Minutes from the 6 November 2018 IBIS Editorial Task Group meeting are attached.
Regards,
Justin
================================================================================
IBIS EDITORIAL TASK GROUP
http://www.ibis.org/editorial_wip/ ;
Mailing list: ibis-editorial@xxxxxxxxxxxxx
Archives at //www.freelists.org/archive/ibis-editorial/
================================================================================
Attendees from November 6, 2018 Meeting (* means attended at least using audio)
ANSYS Curtis Clark
Cadence Design Systems Bradley Brim
Intel Corp. Michael Mirmak*
Keysight Technologies Radek Biernacki*
Mentor, A Siemens Business Arpad Muranyi*
Micron Technology Justin Butterfield*, Randy Wolff
SiSoft Walter Katz, Mike LaBonte
Teraspeed Labs Bob Ross*
Michael Mirmak convened the meeting. No patents were declared.
Justin Butterfield took minutes.
Review of minutes from the October 31 meeting:
Bob Ross noted a spelling error in the minutes, but he did not have the exact
word. Bob moved to approve with this correction. Arpad Muranyi seconded.
Minutes were approved without objection.
Review of ARs:
- Michael to send out the latest draft and checklist.
- Michael reported this was done and thanked Mike LaBonte for posting.
Opens:
- Radek Biernacki suggested we could discuss Michael's question on the AMI file
example. He agrees with the email.
Editorial review of ver7_0_181106.docx and the task checklist:
Michael noted we still need to resolve the default values in the summary
tables. In Table 40, the Required column issues have been resolved, but we
need to resolve the Default column. There are some descriptions of the values
in the footnotes, but the question is if this is sufficient. We also have some
entries of "None" and double dash ("--"), which need to be resolved.
Bob stated he likes "Undefined" as an entry in the default column. Radek
commented that "None" and "Undefined" are okay and can mean the same thing in
this context. He stated double dash is okay for required parameters. Bob
agreed double dash is fine for required parameters and commented this makes the
table more readable. Michael noted we are consistent with having double dash
for required parameters. He asked about the three cases of double dash where
the parameter is not required. Arpad replied the double dash is independent
from required, as it means if the parameter is supplied we have to give a
value. Bob noted we might modify the meaning of double dash to include this.
Radek thought that non-required parameters should have "None" rather than
double dash. Arpad commented that the footnote Michael added addresses this
issue and suggested we should move on.
Arpad asked if the "None" and "Undefined" mean the same thing and if we have
any language in the specification text on this. Michael noted the summary
table should match the other tables, but the default stated in the text can
mean something else. He asked if the text for Tx_Sj_Frequency implies any
default, for instance. Arpad replied there is no mention of default for
Tx_Sj_Frequency. He proposed to change the Tx_Sj_Frequency default to "None".
Radek agreed. Bob thought that "None" has a double meaning of none defined or
zero. Arpad stated the meaning of zero is clear, but if we say "None" the
model does not tell you how the parameter is defined. Bob commented if a
parameter that has not been defined is used in a calculation the tool should
issue an error. But, if you state that "None" can be zero, then the simulation
can use zero. Arpad stated, if an IBIS AMI file does not define a parameter,
the EDA tool can use the value it wants.
Michael asked for Rj, Sj, DCD, and Dj if we should have "None" for the
default. Arpad asked what the text for these parameters says. Michael checked
the text and noted that there is no text about what the parameter should be if
it is not defined.
Bob commented, if a parameter is not specified, then a tool should not add the
effect, but if a parameter exists, the tool could override the value. Michael
thought we should be careful with this. Radek noted this might not be an
editorial change if we were to add text stating this. Arpad gave an example in
his tool where features for adding jitter in the GUI were added before these
parameters were officially available. These settings are only available if the
parameters are not defined. Bob thought it would be okay for the tool to
override these parameters. Michael asked if we need a BIRD to explicitly state
these rules. Arpad thought we could add text to the parameters to describe
this.
Arpad noted that the issue of how to handle things when missing is a classic
problem with IBIS. Michael proposed to not make any changes to the table. He
noted the defaults of zero are clear for use in the equations. Arpad asked if
it would be better to change "None" to "Undefined". Michael agreed and made
this change. He went through each of the parameters that were changed to
"Undefined" to confirm it makes sense. Arpad commented we should check the
text of Rx_Clock_PDF. Radek stated it is okay to say "Undefined" for
Rx_Clock_PDF. Michael asked about Tx_Sj_Frequency and if there is any
assumption about the frequency. Radek stated there is not. Michael also went
through all the zero value defaults in the table. He noted Ignore_Bits,
Max_Init_Aggressors, and PAM4_centerthreshold are defined in the text with
defaults of zero. Michael thought the parameters with zero for defaults look
good. He asked if the double dashes used for Special_Param_Names,
Supporting_Files, Ts4file are correct. Bob proposed to change these to
"Undefined". Arpad agreed noting these are cover in footnote 2. Arpad asked
if we should add a superscript for footnote 6. Michael agreed and added this
to Default column header.
Michael to update the other tables with these changes [AR].
Michael commented he will save a ver7_0_181106.docx draft, but he will not
send it out as there is only a short time until the next meeting.
Arpad moved to adjourn. Bob seconded. The meeting adjourned without
objection.
The next meeting will take place Wednesday, November 7, at 8 PM Pacific.
Open Technical Questions:
1. BIRD182: POWER and GND [Pin] signal_name as [Pin Mapping] bus_label
a. Is a bus_label created even if we don't have [Pin Mapping], [Bus Label],
or [Die Supply Pads] through the [Pin]?
b. Is a bus_label short created for legacy package models based on the
second column of [Pin] when we do not have a [Pin Mapping] entry?
