IBIS Macromodel Task Group

Meeting date: 24 June 2014

Members (asterisk for those attending):
Agilent:                    * Fangyi Rao
                            * Radek Biernacki
Altera:                       David Banas
ANSYS:                      * Dan Dvorscak
                            * Curtis Clark
Avago (LSI)                   Xingdong Dai
Cadence Design Systems:       Ambrish Varma
                              Brad Brim
                              Kumar Keshavan
                            * Ken Willis
Ericsson:                     Anders Ekholm
Intel:                      * Michael Mirmak
Maxim Integrated Products:    Hassan Rafat
Mentor Graphics:            * John Angulo
                            * Arpad Muranyi
Micron Technology:            Randy Wolff
                              Justin Butterfield
QLogic Corp.                  James Zhou
                              Andy Joy
SiSoft:                     * Walter Katz
                            * Todd Westerhoff
                            * Mike LaBonte
Synopsys                    * Rita Horner
Teraspeed Consulting Group:   Scott McMorrow
                            * Bob Ross

The meeting was led by Arpad Muranyi.

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

- Michael M: A draft BIRD has been emailed to address agenda item #7

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

- None

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

- Michael M write BIRD for I/O AMI models.
  - Done

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

Introduction of Rita Horner:
- Technical marketing manager for SerDes product line at Synopsys in Mountain View, CA.

New DLL dependency BIRD:
- Arpad showed a BIRD draft.
- Arpad: This recommends testing for DLL dependencies.
  - This is not submitted yet.
- Fangyi: This is to test compatibility?
- Arpad: It is to find dependencies on things like msvcrt80.dll, etc.
- Todd: Testing on a clean machine is simple.
  - It might be better to test on a simulator.
- Michael M: Is there a listing of libraries provided with each OS?
- Todd: That would be difficult.
- Arpad: This will be submitted to the open forum.
- Bob: Some of the context text might have been replaced by BIRD 155.
- Arpad: That has been incorporated already.

New I/O AMI model BIRD:
- Michael M showed a draft BIRD.
- Michael M: This requires no changes to existing models or tools.
  - A fourth item is added: Platform_Compiler_Bits_Direction.
  - Direction can be TX or RX.
  - This establishes which flow is used where Model_type is *_I/O.
  - The parser can check for problems in the AMI file.
- Bob: This looks good.
  - Can a DLL be I/O?
- Michael M: I don't see any reason why not.
- Radek: Agree. A parameter could tell it the direction.
- Arpad: An enable control could be passed.
- Michael M: Not sure if that's the best way.
- Arpad: It might be convenient for dynamic bus turn-around.
- Ken: A tool might associate two DLLs with that control.
- Michael M: The idea is to give tools the option.
  - But this is more work for the model maker.
- Todd: I have reservations about trying to model bus turn-around.
  - We should say it explicitly if so.
  - I would prefer to exclude that.
  - We can't model the impedance transition.
- Michael M: An IBIS parallel is that we never modeled transitions for Series_switch.
  - The idea here is to be able to handle models that are I/O.
- Arpad: How will the analog models be changed?
- Michael M: The tool should observe the state of the enable as it already does.
- Arpad: This is OK for separate DLLs but not for where one DLL is used.
- Radek: I don't see any conflict between these positions.
- Ken:Tools already figure out who is driving.
- Arpad: How would the DLL switch code sections?
- Michael M: That's a special case.
  - A DLL is a black box, may be need to be told.
- Todd: The waveforms to TX GetWave and RX GetWave are very different.
  - The model would have to be told what it has been given.
- Michael M: This can be just a passed parameter.
- Walter: It would have to be reserved.
- Todd: Are any of these passed to the model?
- Walter: Yes.
- Ken: Is there any value to this?
- Michael M: It might be an optional reserved parameter.
  - This can be tested by changing a model to I/O today.

- Radek: Do we require both TX and RX to be present for these?
- Michael M: It could be I/O but some Executable lines could be missing.
- Walter: Some tools generate IBIS files and make everything I/O.
  - This makes that more possible, but tougher for simulators to use.

Redriver flow:
- Walter showed Redriver Flow Problem.
- Walter: The number of permutations that would have to be handled becomes quite large.
  - Most TXs can't optimize themselves in the field.
  - A few of those might be out there.
  - That should be indicated or at least controllable.
  - That has limited applications today.
  - If we assume it won't, our flows all become simpler.
  - Each redriver stage multiples permutations by 9.
  - We either let tools do what they do now or have optimizing TX as a special case.
- Arpad: Can this be specified without defining a flow?
- Walter: All RX AMI_Inits and AMI_GetWaves optimize.
  - No existing TX has adaptive optimization except under RX control.
  - Kumar wants the TX and and RX to be symmetric.

-------------
Next meeting: 01 Jul 2014 12:00pm PT

-------------
IBIS Interconnect SPICE Wish List:

1) Simulator directives