IBIS Macromodel Task Group Meeting date: 12 December 2017 Members (asterisk for those attending): ANSYS: * Dan Dvorscak * Curtis Clark Broadcom (Avago): Xingdong Dai Bob Miller Cadence Design Systems: Ambrish Varma Brad Brim Kumar Keshavan Ken Willis eASIC: David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker IBM Luis Armenta Trevor Timpane Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Ming Yan Maxim Integrated Products: Hassan Rafat Mentor, A Siemens Business: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff * Justin Butterfield SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte SPISim: * Wei-hsing Huang Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. Curtis Clark took the minutes. -------------------------------------------------------------------------------- Opens: - Arpad reviewed the schedule for upcoming ATM meetings. The group plans to cancel the meetings on December 26th and January 2nd. All other meetings in December will be held as usual. - Arpad noted that he had added a "Topic bin list:" from Michael Mirmak to the agenda. Walter suggested that at a future meeting we might allocate 5 minutes per topic on this list to decide how to move forward. ------------- Review of ARs: - None. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - Arpad: Does anyone have any comments or corrections? [none] - Walter: Motion to approve the minutes. - Michael M.: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: BIRD189 - Groups/Sets - Arpad: [to Michael M.] Do we have a target date goal for completing this? - Michael M.: One goal was the end of the year, but I don't think it's possible. - More realistic, but still tight, would be to have something submitted to the Open Forum before DesignCon. - We probably only have 3 weeks left to hit that deadline. - We still have the Groups/Sets discussion. - We still have a technical issue with Aggressors. - Then it can be cleaned up and prepared for a vote. - Walter: [sharing his latest proposal] - We still have to approve the version to be called draft12 and uploaded. - I started with a version called draft12_wmk as a baseline for this change. - [Note: draft12_wmk is draft12_v1 with the changes accepted.] - [Reviewing changes relative to draft12] - Solution Requirements: - Add the following sentence to #12: - The full path may be described using BIRD 189 Buffer to Pad models and Legacy Pad to Pin Models. - Add the following sentence to #15: - The model user may be required to use Legacy Package models with on- die BIRD 189 models. - These two sentences are basically statements of the same issue. Legacy Package models may be used with on-die BIRD189 models. - Arpad: Would T-coils be an example of this? - Walter: Exactly, one might use a T-coil to represent on-die effects but use a legacy package model. - [Reviewing changes relative to draft12] - New section added at the end of Usage Rules: for [Interconnect Model Group]. - A Group contains Sets and Sets contain Models. There are rules that apply to the combined list of Interconnect Models contained in a Group. - I/O pin_name rules: - Each I/O pin_name may appear only once as a victim. - Applies to Pin_I/O and Buffer_I/O terminals. - Eliminates confusion about which Model to use for a particular victim pin. - List of the valid interface combinations for BIRD189 models. - Pin to Buffer (full path) - Pin to Pad and Pad to Buffer (full path in two parts) - Pin to Pad (package only, on-die assumed to be shorts) - Pad to Buffer (on-die only, Legacy Package Model may be used). - In no Model at all (pin is not covered by a BIRD189 model). - List of the invalid interface combinations. (combinations that overlap) - Pin to Buffer and Pin to Pad. - Pin to Buffer and Pad to Buffer. - Pin to Buffer and Pin to Pad and Pad to Buffer. - If an I/O pin_name never appears as a victim and appears in one Model as an aggressor, then that Model may be used to simulate that I/O pin name, but the user should be aware that not all crosstalk info is included. - If an I/O pin_name never appears as a victim and appears in more than one Model as an aggressor, then there is ambiguity and the user or tool would need to make some choice. - Rail Terminal Rules: - A bit more confusing than I/O terminals. A rail pin can appear in a model as a pin or a rail or a signal name. - A rail terminal (e.g. VDD) can appear in a model for two reasons: - As a connection from a pin to a buffer. (connection in a PDN model) - As a reference. - As a reference it can appear in lots of models. As a connection from the pin to the buffer it can only appear in one model. - Each rail pin_name shall not appear as a Pin_Rail terminal in more than one PDN model in the Group. - Each rail pin_name shall not appear as a Buffer_Rail terminal in more than one PDN model in the Group. - Discussion: Bob noted that he generally agreed with the rules, except that he disagreed with the caveat that allows Pins to appear multiple times as Aggressors. Radek noted terms like "Legacy Package Model" had to be carefully spelled out. He noted that the enumerations of valid interface combinations were confusing. He also said discussion of falling back to legacy modeling approaches for Pins not included in a BIRD189 model was probably redundant. He noted these were just editorial issues. Walter agreed that there was editorial work to be done, and said he thought the Aggressors issue was the only technical issue to resolve. Arpad summarized the technical differences between draft11 and draft12_v1: 1. draft11 had already removed File_TS0. 2. draft11_v2 adds A_gnd for use with File_TS. 3. draft11_v2a adds the rule that subcircuit definitions should not contain node 0 and adds a warning to model makers to be careful about using node 0. 4. draft12_v1 is a glorified version of draft11_v2a in which the tables were also updated and the entire document was made consistent. Bob moved that draft12_v1 be posted to the Interconnect archive as draft12. Michael M. seconded. No one was opposed. Bob moved that Walter's new version be posted to the Interconnect archive as draft13. Mike L. seconded. No one was opposed. - Arpad: Thank you all for joining. AR: Walter to send draft12_v1 and his latest proposal to the ATM and Interconnect lists. Mike L. to post them to the Interconnect archives as draft12 and draft13 respectively. ------------- Next meeting: 19 December 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives