Hi Arpad,
I agree with most of the comments. Only wanted to focus on the File_** and
Number_of_terminals statements in draft9_v2.
------
As a point of clarification, the sentences below can be improved, but not in
the direction that you are suggesting.
The following sentences don't seem to be correct:
"For File_TS, the Number_of_terminals value shall be 2 or greater and equal
to the number of ports N plus 1 (N+1) in the Touchstone file.
For File_TS0, the Number_of_terminals value shall be 1 or greater and equal
to the number of ports N in the Touchstone file."
If we are allowing unused ports, the highlighted words should be changed to
"and less than or equal to the number of".
"For File_TS, the Number_of_terminals value shall be a value equal to N+1
(number of ports plus one) in the Touchstone file. Because a Touchstone file
requires at least one port, the Number_of_Terminals value shall be 2 or
greater.
For File_TS0, the Number_of_terminals value shall be a value equal to N
(number of ports) in the Touchstone file. Because a Touchstone file
requires at least one port, the Number_of_Terminals value shall be 2 or
greater."
.."
We are declaring within the [Interconnect Model] the size or size plus one
of the Touchstone file and not deducting unused terminals corresponding to
disconnected ports. We are also assuming that the Touchstone file has at
least one port. There is no format for a "0" port Touchstone file.
Also the File_IBIS-ISS rule should be modified in a parallel manner from:
For File_IBIS-ISS, the Number_of_terminals value shall be 1 or greater and
equal the number of subcircuit terminals for an IBIS-ISS subcircuit.
To:
For File_IBIS-ISS, the Number_of_terminals value shall be a value equal to
the number of subcircuit terminals in the IBIS-ISS subcircuit. Because an
IBIS-ISS subcircuit requires at least one terminal (node), the
Number_of_terminals value shall be 1 or greater.
Perhaps our terminology is slightly misleading. It is used both for actual
number of terminals in IBIS-ISS connections and maximum number to terminals
for Touchstone connections.
--------
I agree that this paragraph no longer applies and should be deleted.
"Model makers are recommended to ensure that each Interconnect Model Set
contains a complete description, through Interconnect Models, needed for the
path connecting the I/O buffers of interest to their associated pins, and
for connecting all rails related to these I/O buffers. This simplifies
choices to be made by the user or automatically by the EDA tool. It also
assures that the full electrical structure that is simulated matches what
the model provider intends. Some EDA tools may support selecting several
Interconnect Model Sets at once to form a complete path, but this requires
additional user interaction and may risk generating less-accurate simulation
data due to duplicate or missing content."
However, the first sentence should be moved to an Other Notes for
[Interconnect Model Set Group] and adapted such as suggested below.
However, it is difficult to make some statements or rules without bringing
in the terminology introduced later.
"Model makers should ensure that the [Interconnect Model Set Group] contains
all of the information needed for the selected pins of interest within that
group. This means that the complete path for I/O pins of interest should
have connections to their buffers. This also means that any PDN (power
distribution network) structure should have complete connections from the
Pin interface to the buffer interface.
Each [Interconnect Model Set Group] is independent of any other
[Interconnect Model Set Group] and intended only for simulations for the I/O
pin or set of pins within that group. There shall be no duplication of I/O
pins or PDN structures within that group.
---
We do not want to encourage or even support combining any [Interconnect
Model Set Group] with any lower level package model. That introduces too
many issues.
Bob
From: ibis-interconn-bounce@xxxxxxxxxxxxx
[mailto:ibis-interconn-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Monday, October 2, 2017 11:42 AM
To: ibis-interconn@xxxxxxxxxxxxx; IBIS-ATM
Subject: [ibis-interconn] Re: BIRD189_draft9_v2
Bob,
There are a couple of "]" characters with strike-through font, which should
be
changed to normal font. I would like to remove the blue and red colors too,
because they were supposed to indicate older changes which are not relevant
anymore. It is just too confusing to see all these colors when we are
talking
about different things.
This sentence appears twice, one of it should be deleted:
"A [Component] may have zero or more [Interconnect Model Set Group] keywords
(identified by a name) associated with it."
I think "one or more than one" could be just written as "one or more".
The last sentence in the precedence rules still needs work:
"[Package Model] and [Interconnect Model Set Group] may both be present for
the same [Component] but should not both be used at the same time."
The problem is around "same time". I think we would not have any
problems to use legacy and BIRD189 models in the same simulations
(at the same time) if they cover different pins. We just don't want
any overlap (at the same time). Then come the related questions on
what exactly is an overlap when the [Define Package Model] uses the
coupled matrix format, etc.
The following two sentences say pretty much the same thing, and I
think one of them could be removed (or they could merged together):
"Interconnect models may be supplied separately for on-die interconnect and
package interconnect, or may be supplied as a single model for the entire
connection between the package pins and buffers."
"An Interconnect Model may describe the connection between the pins of the
package and the buffers, the pins of the package and the die pads, or the
die pads and buffers."
If we want the user to make selection in one place, i.e. the [Interconnect
Model Set Group], then we will need to consider changing this statement:
"One Interconnect Model Set may contain groups of similar interconnect
models or different interconnect models to describe the complete connections
from the buffer to pin interface."
The highlighted words above imply that the user will have to make a
selection
from the similar interconnect models.
The following sentence doesn't make a lot of sense to me:
"In such a model, the crosstalk on any one route in the model could only be
caused by its nearest neighbors."
This comes after stating "This example assumes that only a few routes out of
a larger bus are shown.",
but what if those few routes do not even include any of the "nearest
neighbors"
that the next sentence is talking about? The paragraphs that follows this
is good,
but this sentence could possibly be deleted. It doesn't seem to add much to
the
discussion.
These sentences touch on the precedence rules again:
"In addition, within that .ibs file, they override any interconnect package
models defined using the [Package], [Pin], or [Define Package Model]
keywords. Interconnect Models in separate .ims files referenced by the
[Interconnect Model Set Group] keyword in a .ibs file also override any
interconnect package models defined in the same .ibs file using the
[Package], [Pin], or [Define Package Model] keywords."
The problem is with override [Define Package Model]. Is this happening
regardless
of what pins the [Define Package Model] is associated with, or does this
apply only
if the [Define Package Model] and the BIRD189 model is touching the same
pins?
I am still puzzled by the following paragraph, especially with the changes
we made
with the [Interconnect Model Set Group] keyword. Do we still need/want this
in
here:
"Model makers are recommended to ensure that each Interconnect Model Set
contains a complete description, through Interconnect Models, needed for the
path connecting the I/O buffers of interest to their associated pins, and
for connecting all rails related to these I/O buffers. This simplifies
choices to be made by the user or automatically by the EDA tool. It also
assures that the full electrical structure that is simulated matches what
the model provider intends. Some EDA tools may support selecting several
Interconnect Model Sets at once to form a complete path, but this requires
additional user interaction and may risk generating less-accurate simulation
data due to duplicate or missing content."
I wonder about the uniqueness requirements for the [Interconnect Model]'s
name argument. It says:
"The [Interconnect Model]/[End Interconnect Model] keyword pair is
hierarchically scoped by the [Interconnect Model Set]/[End Interconnect
Model Set] keywords."
Does this mean that I can have two different [Interconnect Model]s with the
same name (for example, "Coupled") in two different [Interconnect Model
Set]s?
Or does every single [Interconnect Model] have to have unique name?
The following sentences don't seem to be correct:
"For File_TS, the Number_of_terminals value shall be 2 or greater and equal
to the number of ports N plus 1 (N+1) in the Touchstone file.
For File_TS0, the Number_of_terminals value shall be 1 or greater and equal
to the number of ports N in the Touchstone file."
If we are allowing unused ports, the highlighted words should be changed to
"and less than or equal to the number of".
I would still recommend to remove the word "reference" from all of the
IBIS-ISS examples in the comments. I don't want to give any special meaning
to any of the subcircuit terminals, unlike the Touchstone reference
terminal,
IBIS-ISS terminals have no such special meanings.
I know a lot of these comments are a repeat from my previous comments,
but I don't want them to get lost while we keep distributing new drafts of
the document, thinking that my comments have been addressed in the new
draft.
Thanks,
Arpad
=============================================================
From: ibis-interconn-bounce@xxxxxxxxxxxxx
[mailto:ibis-interconn-bounce@xxxxxxxxxxxxx] On Behalf Of Bob Ross
Sent: Wednesday, September 27, 2017 11:46 AM
To: 'Mike LaBonte' <mlabonte@xxxxxxxxxx>; ibis-interconn@xxxxxxxxxxxxx;
IBIS-ATM <ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-interconn] BIRD189_draft9_v2
Mike, and group,
For posting on the Interconnect Group work archives.
Attached BIRD189.5_draft9_v2 is an editorial update per the discussion at
the 9/27 meeting relating to the File_IBIS-ISS, File_TS, and File_TS0
additions and to the Number_of_terminals rules.
I made a few other minor editorial changes.
On page 22, the unused port termination rules have not been changed. They
will be discussed at the ATM meeting, So this document is also sent to the
ATM reflector.
Bob
--
Bob Ross
Teraspeed Labs
www.teraspeedlabs.com <http://www.teraspeedlabs.com/>
bob@xxxxxxxxxxxxxxxxx
Direct: 503-246-8048
Office: 971-279-5325
