IBIS Macromodel Task Group Meeting date: 25 Oct 2011 Members (asterisk for those attending): Agilent: * Fangyi Rao * Radek Biernacki Altera: * David Banas Ansys: Samuel Mertens * Dan Dvorscak * Curtis Clark Arrow Electronics: Ian Dodd Cadence Design Systems: Terry Jernberg Ambrish Varma Celsionix: Kellee Crisafulli Cisco Systems: Ashwin Vasudevan Syed Huq Ericsson: Anders Ekholm IBM: Greg Edlund Intel: Michael Mirmak LSI Logic: Wenyi Jin Mentor Graphics: * John Angulo Zhen Mu * Arpad Muranyi Vladimir Dmitriev-Zdorov Micron Technology: Randy Wolff NetLogic Microsystems: Ryan Couts Nokia-Siemens Networks: * Eckhard Lenski QLogic Corp. James Zhou Sigrity: Brad Brim * Kumar Keshavan Ken Willis SiSoft: * Walter Katz Todd Westerhoff Doug Burns * Mike LaBonte Snowbush IP: Marcus Van Ierssel ST Micro: Syed Sadeghi Teraspeed Consulting Group: Scott McMorrow * Bob Ross TI: Casey Morrison Alfred Chong Vitesse Semiconductor: Eric Sweetman Xilinx: Mustansir Fanaswalla The meeting was lead by Arpad Muranyi ------------------------------------------------------------------------ Opens: - Mike: Europe DST change is this weekend, US Nov 6 -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad prepare slides for today on corners - Done - Mike post Jitter BIRD - Done ------------- New Discussion: Meeting schedule: - Arpad showed a list of future meeting dates - Some conflict with summits - We will not meet Nov 15 and 22, Dec 27 Arpad showed Notes on Corners in IBIS (BIRD 140): - Slide 2: - IBIS 1.1 note on data derivation of corners - Arpad: It was felt that min/max could be scaled from typical - Slide 3: - IBIS 2.1 relaxed the rules - Slide 4: - Page 174 requires min and max values "for all remaining keywords" - Examine which keywords fall after that statement - Arpad: We can't go by that strictly - Recent spec changes may not have observed it - Slide 5: - Arpad: C_Comp is independent even where correlation is known - Defines "conservative approach" as max=slow, min=fast - Slide 6: - Arpad: More ringing could slow down effective performance - We avoided defining it for that reason - Slide 7: - Arpad: Does IBIS min = AMI slow and IBIS max = AMI fast? - C_comp is not the only questionable parameter - Slide 8: - BIRD 133 [C Comp Corner] - Walter's proposal: let EDA tool decide until BIRD 133 approved - Use new keyword after that - Kumar: Does AMI Corner format define this? - Walter: It is typ/slow/fast, not typ/min/max - Bob: There are no GT/LT requirements - Problems: - [C Comp Corner] is optional - What does tool do if absent? - There are other independent parameters - Slide 9: - Other independent keywords: - [Diff Pin] tdelay_*** - [Rgnd], [Rpower], [Rac], [Cac] - [* Series] - [External Model], [External Circuit] - Slide 10: - Arpad proposes extending IBIS to 5 corners, adding slow and fast - Slide 11: - When slow not present max is used for slow - When fast not present min is used for fast - Slide 12: - Why have a default that is opposite of the intuitive approach? - Slide 13: - We may need 6 corners because AMI_typ may not equal IBIS_typ - Slide 14: - Summary - David: On slide 5 it is fundamental that the tool makes the choice - IBIS is a datasheet, not a model - We are getting away from our roots - Walter: People generally have a good idea of slow and fast - David: We know that min is faster for C_comp - Walter: Sometimes C_comp is larger for the fast corner - Walter: Disagree on using min for slow and max for fast - Model makers should document what they are doing - David: This would have us using reverse order of current practice for C_comp - Bob: IBISCHK will warn for min > max - David: On slide 11 is say min is fast corner - Walter: People want more than three corners - We should do this in a general way - Arpad: Simulators can't just do what they do now because the spec is not clear - Walter: In 5.1 they can use [C Comp Corner] - Kumar: Min/typ/max is a numerical concept - It doesn't map to typ/slow/fast - With 5 corners do I simulate all 5 or just three? - David: Complexity creates more work - Walter: In 5.1 there should at least be a comment - Should this be done so the program can read it? - Kumar: A comment is not a major, fundamental change - Walter: We should solve the fundamental C_comp problem with BIRD 133 - Radek: We should consider a long term solution - This needs to address all parameters, not just C_comp - Extending BIRD 133 to other parameters would require more keywords - In Arpad's proposal 4th and 5th columns would only be needed for C_comp - Comments are welcome but users might not see them in large IBIS files - Arpad: We could use 5 columns for C_comp for now - The overwriting mechanism is confusing - Bob: I oppose adding the 2 more corners - Tools would have to handle it forever - Kumar: A new keyword could document the C_comp mapping - Arpad: What about the other problem parameters? - Bob: Params like [R Series] not derived from silicon are a problem - We will not solve those with new keywords - Walter: Associating ISS subckts with typ/slow/fast would be better - We should vote to use either Bob's approach or Arpad's - Bob: I'm prepared to move forward on the C_comp BIRD - Arpad: What does AMI Corner align with? - Walter: It aligns with data derivation - Arpad: C_comp is still needed for AMI models - Walter: [C Comp Corner] fixes that ------------- Minutes by Mike LaBonte Next meeting: 01 Nov 2011 12:00pm PT Next agenda: 1) Task list item discussions ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives