IBIS Macromodel Task Group

Meeting date: 14 November 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
                            * Alaeddin Aydiner
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:

Curtis noted that the meeting scheduled for November 7th had been cancelled.

Arpad reviewed the schedule of upcoming meetings.  The group decided to cancel
the following meetings:
  November 21, 2023
  December 26, 2023
  January  02, 2024
  
-------------
Review of ARs:

Kinger: Prepare some slides to aid in the discussion of BIRD220 and BIRD226 and
        their interaction
        - Done.  See discussion below.
        
Michael: Ask Alaeddin to submit his Ignore Block proposal to the Open Forum as
         an official BIRD
         - Done.  Alaeddin said he would submit his proposal to the Open Forum.

Michael: Send out draft15 of the AMI Test Data proposal including more
         feedback from the ATM meeting.
         - Done.
         
Michael: Develop a full syntactically complete example demonstrating the
         [AMI Test Data] proposal.
         - In progress.

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

- None.

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

Arpad asked for any comments or corrections to the minutes of the October 31st
meeting.  Michael moved to approve the minutes.  Kinger seconded the motion.
There were no objections.

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

BIRD220 and BIRD226:
Kinger reviewed his slide presentation on the interaction between BIRD220 and
BIRD226, which he had sent to the ATM list prior to the meeting.  He noted that
the slides were not intended to go into details about the individual BIRDs.
Instead, his goal was to explore high level scenarios and what the two BIRDs
were able to handle.

Kinger said that platform PDN design is complicated.  There may be multiple
power rails for a given interface, and a single rail may provide power to
multiple buffers.  It is common to attempt to consolidate power rails for cost
tradeoff reasons.  He said it is necessary at a system level to consider the
PSIJ contributions from all rails.

slide 3 - BIRD220
Kinger said that BIRD220 is applicable to a scenario with a single power rail
providing power to a single buffer.  He asked whether it could handle one rail
to many buffers, or many rails to one buffer.  He also asked whether BIRD220
would supersede the [ISSO PU] and [ISSO PD] keywords and whether there would
be double counting of effects if both were used.

Randy said BIRD220 could handle the one rail to many buffers scenario.  If your
PDN model allows you to see the rail voltage output at each buffer, then you
could instantiate multiple instances of the buffer [Model] to connect to them.

Arpad addressed the multiple rails to one buffer scenario.  He said that it is
a fundamental limit of the existing IBIS [Model] keyword that it only considers
one voltage rail.  He said that with [Model], which is what BIRD220 enhances,
it would be difficult to address the case where a buffer requires multiple
power rails.  He noted, however, that [External Circuit] is not limited in terms
of how many terminals it can use.  With an [External Circuit], separate supply
rail terminals for multiple Vdd rails could be provided.  One could then model
IBIS buffers with more than just an output stage supply rail.

Randy, Arpad and Yifan confirmed that BIRD220 and the [ISSO PU] and [ISSO PD]
keywords are complementary and do not result in double counting.  Both are
necessary to improving the correlation between an IBIS buffer [Model]
simulation and a full transistor level SPICE simulation.  The ISSO keywords
characterize the final stage and modulate the drain to source current based on
the VDDQ voltage.  They modulate the drive strength and therefore the edge rate.
BIRD220 characterizes the delay sensitivity in the predriver and shifts the
location of the edge in response to changes in VDD.

Arpad asked whether BIRD220 and BIRD226 would result in double counting of PSIJ
effects.  If an IBIS [Model] containing BIRD220 information were used as part of
a BIRD226 type system simulation, would there be double counting?  Kinger said
there would not be double counting.  Because BIRD226 envisions an initial SI
simulation done with ideal power, there would be no variation in VDDQ, and
BIRD220 would have no effect.  Kinger said he thought BIRD220 and BIRD226 were
complementary and could co-exist without fear of double counting.

Arpad and Randy said their team had started reviewing BIRD220 with an eye toward
implementation.  They said they had some questions from the development team
including:
- How do BIRD220's K(t) equations interact with the ISSO keywords?
- Is there a simpler approach involving shifting the original K(t) as opposed to
  the new A and B terms introduced into the equation in BIRD220?

Randy said they would send Chulsoon and Yifan detailed questions offline, and he
and Arpad asked other EDA vendors to review BIRD220 with implementation in mind.
Chulsoon and Yifan said they would be happy to discuss any feedback on BIRD220.

Arpad noted that the IBIS Open Forum had asked this group to review BIRD220 and
BIRD226 to ascertain their readiness for a vote.  Arpad said he thought BIRD226
might be ready for a vote.  Arpad suggested that we hold off on proposing a vote
on BIRD220 in the Open Forum until we have had a chance to discuss the new
feedback/questions.

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

New ARs:

- None.

-------------
Next meeting: 28 November 2023 12:00pm PT
-------------

IBIS Interconnect SPICE Wish List:

1) Simulator directives
