tried the 
> >> >>> solution,
> >> >>> since the software wasn't available when R&D proposed the solution.
> >> >>> 
> >> >>> > I am not relying on
> >> >>> >a response and don't mean to put pressure on you or bother you on
> >> >>> >vacation.  I just thought I would post in the hopes that maybe others
> >> >>> >are having the same problem and might be interested in the problem and
> >> >>> >have ideas for a possible solution.  It seems like a common issue that
> >> >>> >one would want to not netlist a component for simulation and still
> >> >>> >netlist for LVS.  
> >> >>> 
> >> >>> I agree - however, AMS Designer netlisting is rather different from
> >> >>> other netlisters (see below):
> >> >>> 
> >> >>> > I would have thought that nlAction ignore would have
> >> >>> >worked.  I am surprised that nlAction is used by assura instead of
> >> >>> >lvsIgnore. 
> >> >>> 
> >> >>> All netlisters honour nlAction - it's a fundamental OSS capability, so
> >> >>> each netlister doesn't have to do anything special to support it.
> >> >>> 
> >> >>> > Why does ams designer netlist only the symbol?  Why
> >> >>> >doesn't it use the spectre view?  What if I have 2 models for the same
> >> >>> >component and want to use the view name to specify which model to use
> >> >>> >(in spectre I could use componentName)? 
> >> >>> 
> >> >>> The idea with AMS Designer is that each cellView creates its netlist
> >> >>> independently - this is to avoid extensive netlisting time when you want
> >> >>> to run a simulation. Each cellView can be netlisted, and even compiled,
> >> >>> at check-and-save time, and so when any user wants to simulate, all
> >> >>> that needs to be done is elaboration (if not up to date), and running the
> >> >>> simulation. No netlisting at all! Even if you change the view switching in
> >> >>> the config, all you  need to do is re-elaborate and run. None of that
> >> >>> lengthy partitioning and re-netlisting that needed to be done with
> >> >>> spectreVerilog.
> >> >>> 
> >> >>> The consequence of this is that view switching makes no difference at
> >> >>> netlisting time. Normally you'd put nlAction=ignore on the stopping view - but
> >> >>> since stopping views are not used for AMS netlisting - they only get used
> >> >>> to guide the elaboration - it can't have an impact on netlisting.
> >> >>> 
> >> >>> This is why it needs to be done with a netlisting procedure - either
> >> >>> a specific netlisting procedure for that component, or by overloading the
> >> >>> standard one so that it supports a property to control ignoring of the 
> >> >>> component.
> >> >>> 
> >> >>> > Why does amsd dumpt the
> >> >>> >verilog.vams netlists into the library instead of a netlist directory?
> >> >>> 
> >> >>> For the reasons I said above - the netlist is just another representation of 
> >> >>> the
> >> >>> cellView - and so it makes sense that this derived bit of data goes along
> >> >>> with the cellView. If done at check-and-save time, it can be used by
> >> >>> any user of the cellView, rather than each user having to netlist their own 
> >> >>> copy.
> >> >>> 
> >> >>> > I load the netlist procedure to not netlist the instance for ams
> >> >>> >designer since the nlAction approach won't work, but amsdirect doesn't
> >> >>> >see this loaded function when I run HED->AMS->Design Prep.  It is my
> >> >>> >understanding that amsdirect doesn't use SKILL.  How does it work from
> >> >>> >CIW->Tools->AMS->Netlist?  What is the mechanism to get amsdirect
> >> >>> >aware of loaded SKILL?
> >> >>> 
> >> >>> That I don't know, and will have to research. I've not seen any emails asking
> >> >>> me yet, but perhaps they're waiting becausing they know I'm on vacation...
> >> >>> >
> >> >>> >Please don't feel a need to respond.  I am just posting the above
> >> >>> >questions so that I get them out in a public forum in the hopes that
> >> >>> >others are tackling the same issues.
> >> >>> 
> >> >>> OK. 
> >> >>> 
> >> >>> Andrew.
> >> >>> 
> >> >>> >---
> >> >>> >Erik
> >> >>> >
> >> >>> >Andrew Beckett <[EMAIL PROTECTED]> wrote in message news:<[EMAIL 
> >> >>> >PROTECTED]>...
> >> >>> >> Take this up with customer support. Realistically the only person watching 
> >> >>> >> this
> >> >>> >> news group who would be able to answer this is me, and you shouldn't rely 
> >> >>> >> on
> >> >>> >> me to reply in a short time frame (especially as I'm on vacation at the 
> >> >>> >> moment...)
> >> >>> >> 
> >> >>> >> Andrew.
> >> >>> >> 
> >> >>> >> On 1 Nov 2003 10:25:22 -0800, [EMAIL PROTECTED] (Erik Wanta) wrote:
> >> >>> >> 
> >> >>> >> >I am able to get the netlist procedure to work in 50 33 when
> >> >>> >> >netlisting with CIW->Tools->AMS->Netlist.  I am UNable to get it to
> >> >>> >> >work from HED->AMS->Design Prep as I don't think the netlist procedure
> >> >>> >> >is not getting loaded.  I am told that there used to be an option
> >> >>> >> >called amsDirect initFile that could be placed in the ams.env to load
> >> >>> >> >a file prior to netlisting but it doesn't seem to work.  Any ideas on
> >> >>> >> >how to load the netlist procedure so that it gets used by
> >> >>> >> >HED->AMS->Design Prep?
> >> >>> >> >
> >> >>> >> >(defun AMSNLPROCDoNotPrintInstance (formatterId cellViewId instanceId)
> >> >>> >> >
> >> >>> >> >; do not print the instance, simply return
> >> >>> >> >
> >> >>> >> >t
> >> >>> >> >
> >> >>> >> >) ; defun
> >> >>> >> >
> >> >>> >> >---
> >> >>> >> >Erik
> >> >>> >> >
> >> >>> >> >
> >> >>> >> >
> >> >>> >> >
> >> >>> >> >[EMAIL PROTECTED] (Erik Wanta) wrote in message news:<[EMAIL 
> >> >>> >> >PROTECTED]>...
> >> >>> >> >> When I specify Proc1 as the netlist procedure in the ams simInfo I get
> >> >>> >> >> errors that Proc1 is not callable.  It returns t for
> >> >>> >> >> isCallable('Proc1) however.  Any ideas?
> >> >>> >> >> ---
> >> >>> >> >> Erik
> >> >>> >> >> 
> >> >>> >> >> 
> >> >>> >> >> [EMAIL PROTECTED] (Erik Wanta) wrote in message news:<[EMAIL 
> >> >>> >> >> PROTECTED]>...
> >> >>> >> >> > Andrew:
> >> >>> >> >> > I got the following in response to PCR 601739.  Am I supposed to use
> >> >>> >> >> > amsDirect initFile in ams.env or am I supposed to specify Proc1 as an
> >> >>> >> >> > ams netlist procedure.  I don't see any documentation on ams netlist
> >> >>> >> >> > procedures.  spectre netlist procedures just take inst, correct?  Any
> >> >>> >> >> > ideas would be appreciated.
> >> >>> >> >> > 
> >> >>> >> >> > ;; Get the netlister object
> >> >>> >> >> > Netlister = amsGetNetlister()
> >> >>> >> >> > 
> >> >>> >> >> > ;; Get the Verilog formatter object
> >> >>> >> >> > Vlog = Netlister->vlog
> >> >>> >> >> > 
> >> >>> >> >> > ;; This netlist procedure looks for a property "amsIgnore" on the
> >> >>> >> >> > ;; instance (of corrPoint) and if the value is ignore, does not
> >> >>> >> >> > netlist
> >> >>> >> >> > ;; the instance.
> >> >>> >> >> > 
> >> >>> >> >> > (defun Proc1 (formatter cellview inst)
> >> >>> >> >> >   ;; Make sure that the master is res
> >> >>> >> >> >   (if (equal inst->masterName "res")
> >> >>> >> >> >     ;; Check for property amsIgnore on the instance, netlist only
> >> >>> >> >> >     ;; if the value is not "ignore" ...
> >> >>> >> >> >     (unless (and inst->id->amsIgnore
> >> >>> >> >> >                  (equal inst->id->amsIgnore "ignore")
> >> >>> >> >> >                  )
> >> >>> >> >> >       ;; Not to be ignored ...
> >> >>> >> >> >       (amsPrintInstance formatter cellview inst)
> >> >>> >> >> >       ) ; unless
> >> >>> >> >> >     ;; The else part: master is not corrPoint
> >> >>> >> >> >     (amsPrintInstance formatter cellview inst)
> >> >>> >> >> >     ) ; if
> >> >>> >> >> >   ) ; defun
> >> >>> >> >> > 
> >> >>> >> >> > Vlog->instanceProc = 'Proc1
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > 
> >> >>> >> >> > Andrew Beckett <[EMAIL PROTECTED]> wrote in message news:<[EMAIL 
> >> >>> >> >> > PROTECTED]>...
> >> >>> >> >> > > Hi Erik,
> >> >>> >> >> > > 
> >> >>> >> >> > > Actually the problem here is really AMS Designer. This is something 
> >> >>> >> >> > > that has been
> >> >>> >> >> > > resolved in the IC5033 release by support for AMS Netlist 
> >> >>> >> >> > > Procedures - I have
> >> >>> >> >> > > a solution somewhere which explains how to do this (it was 
> >> >>> >> >> > > suggested by R&D
> >> >>> >> >> > > when I reported a similar problem).
> >> >>> >> >> > > 
> >> >>> >> >> > > Normally for most netlisters you wouldn't put nlAction=ignore on 
> >> >>> >> >> > > the symbol
> >> >>> >> >> > > unless you wanted it ignored in all tools - you'd put the ignore on
> >> >>> >> >> > > the stopping view. The trouble is with AMS Designer is that the 
> >> >>> >> >> > > netlisting of
> >> >>> >> >> > > each cellView is done independently, and so the switch list type 
> >> >>> >> >> > > info is
> >> >>> >> >> > > not used until elaboration time.
> >> >>> >> >> > > 
> >> >>> >> >> > > So you end up having to write a netlisting procedure to explicitly 
> >> >>> >> >> > > support
> >> >>> >> >> > > a special property to prevent netlisting in AMS Designer - and 
> >> >>> >> >> > > netlisting
> >> >>> >> >> > > procedures for AMS are only supported from IC5033 onwards.
> >> >>> >> >> > > 
> >> >>> >> >> > > As I'm out of the office on vacation until next week, I can't 
> >> >>> >> >> > > easily look this up,
> >> >>> >> >> > > but I'm sure somebody in support could find the PCR which asks for 
> >> >>> >> >> > > an
> >> >>> >> >> > > ignore - just found a record in my notes that it was PCR 601739.
> >> >>> >> >> > > 
> >> >>> >> >> > > Andrew.
> >> >>> >> >> > > 
> >> >>> >> >> > > 
> >> >>> >> >> > > On 30 Oct 2003 08:08:25 -0800, [EMAIL PROTECTED] (Erik Wanta) wrote:
> >> >>> >> >> > > 
> >> >>> >> >> > > >I am trying to get a component not to netlist in AMS Designer but
> >> >>> >> >> > > >still netlist for assura.  I added an nlAction ignore property to 
> >> >>> >> >> > > >the
> >> >>> >> >> > > >symbol and it doesn't netlist for AMSD but Assura doesn't netlist 
> >> >>> >> >> > > >it.
> >> >>> >> >> > > >
> >> >>> >> >> > > >I was able to get layoutXL to bring in the testpoints with
> >> >>> >> >> > > >envSetVal("layoutXL" "ignoreNames" `string "vxlIgnore lvsIgnore
> >> >>> >> >> > > >ignore").  That is I removed the nlAction from the ignore name 
> >> >>> >> >> > > >list. 
> >> >>> >> >> > > >I can't find the equivalent for assura.  I would think assura would
> >> >>> >> >> > > >use lvsIgnore instead of nlAction anyway.
> >> >>> >> >> > > >
> >> >>> >> >> > > >Does anyone know of a way to get Assura to not use the nlAction
> >> >>> >> >> > > >property?
> >> >>> >> >> > > >---
> >> >>> >> >> > > >Erik