IBIS Macromodel Task Group

Meeting date: 02 June 2020

Members (asterisk for those attending):
Achronix Semiconductor        Hansel Dsilva
ANSYS:                      * Curtis Clark
                            * Wei-hsing Huang
Cadence Design Systems:     * Ambrish Varma
                              Ken Willis
                              Jared James
Intel:                      * Michael Mirmak
Keysight Technologies:      * Fangyi Rao
                            * Radek Biernacki
                              Ming Yan
                              Todd Bermensolo
                              Stephen Slater
Marvell                       Steve Parker
Mentor, A Siemens Business: * Arpad Muranyi
Micron Technology:          * Randy Wolff
                            * Justin Butterfield
SiSoft (Mathworks):         * Walter Katz
                              Mike LaBonte
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:

- Bob to send BIRD201.1_draft7 to the ATM reflector.
  - Done.
  
--------------------------
Call for patent disclosure:

- None.

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

Arpad asked for any comments or corrections to the minutes of the May 26
meeting.  Michael moved to approve the minutes.  Curtis seconded the motion.
There were no objections.

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

BIRD201.1_draft7 discussion:
Bob shared and reviewed draft7.  He noted that he had added some new section
numbers and titles.  In section 10.9, Bob asked if the intention had been to
delete the entire first paragraph.  Michael recommended removing the first
sentence of that paragraph, which listed the parameter names and would have to
be updated every time a new related parameter was added.  He suggested moving
the remaining sentence elsewhere.  Radek said the second paragraph was only
about the statistical flow, and Curtis and Radek said it could be removed since
10.9 is about back-channel optimization in general.  Bob removed paragraph 2
and moved the remaining sentence from paragraph 1 to the end of paragraph 3.
Radek suggested changing "These Reserved Parameters" to "The following Reserved
Parameters" in that same sentence.  Bob made the change.

Bob noted that in the Allowed Formats: section he had added examples of Lists
with a single value, since these are syntactically legal.  Section 10.9.3
contains the clarifications made to address Radek's original comments.  In
10.9.4, item #4, Radek suggested "BCI simulation" be changed to "BCI training",
since the term "BCI simulation" had not been used before.  Bob made the change.

Bob noted that he had added BCI_Training_Mode to the various BCI training
parameter tables (33, 34, 35, 39).  He had also introduced references to
section numbers (e.g., 10.8.2) instead of page numbers.

Ambrish recalled that he had previously asked if we could see a presentation on
the results and flows provided by this BIRD.  He said it would give him more
confidence in the solution proposed by this BIRD.  Michael confirmed that there
is an industry need for this statistical flow optimization.  He said that IR
based simulations were heavily used for fast simulations across a wide range
of variations for DOE type statistical flows.  He said this resulted in huge
efficiencies in the cases where devices could be adequately modeled by LTI
assumptions.

Walter said that he could provide a demo presentation for a DDR5 channel in
which the Tx had an FFE and the Rx had a DFE, i.e., a typical DDR5 write.
He noted that during discussions about statistical flow optimization, Fangyi
had said that the training control could be in the EDA tool.  Walter agreed
this was possible but had argued that model makers want to model what their
actual hardware is doing in training.  For the purposes of developing the
proof of concept, Walter said he had followed Fangyi's idea.  He said his
demonstration was based on the EDA tool calling the Tx and Rx AMI_Init functions
repeatedly.  The actual flow does not yet make use of the new BIRD, but it is
equivalent to the BIRD supported scenario in which the Tx and Rx provide the
new AMI_Impulse function, and the training logic is in the AMI_Impulse function.
Ambrish said we didn't have to wait for the new BIRD to be approved to implement
it in an example.

Walter agreed to demo his current proof of concept flow at the next meeting.
Walter moved to submit the modified version (draft8) to the Open Forum.
Michael seconded.  There were no objections.  Bob took an AR to send the
modified version to the ATM reflector as draft8.  Randy took an AR to then
submit it to the Open Forum as BIRD201.1.

BIRD198.1 discussion:
Randy noted that he had replied to the authors and the small ATM working group
with the proposed modifications to BIRD198.1.  He had not yet heard back from
the authors.

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

AR: Bob to create bird201.1_draft8 with today's changes and email it to ATM.
AR: Randy to then submit the draft to the Open Forum as BIRD202.1.


-------------
Next meeting: 09 June 2020 12:00pm PT
-------------

IBIS Interconnect SPICE Wish List:

1) Simulator directives
