Hi Arpad,
The primary motivation is completeness of the specification and
clarification of the terminology when we include Interconnect Modeling
connection features to a pin. Rather than trying to defend a need, the
easiest solution is to make the three interfaced available in case there are
needed.
I cannot offer practical guidance regarding why a model-developer would
specify a Buffer or Die location. The practical measurements based on
interfacing to the outside are at the Pin interface. The data book
specification values associated with the Pin locations should be
conservative to deal with different packages.
However, the Buffer (Die) location might be selected because the signal at
the buffer during simulation is closest to where an electrical measurement
parameter has meaning. The Interconnect Model Pad location can be
available by the structure. Pad might be an intermediate location, but I
cannot defend why it might be selected.
-------
BIRD158.6_draft2 has a related concern. Only "Buffer" terminals (TX 2, 4;
RX 1,3) are named in the reference figures for the Ts4file associated
ports.. "Pad" is not shown nor defined. Perhaps it is named in the context
that an actual physical or even simulated Ts4file extraction might include
some on-die interconnect behavior. I think the default should be buffer, as
shown in the figures. This pad location might be informative. However,
"pad" should not be confused with an Interconnect Model location, nor
require the Interconnect Model to be defined. Some clarification might be
needed.
Bob
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Wednesday, June 28, 2017 11:36 AM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] A question about BIRD 191
Bob,
I was trying to find the motivation for this BIRD in the introduction
section, where you wrote this:
The Interconnect Modeling BIRD189 defines three separate interfaces as
Buffer and Pad (for die pad) and Pin. The [Component] subparameters for
Si_location and Timing_location only provide Die and Pin options. A Buffer
option is needed along with the rule that without the Interconnect Modeling
keywords, Die and Buffer are at the same location.
I understand that with the new interconnect/package BIRD we can basically
plot three different
nodes, pin, pad, buffer, but what is the practical need to be able to plot
at the pad and/or the
buffer? In other words, when and why would an SI engineer want to see the
waveform at the
pad or the buffer. Do they really need to be able to look at either one?
Thanks,
Arpad
===========================================================================
From: ibis-bounce@xxxxxxxxxxxxx [mailto:ibis-bounce@xxxxxxxxxxxxx] On Behalf
Of Mike LaBonte
Sent: Wednesday, June 28, 2017 11:53 AM
To: ibis@xxxxxxxxxxxxx
Subject: [ibis] BIRD 191, Adding Buffer Location to Si_location and
Timing_location
BIRD 191, Adding Buffer Location to Si_location and Timing_location
<http://ibis.org/birds/bird191.docx> , has been submitted by Bob Ross of
Teraspeed Labs. It will be introduced as a new BIRD in the next IBIS Open
Forum teleconference.
ID#
Issue Title
Requester
Date
Submitted
191
Adding Buffer Location to Si_location and Timing_location
<http://ibis.org/birds/bird191.docx>
Bob Ross, Teraspeed Labs
June 28, 2017
Mike LaBonte
SiSoft
Chair, IBIS Open Forum
http://www.ibis.org/
