IBIS Macromodel Task Group

Meeting date: 14 February 2023

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

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

--------------------------------------------------------------------------------
Opens:  Arpad noted that he had added a new item to the agenda email:

  9)  AMI_GetWave block size with continually adapting models
	  
Arpad said he had asked about this topic in response to a DesignCon IBIS Summit
presentation involving an adapting Rx AMI model.  The specification currently
says a model should work with any block size.  The potential issue is that the
block size controls the rate at which adaptation results can be returned in
AMI_parameters_out.  Adaptation results could appear with an artificial lag if
the model settles faster than a single block of data, for example.  Is this an
issue we need to address?

Randy, Michael and Kinger reported that attendance at DesignCon and the IBIS
Summit seemed to be back to pre-pandemic levels.  The show floor was full, and
many presentations including the keynote presentation were full.  IBIS's 30th
anniversary was celebrated at IBIS Summit.

-------------
Review of ARs:
  
- Kinger to send out draft8 of the SPIM BIRD containing changes reviewed and
  discussed in the previous meeting.
  - Done.
  
- Kinger to send out draft5 of the PSIJ Sensitivity BIRD containing changes
  incorporated from SPIM BIRD draft feedback.
  - Done.

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

- None.

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

Arpad asked for any comments or corrections to the minutes of the January 24th
meeting.  Randy moved to approve the minutes.  Dian seconded the motion.
There were no objections.

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

Standard Power Integrity Model (SPIM) BIRD draft:
Kinger reported that he had received feedback from Chi-te and Arpad in response
to the draft8 he had sent out.  He reviewed a work-in-progress draft 9 in which
he was incorporating their changes.

The Solution Requirements section was being rearranged in table form, per
Arpad's suggestion.  Arpad said the goal of Solution Requirements is to keep us
focused on the problem being addressed.  Randy agreed that the table form was
important because this BIRD should adhere to the BIRD template.

The BIRD is now relative to the newly approved IBIS 7.2.

The sum of the weighting values of all stimulus ports equaling one was further
explained to mean that the total current is normalized to 1.0A for AC impedance
analysis.  Kinger said for AC impedance analysis we normalize the overall
current to 1.0A and distribute it according to the weighting values.  Then,
measuring the voltage at the observation port yields the impedance directly,
because the total current is 1.0A.  In the future, for transient analysis, the
weighting values may be used as scaling multipliers to distribute the actual
total current.

The use of "Ball Grid Array (BGA) pins" was simplified to simply "pins" or
"balls".

The text now incorporates Arpad's suggestion to clarify that the N+M port model
refers to the number of ports remaining after the VRM ports have been
connected.

Kinger revisited the section describing holistic platform PDN design and
considering 3 major segments: on-die, package, and board.  He said the SPIM
modeling and Unified PI Target support all 3, but chip vendors will typically
only provide target information for the package and board regions, which extend
from DC to about 20MHz.  This allows them to distribute models to platform
vendors and protect their IP by not providing on-die target information.

The last paragraph on page 6 started with, "For supporting DC analysis in the
frequency domain,".  Arpad and Randy said DC is really its own analysis type,
and whether it might be an initial DC for a transient or frequency domain
analysis is probably not important.  Kinger agreed and removed "in the
frequency domain."

In that same paragraph, Kinger noted that language about terminals had caused
some confusion.  He said the point that was intended was that a PDN model
provided in resistance network format to support DC IR analysis would contain
terminals for every power and ground pin.  For the S-parameter models provided
for AC impedance analysis, however, power and ground pins would typically be
grouped together into ports utilizing local ground references.  The model maker
would determine the port count of the S-parameter model based on a tradeoff
between performance and accuracy.

As a general comment, Arpad and Randy suggested that the Description sections of
multiple keywords contained text that was better suited for Usage Rules.  Randy
said the Description should typically contain only one or two sentences.  Kinger
agreed and said the Description could be confined to a high-level definition and
formatting, and the rest could move to Usage Rules.

Arpad had an overall comment about end keywords, e.g., [End Device SPIM Group].
He said the language in the Required section wasn't quite sufficient, e.g. :
  Required: No. But it is required if a [Device SPIM Group] keyword is present.
Arpad said end keywords are required to exist to match corresponding instances
of start keywords, but they are illegal in any other situation.  The existing
language doesn't make the latter point.  Randy said we should adopt whatever
language is used for existing end keywords.  For [End EMD Group] we use:
  Required:  Yes, for each instance of the [EMD Group] keyword
though the specification's language is not consistent across all keywords.
Michael said Required was originally intended to mean that the keyword must
exist in any valid IBIS file.  He said we had never really standardized on the
way to deal with dependency issues.  Randy and Michael said we could leave this
BIRD's language as is, and the Editorial task group can take up the overall
issue throughout IBIS in the next version of the specification.

Kinger said he would incorporate the changes discussed, wait a few days for any
additional feedback on draft8, and then send out draft9 later in the week.
Arpad again asked everyone to review the draft.

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

AR: Kinger to send out draft9 of the SPIM BIRD containing changes reviewed and
    discussed in today's meeting.
-------------
Next meeting: 21 February 2023 12:00pm PT
-------------

IBIS Interconnect SPICE Wish List:

1) Simulator directives
