[ibis-macro] Minutes from the 27 Oct 2009 ibis-atm meeting
- From: "Mike LaBonte (milabont)" <milabont@xxxxxxxxx>
- To: "IBIS-ATM" <ibis-macro@xxxxxxxxxxxxx>
- Date: Fri, 30 Oct 2009 13:50:14 -0400
Minutes from the 27 Oct 2009 ibis-atm meeting are attached.
Mike
IBIS Macromodel Task Group
Meeting date: 27 October 2009
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 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, Cadence Design Systems
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
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
Zhen Mu, Mentor Graphics
------------------------------------------------------------------------
Opens:
- None
--------------------------
Call for patent disclosure:
- No one declared a patent.
-------------
Review of ARs:
- Arpad modify presentation and send to Mike L for posting
- Done but not yet posted
- Walter and Kumar send presentation to Mike L for posting
- Walter sent his
- Mike L post presentations from Walter, Kumar, and Arpad
- Walter's is posted
- Arpad Write a clarification BIRD to discuss accuracy issues related to the
various AMI clock_tick algorithms in an IBIS-AMI DLL
- TBD
- Todd: Update the BIRD for IBIS S-parameter box based on feedback from
discussion
- No update
- 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
- 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 his updated AMI Flow presentation:
- Changes were made on slide 1 "all in one"
- Fangyi: Why is there a line directly from 1 to 3?
- Arpad: Tx Init passes to both under some conditions
- When Rx GetWave doesn't exist Rx Init returns filter
- Slide 2:
- This is the flow when GetWave exists
- The Init flow is basically the same
- Fangyi: How to handle crosstalk?
- Walter: We need an enhancement to specify how to handle crosstalk
- Mike S: We need to apply the Tx impulse response to xtalk
- Fangyi: It is the Rx that cares about xtalk
- Arpad: We should table this for now
- Arpad: The GetWave flow is straightforward
- Todd: Why does Rx GetWave have curly brackets and Tx has square brackets
- Arpad: Needed both because they are nested in one place
- Slide 3:
- The EDA tool now has to convolve
- Fangyi: This is where we need deconvolution
- Arpad: Only if Rx Init can't return a filter response
- Walter: Convolution boxes 3 & 4 could be combined
- Arpad: Agree
- Slide 4:
- The tool may not need need convolution in box 3
- Slide 5:
- Convolution in boxes 3 & 4 are combined here
- Slide 6:
- Deconvolution is needed unless Rx Init returns filter response
Arpad: We are back to square one on slide 3
- Ambrish: We have to support Use_Init_Output for 5.0 compliance
- Todd: What do we tell vendors who have made models?
- Arpad: Vendors could make models for both flows
- Walter: There is no new flow here
- There are 3 recommendations and a new feature
- The recommendations:
- We can recommend that vendors use the Use_Init_Output switch in the future
- They will set Use_Init_Output to false
- When the model has GetWave make a best effort attempt to generate
a modified impulse response
- These are the reference flows
- If vendors continue to use old modes it burdens the EDA tools
- But, EDA vendors can choose what they support
- Ambrish: What is on the screen now is the new flow?
- Walter: This is the "reference" flow
- Mike S: Arpad's slide gives a number of flows
- It may depend on model capabilities
- New flows could be supported based on switches
- Ambrish: So we still support Use_Init_Output
- Todd: The new slides assume Use_Init_Output = False
- Walter and I made a mistake saying it could go away
Fangyi: This doesn't work for statistical analysis
- Walter: Correct, but the spec is inconsistent
- Todd: We did not discuss statistical processes when we wrote this
- Fangyi: Five of the flows don't work for statistical analysis
- We need to clearly state how it works with the new boolean
- Ambrish: We assumed using GetWave meant no statistical analysis
- Todd: At the limit it doesn't support statistical
- In practice it can work sometimes
- Fangyi: If Tx Use_Init_Output is false and Tx return filter what is passed to
Rx Init?
- Walter: The Input to Rx Init is always the channel with best estimate
of Tx initialization
- Walter: There are 3 "shoulds":
- 1 Init returns filter
- 2 Use_Init_Output is false
- 3 Make a best effort to generate an Htei
- Arpad: GetWave doesn't know what came out previously
- Ambrish: Tx Init may just return hAC
- But Rx Init might have EQ
- For the statistical flow there is no Tx Init
- Mike S: It will not do a good job of statistical analysis in some cases
Arpad: There can be some legal combinations that don't really work
- For example if you put an Rx at each end there will be no waveforms
- Walter: The question is if Rx Init optimizes itself or not
- If it does, does it make sense to have an LTI Rx with non-LTI input?
- Todd: Optimization is non-LTI, so you can't have it in Rx Init
- Fangyi: If Use_Init_Output is false what goes in Rx Init?
- Todd: The existing spec is for time domain only
- Ambrish: Not if GetWave is absent
- Walter: This was a kluge because the Tx GetWave input was analog
- Todd: The input to Rx Init is always convolved
- Fangyi got stuck because of the AMI 5.0 text
- The spec was all about deriving a time domain waveform
- Ambrish: So I need a Tx GetWave that is slightly less accurate?
- Mike S: Yes
- Arpad: This is similar to starting with [Ramp] and adding waveforms in IBIS
- Todd: But IBIS was always time domain, this is not the same thing
- Not every model will support both statistical and time domain analysis well
- We still should offer guidance
Ambrish: We need 2 kinds of models?
- Todd: How does the user use that?
- Ambrish: Instead of 2 files they would have 3 including both Tx and Rx
- Arpad: It doesn't help if a model has 2 files
- Ambrish: They don't have to be in separate files
- Mike S: Is that simpler for the user?
- Ambrish: The user needs to plug in the best model he has
- Walter: Would [Model Selector] be used?
- Mike S: The user knows which flow is used in the EDA platform
- It is easier to depend on that control than to choose between models
- Ambrish: If you have a statistical-only model that is the only choice
- Mike S: Arpad showed us how to get around that
- Todd: Dozens of models are out there under the belief they can do both
time domain and statistical
Arpad: This is all about whether to be backward compatible
- These flows try to be compatible while fixing problems
- Starting from scratch will not be easy
- Todd: That is what we have been doing
- Walter: We should go with what we have except:
- Rx Init_Returns_Filter should be forced to true
- Fangyi: If Tx Init_Returns_Filter is true can Tx Use_Init_Output be false?
- Mike S: The current spec text is incomplete on this
- Arpad: This drawing puts Use_Init_Output in a different place
There was no objection to the flows as shown today by Arpad
Walter: Arpad has the flows, I will change the text
- Fangyi: We should specify the other possible combinations
- Todd: The spec has to have text, it can't rely on pictures
- Walter: We need a PDF IBIS spec
Next meeting: 03 Nov 2009 12:00pm PT
--------
IBIS Interconnect SPICE Wish List:
1) Simulator directives
Other related posts:
- » [ibis-macro] Minutes from the 27 Oct 2009 ibis-atm meeting - Mike LaBonte (milabont)
