IBIS Macromodel Task Group Meeting date: 10 October 2023 Members (asterisk for those attending): Achronix Semiconductor: Hansel Dsilva Amazon: John Yan ANSYS: * Curtis Clark * Wei-hsing Huang Aurora System: Dian Yang Cadence Design Systems: * Ambrish Varma Jared James Google: Hanfeng Wang GaWon Kim Intel: * Michael Mirmak Kinger Cai * Chi-te Chen Liwei Zhao 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: - Arpad noted that Michael had a new BIRD proposal "Pin Name Fields Extension" to discuss. - Arpad observed that the workload in ATM is currently light, now that Kinger's PSIJ Sensitivity BIRD has been submitted as BIRD226. Arpad asked whether we should consider using the ATM meeting for editorial work on the upcoming Touchstone 2.1. Arpad noted that the Interconnect task group currently has two major proposals it could continue working on: port mapping and pole- residue format. He asked whether we should consider working on the editorial process for Touchstone 2.1 in the ATM meeting, so Interconnect could continue working on the two new proposals. Bob said he would prefer to set the two proposals aside for the time being and continue to use the Interconnect meeting to work on the editorial issues for Touchstone 2.1. Arpad said this was just a question/suggestion, and we can consider how to use the ATM meeting time slot in the future. ------------- Review of ARs: Michael: Send out draft10 of the [AMI Test Data] proposal including more feedback from the ATM meeting. - Done. Michael had sent out drafts 10, 11 and 12. Michael: Develop a full syntactically complete example demonstrating the [AMI Test Data] proposal. - In progress. Michael said he would work on this after we get the proposal itself more settled. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: Arpad asked for any comments or corrections to the minutes of the October 3rd meeting. Michael moved to approve the minutes. Bob seconded the motion. There were no objections. -------------- New Discussion: AMI Test Data proposal: Michael reviewed draft12, which had been sent to the ATM list prior to the meeting. He said that draft10 included the changes discussed at the previous meeting, draft11 corrected an error in the keyword hierarchy tree, and draft12 contained editorial changes Arpad had suggested. Michael noted two references to "block size", which had not previously been defined, and said he had added two new sentences defining block size in terms of the wave_size parameter. Michael reiterated that he agreed with Ambrish's suggestion that he create a full syntactically complete example. He said he thought it had been hard to find all the relevant information by reviewing the rest of the specification. Arpad agreed and said that he had reviewed IBIS 7.2 to see what it said about waveform input and the number of columns passed in and out of AMI_GetWave. He said he didn't find language explicitly stating that a column of time points is not provided, though he said everyone apparently understood that. Curtis noted that the Golden_waveform_file section contained a sentence, regarding the number of columns of data, that only applied to the Statistical Type simulation, wherein the Golden_waveform_file contained the expected output of the AMI_Init call. Michael and Arpad agreed. Curtis said he thought the definition of Golden_waveform_file left room for confusion because it was still overloaded (it contained different information depending on whether the Type was Statistical or Time_domain). He asked whether we should remove this overloading and replace Golden_waveform_file with two different subparameters, each of which would be specific to one Type of simulation. Arpad agreed. Michael said we could break Golden_waveform_file into two subparameters, and he suggested Golden_impulse_file for the expected impulse response returned by AMI_Init. Ambrish, Arpad and Curtis said Golden_waveform_file should be required for Time_domain and illegal for Statistical. Golden_impulse_file should be required for Statistical and optional for Time_domain. If the model maker chooses, they can provide a Golden_impulse_file in the Time_domain case, and it will be compared to the output returned by AMI_Init when it is called prior to the first call to AMI_GetWave. The group discussed whether Golden_impulse_file should be prohibited if the AMI Reserved Parameter Init_Returns_Impulse is false. Arpad suggested that we should not make the prohibition. He said we might want to be able to test and confirm that the model is not mistakenly modifying the impulse response in a case in which Init_Returns_Impulse is false. Arpad had an additional suggestion regarding parameter names. He said he didn't think "impulse" was sufficiently precise and "impulse response" would be better. Michael and the group agreed, and Michael said he would change Input_impulse_file to Input_IR_file and add a Golden_IR_file subparameter. Michael said he would create a draft13 incorporating the changes. Pin Name Fields Extension: Michael noted that he had received several offline inquiries and requests that we increase the specification's 5-character limit on pin names. Additionally, he reported that Kyle Lake of Cadence had raised the same issue in the most recent Open Forum meeting. Michael said pin names are now often machine generated and represent more than the old data book pin names. He said the proposal is to change the maximum length of pin names from 5 to 20 characters. He said that the change itself is simple, but pin names are used in 10 other keywords, 9 of which have an explicit field length limitation. He said one option was to simply change the explicit field length limitation from 5 to 20 in all cases. He suggested that an alternative might be to remove the per parameter limitations altogether and just use the 1024-character overall line length limit. Curtis said that the ibischk parser currently relies on size allocations based on the various field length limits, and removing them altogether would create complications for the parser. He noted however, that the specification should not be driven by parser development considerations. Arpad said he thought it best to go with the quick and simple 5 to 20 change for now. We can revisit the issue later if this change results in requests to increase other field lengths as well. Michael agreed with this approach and said draft1 of the proposal contains the simple approach. Randy asked whether we also want to expand pin names beyond alphanumeric characters. Michael and Arpad agreed that "_", for example, would be useful. However, Arpad suggested that we handle that with a separate BIRD and keep it decoupled from this length change. Michael said draft1 did not require any changes based on the discussion in today's meeting. Arpad asked everyone to review it. - Curtis: Motion to adjourn. - Michael: Second. - Arpad: Thank you all for joining. New ARs: Michael: Send out draft13 of the AMI Test Data proposal including more feedback from the ATM meeting. ------------- Next meeting: 17 October 2023 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives