IBIS Macromodel Task Group Meeting date: 04 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 noted that he is now affiliated with SiSoft -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad ask Radek to propose a methodology - Done ------------- New Discussion: Arpad: BIRD 140.1 is on the ATM website Backchannel BIRD: - Ken was not present Arpad: BIRD 127.2 was brought back to this committee - Radek: There are just a few issues - The Usage language needs to be more neutral. - Arpad: It says "optionally" but it is not optional if it is a Reserved parameter. - Bob: Does this shut the door for Model_specific params? - Radek: It just doesn't matter if they are reserved params - Bob: Tap params are already reserved - Bob moved to accept the change made today - Curtis seconded - The motion passed by acclamation Arpad scrolled to the pg. 141 changes to the Rx_Clock_PDF example - Radek: The param is float but the Table contains string labels - Bob: We made an exception for Labels - This is allowed - Walter: Type should be Integer, UI, Float - David: Are Labels singular or plural? - It doesn't seem to work singular - Walter: In 5.0 it is plural - Arpad: Any parser issues like this should be reported - David: It is a tool issue - Arpad scrolled to the Tx_Jitter section - It says the first column of a Table is a string param name AR: Arpad look for Table column type issues in BIRD 127.2 Arpad scrolled to the section on Default - Radek: It needs to also have the exceptions used for AMI_parameters_out AR: Arpad check on AMI_parameters_out exceptions and maybe send email Arpad showed the "Clarify sample intervals in IBIS-AMI" BIRD proposal - Arpad: The specification intends for any sample rate to be supported - Enhanced description of impulse_matrix parameter to say more about sample spacing - David: Is it a single matrix, no aggressors? - Arpad: That is a later change Arpad: 3.1.2.4 would add more about sample_interval - Radek: It is not precise to say "data rate" here - David: We are not making a "must " constraint here - Fangyi: Should it be a fraction of the bit time? - David: It should not be any definite requirement - Walter: It is a fraction of the bit_time - Fangyi: Do we allow fractions like 64.7? - Walter: Kumar said we should treat the waveform as continuous - sample_interval should at least be a rational fraction of bit_time - It is well understood how to convert time intervals - Fangyi: The DLL will have to convert again on output - What if a bit pattern is out of the device range? - Walter: DLLs can have speed limit checking for bit rates the device will not support - Arpad added that the DLL can report an error for unsupported sample intervals - Do we need to change lowest_bit_time? - Walter: Yes it should be just bit_time AR: Arpad update BIRD 127 BIRD 140: - Arpad: This has been posted on the website for weeks - The ** section about pg 141 is the trouble - Bob: "In IBIS-AMI" should be "In IBIS" - Format Corner is used and values are 0, 1, 2 - Arpad: This is looking ahead to dependency table - The selector is actually hidden from the user - Radek: That is confusing - We use "corner" for different things - Bob: It is making forward statements - Arpad: I see no forward statements - Bob: We have not yet defined Typ/Min/Max for Table yet - Walter: This can be clarified in the Dependency BIRD - No one is confused by this now - Bob: Agree - We can deleted it for this BIRD - Arpad: We should leave it and pick it up for 5.2 - We can take it off the plate for 5.1 - Bob motioned to delete paragraph 2 - Fangyi: What are we trying to solve? - Arpad: It says "align implicitly to slow and fast corners", but doesn't define them - Walter: Maybe it can say "align with IBIS corners" - Radek: It is a future extension and should be deleted - Arpad: For the use to select it should be a parameter - We should not have two ways to do the same thing - Radek: If we define it now it will have to be supported forever - Fangyi: The issue is about how to select, not who selects - Because of the C_comp issue we have to make it clear what association must not be made - The user should know how the data was generated - Radek: When I need to run Slow I don't know what values to use - Walter: The IBIS spec makes that clear, but uses different column names - Fangyi: That is logically clear - We need to separate that you end up with different individual values - Radek: A strict association may not be right - Arpad: I proposed another solution, with 5 values - It allows users to forcibly select min or max in addition to slow and fast - Fangyi: We do not define them well enough - What is the meaning of "min"? - Arpad explained the "C_comp problem" - Fangyi: The user chooses a condition, and the model choose the settings - Bob: The 5 choice proposal is a step backward - The biggest problem is just C_comp - Slow=max and fast=min works 90% of the time - Walter: Should we remove the C_comp min < max restriction? - Then the fast value could be put in for max, etc. - Radek: That would make IBIS strictly slow and fast - Arpad: There is a statement more or less to that effect - Walter: We could get rid of Typ/Min/Max altogether in this section - Arpad: IBIS defines corner names - In [External Model] we have a definition of corners - Fangyi: Does the spec define how model makers set the values? - Arpad: It is in section 9 - It says params like C_comp do not correlate to I-V and V-T curves - It is then unknown which values to use for fast or slow - Fangyi: At least it is clear when we say we are not solving a problem - It is confusing when we do not solve a problem and don't say it - Can we explain the meaning in section 9? - Arpad: The meanings are already described in section 9 - Fangyi: It doesn't say slow and fast - Arpad: It says high/low voltage and strong/weak - Fangyi: Why does Format Corner have 3 values in one place and 5 in another? - Arpad: If it is only 3 slow and fast have to be decoded from min and max - Walter: There would have to be a 6th "Typ" value - The "Typ" corner might actually require the min value - Arpad: Added a 3rd proposal with Typ/Min/Max_performance labels ------------- Minutes by Mike LaBonte Next meeting: 11 Oct 2011 12:00pm PT Next agenda: 1) Task list item discussions ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives