IBIS Macromodel Task Group

Meeting date: 29 June 2010

Members (asterisk for those attending):
  Adge Hawes, IBM
* Ambrish Varma, Cadence Design Systems
* Anders Ekholm, Ericsson
* Arpad Muranyi, Mentor Graphics Corp.
  Barry Katz, SiSoft
* Bob Ross, Teraspeed Consulting Group
  Brad Brim, Sigrity
  Brad Griffin, Cadence Design Systems
  Chris Herrick, Ansoft
  Chris McGrath, Synopsys
  Danil Kirsanov, Ansoft
  David Banas, Xilinx
  Deepak Ramaswany, Ansoft
  Donald Telian, consultant
  Doug White, Cisco Systems
  Eckhard Lenski, Nokia-Siemens Networks
  Eckhard Miersch, Sigrity
  Essaid Bensoudane, ST Microelectronics
  Fangyi Rao, Agilent
  Ganesh Narayanaswamy, ST Micro
  Gang Kang, Sigrity
  Hemant Shah, Cadence Design Systems
  Ian Dodd, consultant
  Jerry Chuang, Xilinx
  Joe Abler, IBM
* John Angulo, Mentor Graphics
  John Shields, Mentor Graphics
* Ken Willis, Sigrity
  Kellee Crisafulli, Celsionix
  Kumar Keshavan, Sigrity
  Lance Wang, Cadence Design Systems
  Luis Boluna, Cisco Systems
  Michael Mirmak, Intel Corp.
* Mike LaBonte, Cisco Systems
  Mike Steinberger, SiSoft
  Mustansir Fanaswalla, Xilinx
  Patrick O'Halloran, Tiburon Design Automation
  Paul Fernando, NCSU
  Pavani Jella, TI
  Radek Biernacki, Agilent (EESof)
* Randy Wolff, Micron Technology
  Ray Komow, Cadence Design Systems
  Richard Mellitz, Intel
  Richard Ward, Texas Instruments
  Samuel Mertens, Ansoft
  Sam Chitwood, Sigrity
  Sanjeev Gupta, Agilent
  Scott McMorrow, Teraspeed Consulting Group
  Shangli Wu, Cadence Design Systems
  Sid Singh, Extreme Networks
  Stephen Scearce, Cisco Systems
  Steve Kaufer, Mentor Graphics
  Steve Pytel, Ansoft
  Syed Huq, Cisco Systems
  Syed Sadeghi, ST Micro
  Ted Mido, Synopsys
  Terry Jernberg, Cadence Design Systems
* Todd Westerhoff, SiSoft
  Vladimir Dmitriev-Zdorov, Mentor Graphics
  Vikas Gupta, Xilinx
  Vuk Borich, Agilent
* Walter Katz, SiSoft
  Wenyi Jin, LSI Logic
  Zhen Mu, Mentor Graphics

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

- None

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

- No one declared a patent.

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

- Arpad: Update AMI Flows and AMI BIRD text
  - Done

- Arpad:  Update section 2 text based on updated AMI flow
  - In progress

- Arpad:  Write parameter passing syntax proposal (BIRD draft)
          for *-AMS models in IBIS that is consistent with the
          parameter passing syntax of the AMI models

- TBD:    Propose a parameter passing syntax for the SPICE
          - [External ...] also?
          - TBD

- Arpad:  Review the documentation (annotation) in the macro libraries.
          - Deferred until a demand arises or we have nothing else to do

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

Arpad showed a list of changes to IBIS-ISS draft 0.3 mailed by Michael M:
- Arpad: He is asking for feedback
- Bob: This looks like good work
  - There is a lot to look at
- Arpad: There are still no dangling page refs and comments
- Walter: There was an unresolved email comment
  - We are not supporting hierarchical parameters?
- Arpad: There was a question about reserved title lines
  - It does not apply because this is about SPICE fragments
- Walter: HSPICE has a .title statement
- Arpad: The title line can be problematic
- Walter: It could be useful to have standard comments
  - There was also a recommendation to avoid hierarchical parameters
  - Subcircuits can receive parameters and that's all
- Arpad: There should be no global parameters
- Walter: Yes, that was it
- Mike L: It would be good to have comments stay with subcircuits
- Walter: At least ones that start with *ISS
- Arpad: The question about a version keyword is interesting
  - We might never deprecate keywords and not need one
- Bob: Does HSPICE have one?
- Arpad: No global version
  - There is a keyword to say how to round PWL tables, for example
- Walter: Tried to use the document to figure out table W-line model
  - Unable to do it by reading this, or the HSPICE manual
  - We may need to add some examples
  - ICM describes it well
- Arpad: I find no reference to string parameters
- Walter: We may have started with an older version
- Mike: We discussed this Jun 2009 and reached no conclusion

Arpad showed the updated AMI Flow BIRD text PDF:
- Arpad: Anders requested explanation about Init and GetWave algorithm overlap
  - I have added something to address this
- Anders: It should be explained you have to be careful about double-counting
- Arpad: It can be explained where we cover the booleans
- Walter: Arpad's approach with modifications should solve the problems
  - It should be ready for Arpad to present next week
- Arpad: I can introduce it in a general way now
  - General comments added to the beginning of Section 2
  - 2.1 is statistical overview
  - 2.2 TD overview
  - Not very detailed
  - Likewise, 2.3 is now 2.3 and 2.4 for separate statistical and TD
  - The 2 reference flows are independent
  - A new new step 4 has been added 
- Ken: Is this change a cleanup or enhancement?
- Arpad: It is a cleanup
  - I was requested to find a way to avoid deconvolution and ??
  - To satisfy both a duplicate impulse response was needed
- Walter: We are still working this out
  - Existing models will work
- Arpad showed flow diagram #8
  - This is to insure the optimizer gets the modified Init output
- Ambrish: Have we covered all possible scenarios?
- Arpad: We tried that with the truth table
- Ambrish: The truth table doesn't cover what is inside the model
- Todd: Actually it does
- Ambrish: We only discussed one way optimization can be done
- Walter: There can be problems when GetWave does not have all the EQ
- Ken: This assume EQ is in GetWave
- Walter: It is painful when some EQ is in Init
  - We have to leave it because we don't deprecate
  - We should not have to document that flow
  - It would be best to recommend against it
- Ken: Could we decide unanimously to take things out?
- Bob: That would be feasible
- Ken: It would be best to eliminate complexity
- Walter: I would support that
- Todd: It is the TTT cases that would go
- Ken: It would be better to not have Init and GetWave together
- Todd: Customers want the flexibility
- Arpad: Do we want to cover those TTT truth table equations?
- Ken: We should make those illegal
- Walter: We should remove Use_Init_Output from the BIRD
  - It should be ignored
  - Come up with a set of 9 flows based on what models have
- Anders: What about TX Init only or Rx GetWave only?
- Walter: Tx Init output is convolved with stimulus to produce a virtual
  GetWave
- Anders: Then the complication is when part is in each model
- Walter: Might have peaking filter ahead of DFE
  - Need to remember impulse response, use overlap and save method
- Arpad: If we eliminate Use_Init_Output we are saying all of the algorithm has to be in
  either Init or GetWave
- Walter: Does anyone here think we need to keep Use_Init_Output?
- Anders: Don't see why it is so complicated
- Todd: Take TTT off the table and it's more clear what the model does
- Arpad: How do we prevent double-counting?
- Walter: There is a benefit to having dual models
  - The initialization is better
  - Dramatic performance improvement
- Anders: Tx Init  ???
- Todd: It will optimize based on the wrong data
- Walter: SiSoft believes models should have both Init and GetWave
- Anders: What will happen to existing models if we take this out?
- Walter: Only 2 will be affected
- Todd: The flow is undefined anyway
- Bob: It could be a parser warning
- Arpad: This will have to remain unresolved for now

Next meeting: 06 July 2010 12:00pm PT

--------

IBIS Interconnect SPICE Wish List:

1) Simulator directives