All,
The minutes of the task group meeting held this morning are in the attached
file.
Regards,
Weston
IBIS Editorial Task Group Minutes
October 5, 2022
Attendees:
Intel Corp. Michael Mirmak*
Mathworks Walter Katz*
Micron Technology Randy Wolff*
Siemens EDA Arpad Muranyi*, Weston Beal*
Teraspeed Labs Bob Ross*
Minutes by Weston Beal
--------------------------------------------------------------------------------
Call for Patents: none
Review of Minutes:
Minutes from October 5, 2022 were not reviewed since they were not available.
Review of Actions Required:
Michael Mirmak reviewed ARs from the October 5, 2022 minutes. The minutes
were distributed at the beginning of the meeting.
Michael researched the distributed file editing in MS Word. Inserting a
sub-document did not work. The problem might be a specific Word version or
this document. He plans to investigate more.
Michael to convert Visio images to other format for including in the
specification document.
remains open
Arpad made the BIRD for clock_times as draft 3 and sent to the email list.
done
New Actions:
- Walter Katz will discuss the programmers use of clock_times memory to see if
the specification needs to clarify the process of the array passing.
- Michael will ask the EDA tool vendor who violates the AMI_parameters_in
behavior of passing output into the parameter string if they are willing to
correct this in their tool.
Agenda:
Arpad reported that the ATM task group discussed the clock_times issue to be be
clarified. The issue is not resolved yet and needs more discussion. Walter
stated that the time values vector should cover the same time range as the
concurrent waveform vector. This is an output from a DQS model. The DQS model
is almost surely created by the same creator who creates the DQ model. This
creator should verify that the 2 models work together. The difference is that
the specification assumes that the memory pointer from DQS output is the same
memory pointer passed into the DQ model. Arpad reported that HyperLynx SI
accumulates all the clock_times from the full time simulation of the DQS model
and in a subsequent simulation breaks the clock_times array into blocks to pass
into the DQ model. Walter [AR] will discuss this to see if the specification
needs to clarify the process of the array passing. Randy asked about the case
when the clock_times output might go past the end of the input waveform time
range. Does the EDA tool need to buffer these times to maintain contiguous time
into the DQ model? Walter commented that this buffering of time should occur in
the GetWave memory storage of the DQS model.
This discussion of clock_times will continue in the ATM task group instead of
in editorial meeting. Any resulting BIRD will then be presented to the
editorial task group for editing the specification.
AMI_parameters_in: Michael presented an apparent inconsistency in the
description of this parameter. In some places it is described as _all_
parameters and sometimes described as _some_ of the parameters. The confusion
seems to arise from the word, input. Michael sent an email with some suggested
changes to the descriptions to make the descriptions more explicit. Michael has
seen cases where EDA tools interpret the usage differently and behave
differently. The changes to the specification will not require specific
behavior so that currently working tools and models are not invalidated. The
change suggested by Michael will say what a well-behaved tool _should_ do.
Michael [AR] will ask the EDA tool vendor who violates this behavior of passing
output into the parameter string if they are willing to correct this in their
tool. If so, then we don't need to change this requirement by a BIRD. The
discussion will go through the ATM email list to reach a wider audience.
comments on draft2: There were a few comments from reviewers. The word bits and
UI need to be clarified. In this instance, the word bits in correct because the
sentence described calculated bit error rate (BER).
Next Meeting Scheduled:
October 19, 2022
Weston moved to adjourn. Arpad seconded.
Meeting adjourned at 8:52 a.m. PT.