Meeting date: 25 apr 2006
Members (asterisk for those attending):
*Arpad Muranyi,
*Bob Ross,
*Todd Westerhoff,
*Mike LaBonte,
Paul Fernando,
*Barry Katz,
Walter Katz,
Ken Willis,
*Ian Dodd
Lance Wang
*Richard Ward
Doug White
Sanjeev Gupta

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

none

-------------

Topics
- encryption
- leaving SPICE behind

Encryption
- Simple IBIS buffers do not need encryption.
- Modern buffers have netlists and algorithms, and need protection.
- Maybe we can parameterize parts of the model so it is reusable.
- Will IBIS offer encryption?
- Arpad's letter to VHDL and VAMS
  - Replies range from "don't do it" to "we need it".
  - SystemVerilog (or ???) has it already.
  - VHDL is only a programming language.
  - Someone has a website offerring a solution (maybe general).
  - Cadence supports encryption
    - Major EDA companies tend to have that.
    - Cadence is proposing a solution.
    - IEEE group is proposing also.
- IBIS could join other groups to push for a solution.
- Cadence presentation from website:
  - http://www.accellera.org/apps/group_public/download.php/118/VHDL_IP_Encryption.ppt
  - Seems to be discussing both VHDL and VAMS.
  - We could invite Cadence (Steven Dovich) to explain it to us.
  - Need to know more about key escrow.
- Will per-customer keys be required?
- Mentor has encryption for digital simulation only, not AMS.
- IP protection methods:
  - encryption
  - obfuscators
  - compiled models
  - non-disclosure agreement
- TI supplies Matlab P-code
  - If people want to decompile code that badly, so be it.
- Whatever is chosen, the legal departement has to be happy.
- We could start with encryption in HSPICE
  - Will Synopsys will implement a new encryption scheme?
  - If Cisco and Intel require a new mechanism, maybe they will.
  - Major EDA softwares are able to use HSPICE for simulation though.
  - HSPICE is slow though.
- Is this an issue for the IBIS Futures group?
  - We seem to have a great interest in the subject of IP protection.
  - If we get bogged down on encryption, then library work will suffer.
  - Allocate subject time in future meetings?
    - 15 minutes for encryption
    - 45 minutes for library

AR: Arpad compile email responses
AR: Mike investigate commercial solution
AR: Mike ask Cadence (Dovich) to join next call

Richard's DFE-based receiver model
- Should we hypothesize what new building blocks are needed for this?
- Knowing when it is accurate enough is the challenge.
- What does Matlab have that makes it so useful?
  - People think in terms of math and formulas.
- Can TI share the model?
  - May be better to build our own reference model.
- Can Gary Pratt work with Richard to design a model?
  - Start off by writing basic specifications for each block.
  - TI and Mentor might be a good match.

AR: Richard check if he can share the receiver model
AR: Todd, Ian, Richard meet this Wednesday

-------------
Next meeting: Tuesday 02 may 2006 12:00pm PT