IBIS Macromodel Task Group Meeting date: 15 August 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 Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. -------------------------------------------------------------------------------- Opens: - Arpad: We might discuss unused interconnect terminals. ------------- Review of ARs: - Radek to update the AM2 version of BIRD158.6 draft 2 to create a draft 3. - Almost done, can show something today. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - Arpad: Does anyone have any comments or corrections? [none] - Walter: Motion to approve the minutes. - Randy: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: BIRD166.4: - BIRD158.6: - Radek showed his updated BIRD 158.6 draft 3. - Radek described the changes, noting that previous change markup had been accepted. The TX figure shown previously has been updated, changing the source and adding pad and pin. The text is somewhat modified too. The Rx figure is also modified, similarly to the Tx. The Rx text is only slightly modified. The AMI flow figure is modified to remove the box tying channel to packages. The AMI flow text is heavily modified to specify that Ts4file is for die only. New parameters Ts4file_Package_Options and Ts4file_Package_Data are added. - Arpad: Is that renamed? It was Boundary before. - Radek: These are new parameters. - Arpad: They should not be called "new parameters" here. That may seem awkward in a few years. - Radek: Agree, I will change it. - Bob: It would help to have "draft 3" inside the document. Why have Boundary when we already have the directive? Do they work together or not? - Arpad: Could they conflict? - Radek: It is explicitly illegal to have both. - Bob: This replaces the [Model] keyword. - Arpad: I'm not sure we should be able to move the boundary to the pin here. - Radek: The models for both might be in one file. - Arpad: Package and die are usually extracted separately. - Walter: This is an area of premature optimization. The package model could be a short. We should work with the package model we have. - Arpad: If the IBIS file has non-zero package model what happens? - Radek: It explicitly says to ignore that if the AMI file has it. - Arpad: If there is coupling in the IBIS file we can't ignore the coupling. - Bob: If AMI defines package model syntax, that complicates AMI. - Walter: AMI has no concept of package models. We should remove the pin option from this. BIRD 189 can provide this capability. - Radek: I would like people to review this BIRD draft. - Ken: Does this relate to power connections? - Radek: No - Arpad: It is limited to 4 port models. - Ken: Does this preclude modeling power parasitics? - Radek: No IBIS still covers that. - Arpad: If the power system can influence it then we have a non-LTI system. - Walter: We would need to clarify that there is no power coupling. - Ken: This can already be done with [External model], no BIRD needed. - Walter: IC vendors do not like [External model]. - Radek: Ambrish might weigh in on this. - Walter: One use for [External Model] is handling additional reactive elements. But SerDes vendors are using Touchstone to handle T-coils. - Ambrish: Does this take precedence over External Model? - Radek: AMI takes precedence. - Walter: This applies to impulse matrix calculation for crosstalk terms. Pin to buffer coupling needs to be handled there. The package introduces crosstalk. - Arpad: It specifies 4 port s-parameters. You can't do coupling. - Walter: You can't do on-die coupling. - Arpad: Then not package either. - Walter: The pin boundary should be removed from this BIRD. - Radek: It might not be a prevailing use, but there is no harm to have it. Bob Miller wanted to give package data. This was put in for posting so that we can decide. - Bob: Are coupled package models used for AMI? - Walter: Yes. - Ken: Interposers can be in the string too. - Walter: That might be EBD, someday EMD. AR: Radek send updated BIRD 158.6 draft 3 C_comp_model_Using_IBIS_ISS_BIRD: - Randy showed C_comp_model_Using_IBIS_ISS_BIRD_rev1_WIP.docx - Randy: The Mode rules have been updated. This might be handed similar to submodel, with driving and non-driving modes. - Arpad: Would there need to be two models? - Walter: When driving you have VT curves, otherwise only IV. Really there are two models. So far we have used [Model Selector] for this. That is supported today. - Bob: Not every tool supports [Model Selector], we have Model_type I/O for this. I don't recommend [Model Selector] for this. C_comp can be different across the modes. - Walter: [Model Selector] handles that. Each [Model] has its own C_comp. - Bob: This would allow bringing in different C_comp models. This resolves the problem that the receiving point is not the driving point. - Radek: There is a third option, to have both. - Bob: In some DDR devices terminators are switched in only for receiving mode. - Arpad: I would like the "all" mode. If the same model is used for both modes it would be better not to need to duplicate the model. - Randy: Tools already deal with models that only have pullup/pulldown references defined. connect the pullup and pulldown terminals to the clamps. - Mike L: Does this need to handle transitions from driving to non-driving states? - Michael M: We should consider IBIS handling protocol sensitive simulations, but it's not easy. AR: Randy to send C_comp_model_Using_IBIS_ISS_BIRD_rev1_WIP.docx to Mike L for posting. - Walter: Motion to adjourn. - Arpad: Second. - Arpad: Thank you all for joining. ------------- Next meeting: 22 August 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives