[ibis-macro] Re: [SI-LIST] Your blog post: Four Things That Drove Me Nuts About IBIS 5.1
- From: <colin_warwick@xxxxxxxxxxx>
- To: <twesterh@xxxxxxxxxx>, <si-list@xxxxxxxxxxxxx>
- Date: Fri, 26 Oct 2012 17:24:00 -0600
Hi Todd,
Thanks for the feedback, which I'll consider for the next version. I hope to
converge on perfection. :-)
Let me explain the (possibly misguided) rationale behind the present version:
The "Case 1-5" labels I used were intend as shorthand internal to the document.
In the caption of Table 2, I even said explicitly:
"For the purposes of this article, let's give them names --Cases 1-5, but
bear in mind that IBIS 5.1 doesn't use these names."
A book critic reviewing a book doesn't feel obliged to agree with everything
the author said. Similarly here, I was trying to give a different perspective
aimed at newcomers. Simply replicating the language of the standard would have
made our paper redundant.
That said, I do like your proposal for naming the four model flag combinations,
specifically:
a) "Illegal" for Init_Returns_Impulse=FALSE & GetWave_Exists="FALSE"
b) "LTI-only" for Init_Returns_Impulse=TRUE & GetWave_Exists="FALSE"
c) "NLTV-only" for Init_Returns_Impulse=FALSE & GetWave_Exists="TRUE"
d) "NLTV/LTI" for Init_Returns_Impulse=TRUE & GetWave_Exists="TRUE"
...I'll adopt those in the text below...
I agree that the combinations where one or both models are "illegal" divides
the 32 combinations (four flags by one simulator mode setting) into 18 legal
and 14 illegal combinations.
The 14 illegal combinations are color coded red in my table:
http://signal-integrity.tm.agilent.com/wp-content/uploads/2012/09/IBIS_AMI_5_1_combinations.pdf
...the 18 legal ones are colored green or yellow.
Although there are nine 9 legal stat mode combinations, five of those involve
one or both models being "NLTV-only" which are implicit pass-throughs in stat
mode, and therefore not very useful IMHO. I didn't consider them in the paper.
In addition, in stat mode a "NLTV/LTI" model behaves like an LTI-only model
because stat mode ignores GetWave, hence my coining "Case 1", which is a short
hand for mode=statistical & Tx Init_Returns_Impulse=TRUE & Rx
Init_Returns_Impulse=TRUE
Similarly in bit-by-bit mode, "NLTV/LTI" model behaves like a "NLTV-only"
model because bit-by-bit mode ignores the LTI in favor of GetWave for a
"NLTV/LTI" model. So you can boil down the nine legal cases to four (my "Cases
2-5") with each Case corresponding to a particular Tx and Rx GetWave_Exist
setting pair.
Hope this helps!
-- Colin Warwick
Product manager for High Speed Digital, Agilent EEsof EDA
...feeds blog @ http://Signal-Integrity.TM.Agilent.com/feed/
...tweets @signalintegrity
...it's sooo 19th century, but he even sends and receives phone calls @ +1 978
681 2406
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Todd Westerhoff
Sent: Friday, October 26, 2012 9:57 AM
To: si-list@xxxxxxxxxxxxx
Cc: ibis-macro@xxxxxxxxxxxxx
Subject: [SI-LIST] Your blog post: Four Things That Drove Me Nuts About IBIS 5.1
Colin,
Your paper "Explore the SERDES design space using the IBIS AMI channel
simulation flow" and blog entries redefine pre-existing terms, which I think is
going to increase confusion, not reduce it. If you want to clarify IBIS 5.1
flows, I think you should stick with the terminology the committee already uses.
There are, indeed, 4 possible combinations of the parameters
"Init_Returns_Impulse" and "Getwave_Exists", but only three possible model
states, because invalid states don't count.
Your paper redefines the word "Case" and proposes independent case numbering
for Statistical and Time-Domain analysis. That's not how the spec works. There
are 3 legal states for an AMI model, thus 9 possible combinations (cases) of TX
and RX models. Each case is completely defined for both Statistical and
Time-Domain analysis. That makes 9 flows, not 32.
This was laid out in the July 2010 presentation "IBIS-AMI Flows", which is on
the IBIS-ATM committee website at http://tinyurl.com/28bj9nm. This is where
the equations for the 9 different AMI flows in IBIS 5.1 were originally spelled
out. The text on pages 158-162 of the IBIS 5.1 spec describes how these
equations are realized for Statistical and Time-Domain simulations in step by
step fashion. If you choose to create new material that presents IBIS-AMI
flows, I think you should be consistent with the committee's prior work and the
written spec.
One area that's probably worthy of new discussion is how we name the different
AMI model states in a user-friendly manner. Those of us who are close to IBIS
have become accustomed to the definitions
Init = Initialization and linear processing,
Getwave = Nonlinear and time-varying waveform processing
And thus tend to refer to models as "Init-only", "Getwave-only" and "Dual or
Advanced". If you think associating "LTI" with "Init" and "NLTV" with "Getwave"
will help reduce confusion, then I think you should propose an IBIS BIRD to
that effect. I'll even help you write it.
Todd.
Todd Westerhoff
VP, Software Products
Signal Integrity Software Inc. . <http://www.sisoft.com/> www.sisoft.com
6 Clock Tower Place . Suite 250 . Maynard, MA 01754
( <callto:978%29%20461-0449> 978) 461-0449 x24 .
<mailto:twesterh@xxxxxxxxxx> twesterh@xxxxxxxxxx
"I want to live like that"
-Sidewalk Prophets
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
---------------------------------------------------------------------
IBIS Macro website : http://www.eda.org/pub/ibis/macromodel_wip/
IBIS Macro reflector: //www.freelists.org/list/ibis-macro
To unsubscribe send an email:
To: ibis-macro-request@xxxxxxxxxxxxx
Subject: unsubscribe
Other related posts:
