IBIS Macromodel Task Group Meeting date: 25 June 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 Sai Zhou 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 asked whether we should hold a meeting on July 2nd. Walter moved to cancel it. Ambrish seconded. There were no objections. The meeting scheduled for July 2nd will not be held. ------------- Review of ARs: Arpad: Send his C-phy presentation and Alaeddin's to the ATM list. - Done. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: Arpad asked for any comments or corrections to the minutes of the June 18th meeting. Ambrish moved to approve the minutes. Arpad seconded the motion. There were no objections. -------------- New Discussion: SPIM transient support in IBIS: Kinger reviewed a presentation on a proposal to add more to SPIM (Streamlined Power Integrity Model) to support transient analysis. He noted that BIRD223.1, approved in December of 2023, added SPIM and support for AC impedance and DC IR drop analyses. Kinger and Chi-te are now proposing additional information be added to SPIM to support transient analysis. [SPIM Transient Target] is a proposed new keyword that contains Vpp, Vmin and Vmax values. Kinger said the Vmin is the minimum voltage required for the circuit to function properly, Vmax defines the upper limit (e.g., a max limit on overshoot), and Vpp is the peak-to peak-noise that would affect Power Supply Induced Jitter. [SPIM icct File] is a proposed new keyword that contains the name of an IBIS-ISS file and an included subcircuit. The subcircuit would define the current profile to be drawn through the PDN. Kinger said they would like to add VCCS Poly (polynomial) to IBIS-ISS for this. Kinger described the stages of development for the icct model. He said the 0th order ideal icct profile is generated by connecting the Silicon model to an ideal voltage source. The 1st order modeling involves pulling the icct profile from a PDN model connected to an ideal voltage source. The 2nd order approach also pulls the icct profile through the PDN, but the VCCS used to generate the icct also incorporates the effects of the voltage drop through the PDN. To support this second order approach to the PDN testing, they propose using a VCCS Poly and an I PWL, both of which would have to be added to IBIS-ISS. The nominal icct current would be additionally scaled according to the voltage fluctuation on the power rail. Arpad noted that these PWL and Poly additions had been discussed on the reflector this week. He said that neither of them are currently available in IBIS-ISS. He said it probably wouldn't be too hard to add them, but we might need to ask permission from Synopsys, as the IBIS-ISS is made up of a subset of their netlist syntax. Randy said IBIS-ISS (ISS = interconnect SPICE subcircuit) was originally designed for passive interconnect. He said that's why it doesn't include support for these types of devices, and he expressed concern that people might abuse the definition of IBIS-ISS if we added these device types. Arpad, Randy, and Walter suggested that we might instead provide the PWL data and polynomial definition in the IBIS file (SPIM) and leave it to the EDA tool to implement the current source and connect it. Kinger agreed that this solution might work. Special characters in the Executable sub-param of [Algorithmic Model]: Walter said he was aware of one EDA tool that complained about the appearance of a plus sign ('+') in the Executable sub-param. Walter noted that everyone else he'd talked to thought '+' was perfectly legal. Arpad asked whether the specification explicitly states anything about it. Walter said he believed the specification said that '+' can't be used in a filename, but doesn't prohibit its use in the Executable sub-param. Arpad reviewed the specification's definition of alphanumeric, but he noted that the Executable sub-param is not restricted to alphanumeric entries. Arpad reviewed the Usage Rules for [Algorithmic Model] (IBIS 7.2, pg. 211). The Platform_Compiler_Bits description doesn't place any restrictions on the characters, except to state that '_' is the separator between the three fields. The Examples include '.' and '-' characters, and the group agreed that '+' is a legal character in the Executable sub-param. Ambrish noted that '+' is also legal in signal_name (IBIS 7.2, pg. 356, Example). Use of Reserved Words as Model_Specific AMI parameter names: Walter noted that "Usage, Type, Format ... are reserved names" in the .ami file (IBIS 7.2, section 10.3.4, pg. 237). However, other IBIS keywords, for example "corner", are not explicitly called out as reserved. Walter said that IBIS 7.2 leaves it to interpretation, and we might want to clear it up and state that "Corner", for example, can't be used as a Model Specific parameter name. Ambrish agreed that we shouldn't allow any of the reserved words in IBIS to be used as Model Specific parameter names. Arpad reviewed the General Rules for the .ami file (IBIS 7.2, pg. 234), and the group saw nothing that puts any further restrictions on Model Specific parameter names beyond the statement on page 237 mentioned by Walter. Arpad said he had once seen an AMI model with a Model Specific parameter named "Corner", which took additional options beyond normal corner values. Arpad said that if we want to make it clear we should add a bullet to the General Rules section explicitly stating that reserved words can't be used as Model Specific parameter names. He asked whether anyone wanted to write a clarification BIRD. Walter said he just wanted to find out what people thought about the current state of the specification. The group agreed that there is no clear answer, but the specification doesn't seem to restrict the names right now. Ambrish said that model makers should exercise some common sense when creating their Model Specific parameters' names. T-coil in [C Comp Model]: Walter noted that it's possible to model a T-coil in an AMI model by providing a Touchstone file in the Ts4file Reserved parameter. For a standard IBIS model, we could use [C Comp Model] to provide the Touchstone file for a T-coil. However, a T-coil model would require an additional series model connection to the output, and currently the [C Comp Model] output is always tied directly to the I/O buffer pad. Walter said he had drafted a BIRD in 2022 to address this requirement. He said the open question is whether we should proceed to work on that BIRD or wait for a pressing need from the industry. - Ambrish: Motion to adjourn. - Curtis: Second. - Arpad: Thank you all for joining. New ARs: Kinger: Send his "SPIM transient support in IBIS" presentation to the ATM list. ------------- Next meeting: 09 July 2024 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives