[ibis-macro] Re: IBIS AMI Specification Questions/Proposal
- From: "Walter Katz" <wkatz@xxxxxxxxxx>
- To: <Arpad_Muranyi@xxxxxxxxxx>, "IBIS-ATM" <ibis-macro@xxxxxxxxxxxxx>
- Date: Tue, 25 Aug 2009 09:02:09 -0400
Arpad,
You are correct. The current 5.0 is confusing on Default and Format, and
Ambrish suggests on e solution for Format but does not clarify how to use
Default.
I personally would much rather deprecate Format, and make Value, Range,
List, Corner, Increment, and Steps reserved parameters. Ambrish suggests:
| (Format <data format>) (Value <values>) (Default <values>)
I much prefer
| (<format> <values>) (Default <values>)
Because of the inconsistent definition of Format in the IBIS 5.0, IC Vendors
have been all over the map in generating .ami files, and also in generating
the strings that get passed into the Init call. We have solved this problem
by simply always stripping off Format from the .ami files when we read them
in. Just removing Format is what our customers have recommended to us.
And to answer your specific concern about Default, I suggest the following
language:
| Default <value>:
| Depending on the Type and <format>, <value> will provide a default
value for the
| parameter. For example, if the Type is Boolean, <value> could be True
| or False, if the Type is Integer, the <value> can be an integer value.
| Default is not allowed if <format> is Corner. Computed is a reserved
value for
| Default. If {Default Computed), then the value of this parameter is
computed
| from the values of other parameters by the Computation Engine. If
Default is not
| specified, then the default value of a parameter shall depend on the
<format> as
| described below
| Value: (Only Computed is allowed)
| Range: <typ value>
| List: <value1>
| Label: NA
| Corner: NA
| Increment: <typ value >
| Steps: <typ value >
Note that Label and Computation Engine are enhancements to AMI that we were
planning to introduce shortly.
Note that the answer to the formulation of the string passed to Init is now
very simple. A "leaf" is either a parameter or a reserved keyword. Only
leaves that are "parameters" and In or InOut are passed to the DLL, and the
leaves are always passed in the form: (<parameter> <value>). <value> is
determined by the EDA tool. If the <format> is Value, then there can only be
one value that the EDA tool can pass. If the <format> is Corner, then the
process Corner determines <value> selected for the simulation. Otherwise,
the EDA tool can choose any mechanism for determining <value> based in its
allowed values determined by <format> being Range, List, Increment or Steps.
Note that there is no <format> that allows more than one <value>. Gaussian,
Table, Dual-Dirac and DjRj that are only used for Reserved parameters
Clock_PDF and Tx_Jitter, and were not well thought out when we wrote the
original AMI specification, and in fact are not useful and just add to
confusion on using AMI. We are planning on correcting this in our proposed
changes to AMI.
In what I am planning to propose, the new reserved leaf Vector can be used
combine the other sibling leaves into a list of values of the parent leaf.
This is the way that the following would occur (<parameter> <value> <value>
<value> <value>). Using Vector has two applications. The first application
is Taps:
Without Vector
(Tap (-1 .1) (0 .8) (1 -.1))
With Vector
(Tap .1 .8 -.1)
Similarly, for clock PDF, we have found that IC vendors need to control the
mean and standard deviation of the clock independently. Our current
formulation of clock PDF does not allow the IC Vendors to do this. That is
why we will offer new reserved parameters for Clock_PDF_Mean and
Clock_PDF_Sigma. This can be done in either of the two following forms:in
the ami file:
(Clock_PDF (Mean (Usage In) (Range 0. -.5UI +.5UI)) (Sigma (Usage In)
(Range 0. -.3UI +.3UI)) )
(Clock_PDF (Mean (Usage In) (Range 0. -.5UI +.5UI)) (Sigma (Usage In)
(Range .1UI. 0. .3UI)) (Vector (Value True)) )
The strings passed to the DLL would be (assuming UI is 1.e-10, and using
Default values):
(Clock_PDF (Mean 0.) (Sigma .1e-11))
(Clock_PDF 0. 1e-11)
.As I have said in previous meetings, all of these deficiencies in the IBIS
5.0 are interrelated, and need to be fixed in a unified way,
Walter
Walter Katz
303.449-2308
Mobile 720.333-1107
wkatz@xxxxxxxxxx
www.sisoft.com
-----Original Message-----
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx]On Behalf Of Muranyi, Arpad
Sent: Tuesday, August 25, 2009 12:17 AM
To: IBIS-ATM
Subject: [ibis-macro] Re: IBIS AMI Specification Questions/Proposal
Ambrish,
Thanks for pointing this problem out. I am sure we will address
these issues as our discussions on AMI improvements continue.
I would like to ask you a couple of questions about the usage of
"default".
1) Considering your statement that "we have used the word Default
to specify the value for that model/simulation", what would you say
is the difference between the typical value of a range and the default?
2) Your statement "we have used the word Default to specify the
value for that model/simulation" opens up the question I asked a
several weeks ago about parameter strings. Remember, my impression
about the specification was that the entire .ami file was supposed
to be passed into the DLL's parameter string and consequently
the DLL was supposed to parse it and extract the values from it.
Your statement that "we have used the word Default to specify the
value for that model/simulation" seems to underline that thinking,
because it seems that you are saying that the DLL will extract the
default value and simulate using that. In this case, the EDA tool
could modify the original .ami file's default values before passing
it to the DLL according to the user's wishes for each specific
simulation.
On the other hand, if this was not the intent, in other words, if
the DLL is supposed to get only a subset of the .ami file in the
parameter string, I am still looking for a description in the AMI
specification that explains how to formulate that subset.
Thanks,
Arpad
======================================================================
-----Original Message-----
From: Ambrish Varma [mailto:ambrishv@xxxxxxxxxxx]
Sent: Monday, August 24, 2009 3:03 PM
To: IBIS-ATM
Cc: Muranyi, Arpad
Subject: RE: [ibis-macro] Re: IBIS AMI Specification Questions/Proposal
Hi All,
I feel there is some real confusion about "Values" and "Default".
Defaults, generally mean that if there is no value provided, the tool
will use the default value. In the spec, we have used the word Default
to specify the value for that model/simulation (particularly in the
reserved parameters section).
There is also confusion because we have defined the format for declaring
the parameters at 2 places for the reserved parameters (one with format,
one without).
Please refer to the attached document to see the changes that I am
proposing to the parameter section.
Thanks,
-Ambrish.
-----Original Message-----
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Tuesday, August 18, 2009 6:51 PM
To: IBIS-ATM
Subject: [ibis-macro] Re: IBIS AMI Specification Questions/Proposal
Bob,
Thanks for posting this summary of the problem.
I read trough the attached file and I am not sure what
the reasons are for these inconsistencies. In order to
make the right decisions, I feel we need to ask those
who wrote these portions of the AMI spec to find out
what was their intent. Authors, please speak up...
Thanks,
Arpad
========================================================
-----Original Message-----
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Bob Ross
Sent: Tuesday, August 18, 2009 5:11 PM
To: IBIS-ATM
Subject: [ibis-macro] IBIS AMI Specification Questions/Proposal
All:
This is being sent out per the issue raised at the Aug. 18
meeting concerning parser interpretation of the AMI Spec.
The ibischk5 parser with the -ami flag will check the
syntax of the <file name>.ami file. The attached document
shows several forms for <parameter_name> that are in the
specification. Some of the forms imply a restricted set of
parameters, but then this is contradicted by the example
at the end.
This implies that some parameters are really optional and
not excluded according to the syntaxes given.
The propposal is simply to make all parameters at least
optional, and some required according to a draft Table 4
that is NOT in the specification.
Do the choices in Table 4 look reasonable?
Bob
--
Bob Ross
Teraspeed Consulting Group LLC Teraspeed Labs
121 North River Drive 13610 SW Harness Lane
Narragansett, RI 02882 Beaverton, OR 97008
401-284-1827 503-430-1065
http://www.teraspeed.com 503-246-8048 Direct
bob@xxxxxxxxxxxxx
Teraspeed is a registered service mark of Teraspeed Consulting Group LLC
---------------------------------------------------------------------
IBIS Macro website : http://www.eda.org/pub/ibis/macromodel_wip/
IBIS Macro reflector: //www.freelists.org/list/ibis-macro
To unsubscribe send an email:
To: ibis-macro-request@xxxxxxxxxxxxx
Subject: unsubscribe
---------------------------------------------------------------------
IBIS Macro website : http://www.eda.org/pub/ibis/macromodel_wip/
IBIS Macro reflector: //www.freelists.org/list/ibis-macro
To unsubscribe send an email:
To: ibis-macro-request@xxxxxxxxxxxxx
Subject: unsubscribe
---------------------------------------------------------------------
IBIS Macro website : http://www.eda.org/pub/ibis/macromodel_wip/
IBIS Macro reflector: //www.freelists.org/list/ibis-macro
To unsubscribe send an email:
To: ibis-macro-request@xxxxxxxxxxxxx
Subject: unsubscribe
Other related posts:
