[ibis-interconn] Minutes, Jan. 23 IBIS-Interconnect Task Group and Feb. 6, 2013 Agenda
- From: "Mirmak, Michael" <michael.mirmak@xxxxxxxxx>
- To: "IBIS-Interconnect (ibis-interconn@xxxxxxxxxxxxx)" <ibis-interconn@xxxxxxxxxxxxx>
- Date: Wed, 6 Feb 2013 06:03:03 +0000
======================================================================
IBIS INTERCONNECT TASK GROUP MEETING
http://www.eda.org/ibis/interconnect_wip/
Mailing list: ibis-interconn@xxxxxxxxxxxxx<mailto:ibis-interconn@xxxxxxxxxxxxx>
======================================================================
Next meeting: Feb. 6, 2013
8 AM US Pacific Time
Agenda:
Attendance
Call for Patents
Agenda and Opens
Review of Task Group Status after Summit, ATM Meeting
Next Meetings' Schedule/Agenda:
* Feb. 13
* Feb. 20
For international numbers, please contact Michael Mirmak.
Note: in case of issues with Lync we will use the WebEx noted at the bottom of
this message
.........................................................................................................................................
Join online meeting<https://meet.intel.com/michael.mirmak/QZ193W0C>
https://meet.intel.com/michael.mirmak/QZ193W0C
First online
meeting?<http://r.office.microsoft.com/r/rlidOC10?clid=1033&p1=4&p2=1041&pc=oc&ver=4&subver=0&bld=7185&bldver=0>
[!OC([1033])!]
.........................................................................................................................................
<--- Reservationless Bridge - Do not edit or remove ---
916-356-2663, 8-356-2663, Bridge: 2, Passcode: 8625431
Speed dialer: inteldialer://2,8625431
----------------------------------------------------------------->
Note: in case of issues with Lync we will use the WebEx noted at the bottom of
this message
======================================================================
Attendees, Jan. 23
Agilent Technologies Radek Biernacki*
Altera David Banas*
ANSYS Luis Armenta, Steve Pytel
Cadence Design Systems Brad Brim, Ambrish Varma
Intel Michael Mirmak*
Mentor Graphics Arpad Muranyi*
Micron Technology Justin Butterfield*, Randy Wolff*
QLogic Jason Zhou
Signal Integrity Software Walter Katz*
Teraspeed Consulting Group Bob Ross*
Minutes
No patents were declared. No opens were raised.
Michael Mirmak summarized the proposed Summit material from Walter Katz. Bob
Ross stated that the thrust of this presentation is EMD, which hasn't been
discussed in a summit before. Michael agreed, suggesting that a separate
presentation on EMD would generate interest, discussion and expansion. Another
slide set would be provided to show Interconnect Task Group progress.
Walter suggested that the final/summary EMD slide be used as the introductory
Task Group summary slide. Walter suggests that a topology slide from a
previous presentation from ANSYS to show issues facing the industry.
Additionally, a list of presentations should be provided. Michael volunteered
to complete the slides and present them.
Walter requested that his document stating the relationship between "MCI" (a
generic term) and EMD be posted to the web page. He referred to a statement
from Brad Brim regarding "MCI" as a tool for IC vendors to build an EMD file.
It is unknown whether this applies to Si2 efforts or not.
Bob asked whether the Si2 efforts are still on-track. Walter noted that the
dividing line between on-die interconnect and the buffer model is the
fundamental issue.
Walter pointed out the IBIS 5.1 limitation slide in his EMD presentation. The
relationship between the buffer and bump pad being a single point/node rather
than the reality of a separation between the bump pad and the buffer. For AMI,
the common practice is to put this information into a Touchstone file; in AMI,
there's no such thing as a high or low state (in contrast to [Model]). Michael
added that there's interaction between ATM and this group because of
controversy regarding how that analog information is tied into AMI and [Model],
including TX stimulus. Walter stated that C. Kumar had earlier pointed out
that the impulse response of the channel should not constrain how it is
generated; the user can generate it any way he wants. It wasn't clear though
how to generate the analog models for the TX and the RX and how they interact
with the impulse response.
Bob asked when you say "channel", this includes the package portion of the TX
and RX and the interconnect? Walter replied not, it also includes the analog
model that the IC vendor says to use at both the TX and the RX. If you are an
IBIS person (as opposed to an S-parameter person), you need a [Model]; in any
case, it must include the reactive portion of the TX and RX that will affect
the generation of the TX and RX. James's ATM comment was that the IC vendor
decides where that point is in the silicon. Walter wants to add to that that
the algorithmic model is to the right and analog modeling is to the left. The
IC vendor has to supply the partial data; a classic example is a peaking
filter, which is not reactive. It can be in either section, but it can't be in
both sections simultaneously.
Bob noted that one interpretation of AMI is that the IBIS AMI paths can be
interpreted as separate or having separate paths than traditional IBIS. Actual
signals in analog AMI are representative, whereas the AMI flow may include a
bit pattern beyond what a standard model provides . Walter described a
traditional two-tap TX approach with Driver Schedule and stimulus. Nowadays
hundreds of submodels would be needed to sequence modern buffers.
David Banas reported on his GetWave survey results - 9 total respondents; all
but 1 answered "A" (input to analog channel); the sole discrepancy was Arpad
Muranyi, who answered F. Radek Biernacki noted that question A was not 100%
precise; the intent was the combination of TX backend, package, interconnect
channel, etc.; that could be interpreted differently, so answer F could be
acceptable. David replied that the group could use a good reference diagram.
Bob added that output of GetWave is to the input of the channel; therefore,
that includes the TX analog model. Radek replied that the convolution of a
single output response is subject to actually being able to perform the
convolution, so we need an ideal voltage source to drive the channel. Those
three pieces are driven by a voltage source, or at least something like one.
Model makers decide where to put that point. The output of GetWave needs to be
convolved with impulse response. Michael summarized responses he has received
in the past on real number inputs to SPICE-style IBIS implementations. Walter
noted that this is the crux of the discrepancy between visions of the TX
buffer: an S4P representation of buffer in SPICE vs. numerical calculation
environment approach.
Radek asked, in analog simulators like SPICE, you need an analog source to
drive the buffer; the intent, however, was to have a digital stimulus to toggle
the buffer. How is that implemented in an analog simulator? With an analog
source, the trigger events occur when you cross a threshold. In *that*
context, Radek agrees with Arpad. The analog portion comes from how we have to
handle the AMI simulations. We are not clear on descriptions and relationships
for all cases; if you have S4P, ports are clear, but if you have an IBIS
buffer, it's not particularly clear. Walter replied that the reason why
5.0/5.1 is in fact correct for traditional models is because we are using IBIS
with a step response and measuring/creating an impulse response for a
*particular* voltage swing, the one that the buffer can supply to the rest of
the channel. AMI assumes LTI, so this impulse response can be scaled to match
the actual response of GetWave.
David noted that four categories were identified: EDA, silicon vendor,
consultant, ?; 5 EDA vendors, 2 vendors, 1 consultant, 1 ?. On question 2,
there were 5 answers for "normative", 2 answers for "informative"; 2 for a mix
(on Section 10). Walter summarized the differences between normative and
informative.
David stated that a diagram will be sent out to stimulate discussion. Walter
observed that the derivation methods are in the specification; this is murky on
informative vs. normative. Walter suggests specific sections be labeled
normative vs. informative; David seconds.
======================================================================
In case of Lync issues only, we will switch to WebEx as noted below.
Meeting Number: 732 940 715
Meeting Password: IBIS
-------------------------------------------------------
To join this meeting (Now from mobile devices!)
-------------------------------------------------------
1. Go to https://sisoft.webex.com/sisoft/j.php?J=732940715&PW=NNWY2NmRmZTY0
2. If requested, enter your name and email address.
3. If a password is required, enter the meeting password: IBIS
4. Click "Join".
5. Follow the instructions that appear on your screen.
-------------------------------------------------------
Audio conference information
-------------------------------------------------------
Call-in toll number (US/Canada): 1-650-479-3208
Access code:732 940 715
http://www.webex.com
IMPORTANT NOTICE: This WebEx service includes a feature that allows audio and
any documents and other materials exchanged or viewed during the session to be
recorded. By joining this session, you automatically consent to such
recordings. If you do not consent to the recording, discuss your concerns with
the meeting host prior to the start of the recording or do not join the
session. Please note that any such recordings may be subject to discovery in
the event of litigation.
Other related posts:
- » [ibis-interconn] Minutes, Jan. 23 IBIS-Interconnect Task Group and Feb. 6, 2013 Agenda - Mirmak, Michael
