IBIS Macromodel Task Group Meeting date: 20 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 send Vladimir's presentation to Mike L. for posting - Done - 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: Kumar gave a presentation "IBIS AMI: Flows" (Kumar and Ambrish): - The question is whether AMI will support statistical simulation - Slide 2: - A model with GetWave is intended for time domain simulation - Slide 3: - Tx Init, Rx Init will support statistical analysis - Slide 4: - Tx Init, Rx GetWave will not support statistical analysis - Mike S: We have many models that support both time domain and statistical - Kumar: The proposal is to separate them out - Slide 5: - Tx GetWave, Rx Init - Walter: If Tx and Rx model writers are different people, do they both assume LTI? - What does Rx Init expect as input? - Kumar: Coefficients - Walter: Another slide says it will be the channel input - Fangyi: The Rx model writer will not be able to know what the input is - Slide 6: - Tx GetWave, Rx GetWave - Minimum complication - Slide 7: - Tx GetWave, Wfm at Rx pad - Rx Init receives unit channel response, normalized to 1 - Walter: Rx Init does not know what is being passed in - Slide 8: - Statistical analysis is supported only without GetWave - Time domain analysis is supported in all cases - Walter: How does Rx Init know what to expect as input? - Arpad: Will Rx Init have 3 different functions? - Kumar: Different models will be used Walter show his presentation "IBIS 5.0 AMI Basic Principles": - Slide 2: - Serdes have simplifications over normal simulations - LTI - Interconnect can be included - Vladimir's method is interesting - Complications: - Tx and Rx stimulus signal processing - Signal algorithms are proprietary - Need to simulate millions of bits - Slide 3: - BIRDs 104 and 107 based on knowledge of how AMI already being done - Successful IBIS 5.0 AMI models exist - Existing IBIS models remain legal for future IBIS releases - Slide 4: - SerDes channel has Tx SP, analog channel, Rx SP - Traditional IBIS does not model differential LTI buffers - Slide 5: - Many buffers are close enough to LTI - Sometimes there is a useful LTI approximation for non-LTI - Slide 6: - Same as slide 5 - Slide 7: - All AMI DLLs have AMI_Init and AMI_Close - If it has AMI_GetWave then it is non-LTI - The ASCII file tells the tool how to use the model - Slide 8: - LTI models do not have AMI_GetWave - Page 5 from Walter's flow: - Can be used for Stat-eye or convolved - Page 5 from Arpads's flow: - Requires 5.1 AMI to get impulse response - Kumar: What happens if there is no impulse response? - Walter: One of 3 things has to be provided - Arpad: The boolean switches control this - Fangyi: This adds the need for deconvolution in the flow - Slide 10: - If Tx AMI_Init does not return modified imp resp, Rx AMI_Init may not be able to optimally initialize - If ether Tx or Rx AMI_Init does not return modified imp resp, statistical is impossible - Slide 11: - Rx also returns clock ticks - Slide 12: - An overlap and save algorithm is needed - Slide 13: - Page 2 from Walter's flow: - Tx GetWave is independent of LTI approximation - If Tx Init does not return modified impulse response it hampers Rx Init - Fangyi: This assumes Use_Init_Output = false - Walter: Yes, 2 IC vendors use this param - Arpad: The latest flow has no booleans at the suggestion of Fangyi - The only flow that works is with a Init_Returns_Filter_Only boolean - In this case the EDA tool performs all convolutions - We also need the boolean for models that want something other than convolution performed - This brings us back to an earlier version of the flow - Ambrish: Do you ever use Use_Init_Output = true - Walter: No - Kumar: Use_Init_Output = false when GetWave is present - LTI models have to work with non-LTI models - Slide 14: - Need to support waveforms from external sources - EDA too must determine equalization - Slide 15: - Every EDA platform must give the same results given the same inputs - They may differ on how they choose inputs - DCD could be implemented in different places - Ambrish: Two separate models are needed for Use_Init_Output = true and false - Walter: We have single models that do this - Fangyi: Rx EQ Filter is needed if Tx GetWave_Exists (AMI_Flows_3W page 3) - Walter: Rx Init requires deconvolution - Should have options for Rx to return just EQ - Not as critical in the Tx side - Arpad: The problem is when something other than convolution must be done - Walter: This is driven by IC vendors, not programmers Kumar: A model either supports statistical analysis or it does not - Walter: You may have to write 4 to 6 models - Arpad: Having more model files will not help users - Kumar: Model developers will decide which modes they will support - Arpad: We need more discussion on this, email and at least one more meeting AR: Walter and Kumar send presentation to Mike L for posting AR: Arpad modify presentation and send to Mike L for posting AR: Mike L post presentations from Walter, Kumar, and Arpad Next meeting: 27 Oct 2009 12:00pm PT -------- IBIS Interconnect SPICE Wish List: 1) Simulator directives