IBIS Macromodel Task Group

Meeting date: 12 September 2023

Members (asterisk for those attending):
Achronix Semiconductor:       Hansel Dsilva
Amazon:                       John Yan
ANSYS:                      * Curtis Clark
                            * Wei-hsing Huang
Aurora System:              * Dian Yang
Cadence Design Systems:     * Ambrish Varma
                              Jared James
Google:                       Hanfeng Wang
                              GaWon Kim
Intel:                      * Michael Mirmak
                            * Kinger Cai
                              Chi-te Chen
                              Liwei Zhao
Keysight Technologies:        Fangyi Rao
                              Majid Ahadi Dolatsara
                              Stephen Slater
                              Ming Yan
                              Rui Yang
Marvell:                      Steve Parker
Mathworks (SiSoft):           Walter Katz
                              Graham Kus
Micron Technology:            Justin Butterfield
Missouri S&T:                 Chulsoon Hwang
                              Yifan Ding
                              Zhiping Yang
Rivos:                        Yansheng Wang
SAE ITC:                      Michael McNair
Siemens EDA (Mentor):       * Arpad Muranyi
                            * Randy Wolff
Teraspeed Labs:             * Bob Ross
Zuken USA:                  * Lance Wang

The meeting was led by Arpad Muranyi.  Curtis Clark took the minutes.

--------------------------------------------------------------------------------
Opens:

- None.

-------------
Review of ARs:

Kinger: Send out draft1 of BIRD223.1 with the changes and updates noted during
        the ATM meeting.
        - Done.

Michael: Send out draft7 of the [AMI Test Data] proposal including more feedback
         from Fangyi and the ATM meeting.
         - Done.

--------------------------
Call for patent disclosure:

- None.

-------------------------
Review of Meeting Minutes:

Arpad asked for any comments or corrections to the minutes of the August 29th
meeting.  Michael moved to approve the minutes.  Ambrish seconded the motion.
There were no objections.

--------------
New Discussion:

BIRD223.1:
Kinger reviewed the changes in draft1 of BIRD223.1.  These were enumerated in
the email he had sent with draft1 (also see ATM minutes from August 29, 2023).
He noted that he had decided to change the S in SPIM from "standard" to
"streamlined".  He also noted that [Interconnect Model Group] had been removed
from the proposed changes to the .ibs parameter tree.  He said it had originally
been included to indicate where the new [Device SPIM Group] parameter should be
added in the tree, but it had caused confusion and [Interconnect Model Group]
might not even exist in each .ibs file.

Kinger said that he was already working on draft2 and incorporating feedback Bob
had sent in reply to draft1.  He had updated the various date fields in the
header per Bob's suggestion, and he said he would add the enumerated list of
changes in BIRD223.1 to the Background Information section.  Bob requested that
Kinger remove the italics from equations 1-7 to make them easier to read.
Kinger agreed.

Kinger said he would send draft2 to the ATM list and plan to submit it to the
IBIS Open Forum prior to its next meeting on Friday, September 15th.  Arpad
asked everyone to review draft2 and provide any feedback prior to Friday.

[AMI Test Data] proposal:
Michael said the changes in draft7 were relatively minor and captured the
feedback from the previous ATM meeting and Fangyi's email reply to draft6.

Michael moved on to discussion of the suggestions Ambrish had provided in an
email reply to draft7.  Ambrish said he thought draft7 was still a bit difficult
to follow because of the cross referencing between statistical and time domain
Sub-params.  He said that requiring even a purely time domain model to use the
Statistical_configuration Sub-param seemed convoluted and confusing.  He
proposed a new Impulse_matrix_file Sub-param, which would contain the impulse
response passed into AMI_Init() regardless of the Type of the simulation.  This
would eliminate the need for overloading the Stimulus_file Sub-param to provide
an impulse response in the statistical case.  It would also eliminate the need
for the Statistical_configuration Sub-param, which had been introduced solely to
provide the impulse response to AMI_Init() for a time domain simulation.

Michael said his recollection was that the initial intent of AMI_Init() had been
initialization of the model, and the modification of the impulse response to
support statistical simulation had been added later.  (Michael noted that others
did not necessarily agree with his recollection of the evolution of AMI_Init()).
Michael asked if Ambrish was trying to go back to that original intent and
isolate the difference between statistical flow and the initialization process
required for the time domain flow.  Ambrish agreed that he wanted to keep that
distinction clear, and he said he thought we had maintained that distinction
in the specification.  He said the AMI flows carefully note that a model can
support statistical, or time domain, or both.

Michael asked for confirmation that Ambrish's proposed changes seeking to
decouple the statistical and time domain Sub-params would still support
models for which the time domain flow depended on what was passed to AMI_Init().
Ambrish said his changes did not affect this at all.  He said models that
adapt in AMI_Init() and then use the adaptation as a starting point for time
domain simulations maintain all that information internal to the model.

Ambrish noted that we would no longer need the Stimulus_file Sub-param when the
Type is Statistical because the Impulse_matrix_file is used.  In addition, we
no longer would need the extra Statistical_configuration Sub-param in time domain
simulations because the Impulse_matrix_file provides the IR passed to AMI_Init().

Bob asked about corners and the fact that "typ" appeared in the names of many
Sub-params in the examples.  Michael said that we had previously decided that
corners were not really relevant to these keywords and Sub-params.  In this
proposal we are naming a specific model and providing the test inputs and the
expected outputs.  We want the model maker to provide filenames that are
descriptive.

Michael said he would adopt the other two changes Ambrish had proposed in his
email response.  One was editorial in nature, and the other was to change the
Sub-param Waveform_file to Golden_waveform_file to make its intent more clear.

Ambrish again noted that params such as sample_interval, which implicitly rely
on the descriptions of the AMI_Init() and AMI_GetWave() function signatures,
should not appear until after the function definitions are introduced.

Michael said he would send out a draft8 incorporating the changes discussed in
the meeting.

- Curtis: Motion to adjourn.
- Ambrish: Second.
- Arpad: Thank you all for joining.

New ARs:

Kinger: Send out draft2 of BIRD223.1 with the changes and updates noted during
        the ATM meeting.

Michael: Send out draft8 of the [AMI Test Data] proposal including more feedback
         from Ambrish and the ATM meeting.

-------------
Next meeting: 19 September 2023 12:00pm PT
-------------

IBIS Interconnect SPICE Wish List:

1) Simulator directives
