IBIS Macromodel Task Group Meeting date: 05 March 2024 Members (asterisk for those attending): Achronix Semiconductor: Hansel Dsilva Amazon: John Yan ANSYS: * Curtis Clark * Wei-hsing Huang Aurora System: * Dian Yang Raj Raghuram Cadence Design Systems: * Ambrish Varma Jared James Dassault Systemes: Longfei Bai Google: Hanfeng Wang GaWon Kim Intel: * Michael Mirmak Kinger Cai * Chi-te Chen Liwei Zhao Alaeddin Aydiner Keysight Technologies: Fangyi Rao Majid Ahadi Dolatsara Stephen Slater Ming Yan Rui Yang Marvell: Steve Parker Mathworks (SiSoft): Walter Katz Graham Kus Micron Technology: Justin Butterfield Missouri S&T: Chulsoon Hwang Yifan Ding Zhiping Yang Rivos: Yansheng Wang SAE ITC: Michael McNair Siemens EDA (Mentor): * Arpad Muranyi * Randy Wolff Teraspeed Labs: Bob Ross Zuken USA: Lance Wang The meeting was led by Arpad Muranyi. Curtis Clark took the minutes. -------------------------------------------------------------------------------- Opens: - None. ------------- Review of ARs: Michael: Review draft2 of the Block Concepts Clarification BIRD and attempt to streamline the language. - Done. Michael reported that he had a cleaned-up version to review in today's meeting. Arpad: Draft an editorial clarification BIRD to make it clearer to the reader when an AMI parameter's value in the .ami file is merely a placeholder. - Not Done. Arpad said he had started to work on it, but it began to look more complicated than we first assumed. See discussion below. Arpad said he would remove this AR. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: Arpad asked for any comments or corrections to the minutes of the February 27th meeting. Ambrish moved to approve the minutes. Michael seconded the motion. There were no objections. -------------- New Discussion: Clarification BIRD for placeholder parameter values in the .ami file: Arpad said he had begun work on this. However, he had run into issues that made it appear to be more significant effort than we first assumed. For example, in IBIS 7.2, we find values such as "ignore_me", "0.00 0.00 0.00", "placeholder", and other values used in placeholder situations. Arpad said he had conferred with Randy, who had originally raised the issue. Randy agreed this was probably a topic for the Editorial task group to take up later. Randy noted that there are already two issues in the known issues document for IBIS 7.2 related to placeholder values. So, the Editorial task group will eventually address this issue. Arpad moved to remove this item from the ATM agenda. Ambrish seconded. There were no objections. AMI Block Concepts Clarification BIRD: Michael reviewed his new cleaned-up version. He said the changes were mostly incorporated from Arpad's private email. He noted one other change based on feedback from Curtis that the definition of symbol used in draft2 was incorrect. Curtis said that draft2 included language changes for page 216 that included the following misleading text: ...NRZ (non-return-to-zero) system where only one of two logical states or symbols is possible in each Unit Interval... Curtis said this mistakenly implied that symbols were the same as the logic levels and there were two symbols per UI for NRZ. Michael had agreed and adopted new language. Curtis noted, however, that he did not think this definition of NRZ should be added in this location at all. He said defining every specific term in the flow of the text made it cumbersome, and it risked alienating the reader by defining basic concepts in the middle of the AMI flow descriptions. Curtis said that if we felt it necessary to define everything so precisely, then he agreed with Michael's suggestion (previous meeting) that we should add a "Definitions" section. This would then allow the regular text to flow much more freely. Curtis said he was still not entirely sure that this clarification BIRD was necessary at all, however. Curtis noted other language from the page 216 changes in draft2: ...time segments (groups of sequential waveform samples, elsewhere called blocks)... He said we could streamline this language by choosing one or the other. Why not just use the term blocks all the time? Arpad agreed that we should reduce the use of synonyms in the specification. Michael said there were subtle differences in usage. He said he'd drawn the equivalence in this text, but that people often talked about time segments when thinking in terms of samples and often used block when thinking in units of bits. Curtis said this was one more reason to remove the synonyms. He said we should not use different names for the same concept just because we are using different units. Arpad asked why we need to explicitly emphasize "sequential" in: ...groups of sequential waveform samples... Michael said the goal was to explicitly state that they would be adjacent samples and monotonically increasing in time. The tool would not be subsampling or otherwise straying from the sample_interval, for example. Michael suggested that he set this BIRD aside and start work on a different BIRD to introduce a Definitions section. Once that new BIRD is finished, we could significantly streamline the language in this BIRD. While most of the terms we need to define initially would be from Chapter 10 (AMI), Michael suggested we add the Definitions in Section 3, since we have many specification-wide definitions already there. Michael said the initial list of definitions we need wouldn't be too long, e.g., bit, symbol, segment, block, NRZ, PAMn. [Clock Group] Proposal: Michael said he did not have a new update relative to last week's meeting. He said the goal will be to define the physical relationships between pins. For example, in a DDR context, which group of DQ pins utilizes a given DQS strobe. He said defining these relationships might be complicated for a DDR controller model, where the grouping might be configurable instead of fixed. The timing relationships might also be important, and it might be helpful to define them within IBIS. One example is data sheet data such as the range of possible deviations in timing between the data and strobe. He said the downside is that we have tended to avoid those types of explicit timing relationships in IBIS. He said it becomes difficult to accommodate corners and other variables. Arpad noted that IBIS had generally avoided internal and even buffer timing questions in the past. Michael said there is always interest in delays at the outputs. At a minimum, however, this BIRD will seek to define which groups of pins need to be simulated. He said we could define the grouping relationships now and leave the timing values to a separate future BIRD if necessary. - Michael: Motion to adjourn. - Curtis: Second. - Arpad: Thank you all for joining. New ARs: Michael: Draft a new BIRD proposal to introduce a "Definitions" section. ------------- Next meeting: 12 March 2024 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives