Ambrish, How can you make such a bold statement about the future: "However for non-linear models, there will be no equalization (that approximates the equalization in the GetWave) in the Init function..."? It makes me feel that I should come to you for advise on how to make my investments... :-) I think this is exactly what some of us disagrees with you. Walter may even come back saying that they have already made models like that, but I will let him speak for himself... From this I am starting to get the impression that we are really not talking about a flow issue here, but the question of what the intent was between the Init and GetWave functions regarding supporting LTI and non-LTI devices in the same model. I thought this question was settled before in my message to Ken (included below). Why are we still dwelling on it? Arpad ================================================================ -----Original Message----- From: Muranyi, Arpad Sent: Wednesday, March 31, 2010 12:35 AM To: 'IBIS-ATM' Subject: Comment on Sigrity's presentation Ken, I would like to make a comment on slide 3 of the presentation you gave us in the ATM teleconference a week ago and the discussion that followed. http://www.vhdl.org/pub/ibis/macromodel_wip/archive/20100323/kenwillis/S igrity%20AMI%20BIRD%20Feedback/Sigrity_AMI_BIRD_fback.pdf If I understood the discussion correctly, you and Kumar are concerned that one model could produce different results when only its Init functions are used vs. when its GetWave functions are also used. The discussion between you and Todd ended in a disagreement. However, it just occurred to me that the AMI portion of the IBIS 5.0 specification already talks about the possibility of having two types of analysis paths, one that only uses the Init functions (for LTI systems), and another that also involves the GetWave function when non-LTI effects are present. This is found in Section 10, 2.1 and 2.2 in the specification. Doesn't this close the discussion we had in the meeting when you were presenting in a sense that you are challenging an intent that is already clearly spelled out in the specification? Thanks, Arpad ================================================================= ________________________________ From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Ambrish Varma Sent: Tuesday, April 20, 2010 7:08 PM To: IBIS-ATM Subject: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Hi Arpad, Sorry - I did mean Nonlinear, and/or Time Variant. Well, there is a link. For a linear model, there will be no getwave, so slide 5 will correctly represent the scenario. However for non-linear models, there will be no equalization (that approximates the equalization in the GetWave) in the Init function, as such Tx (and RX) Use_Init_Output will always be true. This would also mean that these models (with Getwave) would not be suitable for Stateye type simulations. Only models that would like to perform both Stateye type simulation as well as (meaningful) Time Domain simulation via the same model will have issues as a straight path from Tx_Init to Rx_Init is needed for Stateye type sim. In my opinion, if we follow the simple rule set out by the spec, a lot of the confusion will automatically be resolved. Please let me know if I was clear enough. :-) Thanks, Ambrish. ________________________________ From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad Sent: Tuesday, April 20, 2010 6:38 PM To: IBIS-ATM Subject: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Ambrish, I think you meant "Time-variant" in a few places where you wrote "Time-invariant"... but aside from that I am not sure what this has to do with Walter's comment about what goes into Rx_Init. Can you explain how these two topics are related? Thanks, Arpad ================================================================== ________________________________ From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Ambrish Varma Sent: Tuesday, April 20, 2010 5:32 PM To: IBIS-ATM Subject: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Hi Arpad, The spec is very clear in delineating between Linear, Time-invariant model and Nonlinear, and /or Time-invariant models. (section 2, chapter 10). It was expected that a non linear/time-invariant model would not try and model an approximation of the same algorithm in the Init function. Also, a linear model would not have a getwave function. Thanks, Ambrish. ________________________________ From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad Sent: Tuesday, April 20, 2010 6:08 PM To: IBIS-ATM Subject: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Walter, Thanks for your feedback. I think you just made us discover another discrepancy in the spec. Contrast what you quoted with this from the IBIS specification: What do you suggest we should do about this? Arpad ======================================================================== ________________________________ From: Walter Katz [mailto:wkatz@xxxxxxxxxx] Sent: Tuesday, April 20, 2010 4:27 PM To: Muranyi, Arpad; IBIS-ATM Subject: RE: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Arpad, In the IBIS 5.0 specification: | Use_Init_Output: | | Use_Init_Output is of usage Info and type Boolean. When | Use_Init_Output is set to "True", the EDA tool is | instructed to use the output impulse response from the | AMI_Init function when creating the input waveform | presented to the AMI_Getwave function. | | If the Reserved Parameter, Use_Init_Output, is set to | "False", EDA tools will use the original (unfiltered) | impulse response of the channel when creating the input | waveform presented to the AMI_Getwave function. | | The algorithmic model is expected to modify the waveform in | place. | | Use_Init_Output is optional. The default value for this | parameter is "True". | | If Use_Init_Output is False, GetWave_Exists must be True. In what was agreed to in November, the input the Rx_Init was always hAC(t) X hTEI(t). In what you presented this week, the input to Rx_Init is either hAC(t) or hAC(t) X hTEI(t), depending on the value of Tx Use_init_Output. I believe based on the IBIS 5.0 specification above that the November flow is correct and the Spec. correction flow that you presented this week is incorrect. Walter Walter Katz 303.449-2308 Mobile 720.333-1107 wkatz@xxxxxxxxxx www.sisoft.com -----Original Message----- From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx]On Behalf Of Muranyi, Arpad Sent: Tuesday, April 20, 2010 12:48 AM To: IBIS-ATM Subject: [ibis-macro] Re: AMI_Flows_6.pdf for today's ATM teleconference Here is the AMI_Flows_6.pdf file once again. I made the changes which were suggested to me in the last ATM meeting. This flow includes only the correction we wanted to make on the existing spec flow. I am not sure what the decision was about the last two slides which deal with the Rx pad waveform. Did we say we would delete these slides altogether and not address this capability in this BIRD? As far as I can tell, we can't achieve this flow without deconvolution... Please familiarize yourselves with these slides, because I would like to achieve closure on this flow in the ATM teleconference tomorrow. Comments are welcome before or at the meeting. Thanks, Arpad ========================================================= ________________________________ From: Muranyi, Arpad Sent: Tuesday, April 13, 2010 1:39 PM To: 'IBIS-ATM' Subject: AMI_Flows_6.pdf for today's ATM teleconference For those who are unable to join the meeting via LiveMeeting, here is a new flow diagram to aid the discussion on the subject. Arpad ==================================================