Arpad,
There are a large number of e-mails relating to optimization and IBIS
models. I pulled out the following quote from this e-mail from you to me on
12/1/2010
1. However, let’s consider an example when the Init function contains
an optimizer which takes initial values for the tap coefficients
and after some number crunching returns better coefficients.
I would think that in this case the tap coefficients will be
declared as InOut arguments, and the initial values given to the
Init function will be overwritten by the Init function when the
optimizer is done. Now, how is this result propagated to the
GetWave function? Notice that the GetWave function doesn’t have an
AMI_parameters_in argument. This makes me think that the
AMI_parameters_in argument of the Init function is also visible to
the GetWave function, and the modified tap coefficients from the
Init function are passed to GetWave through the AMI_parameters_in
memory location. This implies that the return value of an InOut
argument should be returned in place (which is #1 above). Using
this thinking we won’t need the AMI_parameters_out argument for
returning the values of InOut arguments.
This describes exactly how Bob Miller’s models work, and it is in words you
wrote.
There are tons of references to this in many e-mails. It is clear that the
original intent of IBIS AMI models were to enable the AMI_Init, the
AMI_Getwave and both combined to optimize tap coefficients.
I am also included a random relevant e-mail from 2009 from Kumar.
There are literally hundreds of e-mails between 2007 and today talking about
various methods of optimization. It is very interesting to see some of these
discussions in retrospect.
Walter
Walter Katz
wkatz@xxxxxxxxxx <mailto:wkatz@xxxxxxxxxx>
Phone 303.449-2308
Mobile 303.335-6156
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Monday, June 19, 2017 8:18 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: Response to BIRD-190
Aren’t those Model_Specific parameters? If so, no one else but
the model maker knows what they mean or can be used for…
Thanks,
Arpad
==================================================
From: Walter Katz [mailto:wkatz@xxxxxxxxxx]
Sent: Monday, June 19, 2017 7:01 PM
To: ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx> ; Muranyi,
Arpad <Arpad_Muranyi@xxxxxxxxxx <mailto:Arpad_Muranyi@xxxxxxxxxx> >
Subject: Re: [ibis-macro] Re: Response to BIRD-190
Why would an initial function return tap coeficients based on the input ir?
Get <https://aka.ms/ghei36> Outlook for Android
--- Begin Message ---Walter,
- From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
- To: <ibis-macro@xxxxxxxxxxxxx>
- Date: Wed, 1 Dec 2010 20:47:37 -0400
(I changed the subject line so we know what this thread is all about).
Thanks for your comment, I agree partially, but not completely.
You say ".if it has one of the following leaves.", which doesn't
make sense to me. We can't have Usage or Type alone. What did
you refer to by "one of"? The bottom part of the list under
"Allowed-Value Method" or the whole list? Even with considering
the bottom part of the list, this interpretation is flawed because
under certain circumstances Default and Value are mutually exclusive.
On the other hand, I like the sentence "A branch that is an AMI
parameter may not contain a branch." I think we should use that
in the BIRD.
How about if the Typos BIRD would say the following:
|* A branch in the .ami file is an "AMI Parameter" if it
|* contains the leaves Type, Usage, and any of the following
|* leaves:
|*
|* Default
|* <data_format> or Format <data_format>
|*
|* and does not contain another branch.
Note that I left out the word "only" Fangyi suggested, but added
the words in red at the end.
To address comment #3 from Fangyi, I added a few words to the end of the
paragraph on the top of pg. 5 of the BIRD draft shown in red:
|* The tree data structure passed in and out of the DLL
|* described in section 3.1.2.6 of Section 10 of this
document
|* is similar to the tree data structure in the .ami file
except
|* the 'Reserved_Parameters' and 'Model_Specific' branches
are
|* not included, the "AMI Parameter" branches become leaves
and
|* the "AMI parameters" of Usage Info and Out are not
included.
This actually raises a few more questions in my mind, especially
with the other email thread I started not too long ago about the
AMI_parameters_out argument in mind.
I can't find anything in the specification that describes how the
DLL returns an Out or InOut parameters and how the EDA tool is supposed
to look for them. Here are the possibilities that come to my mind.
Let's start with an InOut example first to make my question easier to
understand.
The AMI_parameters_in argument is a pointer to a string, and the
content of the string contains the InOut parameter name with a value.
The DLL reads this value, and after some number crunching decides to
return a different value. Where should this output value go?
1) The DLL could overwrite the string that it was given by the
EDA tool in the same memory location. Recall, the argument
is a pointer to the string, and it could be processed "in
place", just as we do with the impulse matrix.
2) On the other hand, we have another argument for AMI_parameters_out.
Is this where the DLL is supposed to put the output value it
generated? If so, I would imagine that the DLL has to make a
copy of the AMI_parameter (leaf value) string with the new value
in the value location and put it in the memory location of
AMI_parameters_out.
3) However, let's consider an example when the Init function contains
an optimizer which takes initial values for the tap coefficients
and after some number crunching returns better coefficients.
I would think that in this case the tap coefficients will be
declared as InOut arguments, and the initial values given to the
Init function will be overwritten by the Init function when the
optimizer is done. Now, how is this result propagated to the
GetWave function? Notice that the GetWave function doesn't have an
AMI_parameters_in argument. This makes me think that the
AMI_parameters_in argument of the Init function is also visible to
the GetWave function, and the modified tap coefficients from the
Init function are passed to GetWave through the AMI_parameters_in
memory location. This implies that the return value of an InOut
argument should be returned in place (which is #1 above). Using
this thinking we won't need the AMI_parameters_out argument for
returning the values of InOut arguments.
4) Now let's put another twist to this story and consider an Out
argument. Is the parameter string supposed to contain the name
of the AMI parameter name and a place holder for the return value
so that the Init function can modify this string in place when it
outputs the value?
Or, is the parameter string not supposed to include the name of
the Out argument, and is the Init function supposed to modify this
string in place and add that AMI parameter to it?
Or, is the Init function supposed to use the AMI_parameters_out
argument for returning Out parameters?
5) Studying the spec I also noticed that the AMI_parameters_out argument
is optional for GetWave, but for Init we don't say that it is
optional, so I assume it is required (correct me if I overlooked
something). Shouldn't the AMI_parameters_out be consistently
optional or required for both Init and GetWave?
Depending on the answers above, if the output should be placed in
the AMI_parameters_out argument, we can't say that it is optional,
at least not if the input parameter string includes Out or InOut
parameters.
Sorry for the lengthy discussion here. I almost feel like I may not
see the tree from the forest here, and if that is true, please correct
my "observations" above. On the other hand, I the tree is indeed not
visible in our "spec forest", I think we need to do a little more
cleaning.
Thanks,
Arpad
==========================================================================
From: Walter Katz [mailto:wkatz@xxxxxxxxxx] ;
Sent: Wednesday, December 01, 2010 2:42 PM
To: fangyi_rao@xxxxxxxxxxx; Muranyi, Arpad; ibis-macro@xxxxxxxxxxxxx
Subject: RE: [ibis-macro] Re: IBIS-ATM teleconference - Agenda for
11/30/2010
My definition of a parameter is:
A parameter tree contains a root, branches and leaves. A branch of the
parameter tree is an AMI Parameter, if it has one of the following leaves.
A branch that is an AMI parameter may not contain a branch.
Usage
Type
Default
Allowed-Value Method
Value
List
Range
Increment
Corner
Steps
Walter
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of
fangyi_rao@xxxxxxxxxxx
Sent: Tuesday, November 30, 2010 2:53 PM
To: Arpad_Muranyi@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: IBIS-ATM teleconference - Agenda for 11/30/2010
Hi, Arpad;
I have following questions regarding the Typo BIRD.
1. On page 2, shall we merge the two modifications on line 140 into
one?
2. At the bottom of page 4, shall we say "A branch in the .ami file
is an "AMI Parameter" if it only contains the leaves Type, Usage, and any
of the following leaves"? Shall we explicitly state that a branch having
Usage and/or Type and a sub-branch is illegal?
3. The first paragraph on page 5 seems to suggest that parameters of
Usage Info and Out are also passed into the DLL by the parameter string
for AMI Init.
Thanks,
Fangyi
--- End Message ---
--- Begin Message ---fangyi:
- From: "C. Kumar" <kumarchi@xxxxxxxxx>
- To: <wkatz@xxxxxxxxxx>, <ibis-macro@xxxxxxxxxxxxx>, <fangyi_rao@xxxxxxxxxxx>
- Date: Mon, 12 Oct 2009 22:36:15 -0400
i agree the case of tx with getwave and rx just with init is problematic.
the tx has getwave because the init alone cannot provide equalization
information. but rx init is not very useful without tx equalization factored
in
in this case the minimum rx init can do is to just allow the user/eda tool
supply the coefficients
--- On Mon, 10/12/09, fangyi_rao@xxxxxxxxxxx <fangyi_rao@xxxxxxxxxxx> wrote:
From: fangyi_rao@xxxxxxxxxxx <fangyi_rao@xxxxxxxxxxx>
Subject: RE: [ibis-macro] Re: An AMI Overview
To: kumarchi@xxxxxxxxx, wkatz@xxxxxxxxxx, ibis-macro@xxxxxxxxxxxxx
Date: Monday, October 12, 2009, 10:15 PM
Hi, Kumar;
As Walter pointed out Rx Init does need a modified impulse from Tx to
prepare GetWave or to optimize taps for optimal performance. The problem of
all these confusion and broken flows is that AMI is letting the Init
function provide multiple functionalities including
Prepare for GetWave or optimize taps
Return approximate LTI model for statistical simulation
Return Tx/Rx EQ information
It is too much to put in one single function.
Fangyi
From: C. Kumar [mailto:kumarchi@xxxxxxxxx]
Sent: Monday, October 12, 2009 6:54 PM
To: wkatz@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx; RAO,FANGYI (A-USA,ex1)
Subject: Re: [ibis-macro] Re: An AMI Overview
instead of knotting up all over the place, we have to simplify the flow
i.e if a model has getwave , it cannot modify the impulse response
if the vendor wants to provide approximate models he has to provide separate
models containing just init,
any other method of prescribing both getwave and init will need to confusion
and broken models/flows
--- On Mon, 10/12/09, fangyi_rao@xxxxxxxxxxx <fangyi_rao@xxxxxxxxxxx> wrote:
From: fangyi_rao@xxxxxxxxxxx <fangyi_rao@xxxxxxxxxxx>
Subject: [ibis-macro] Re: An AMI Overview
To: wkatz@xxxxxxxxxx, ibis-macro@xxxxxxxxxxxxx
Date: Monday, October 12, 2009, 8:34 PM
Hi, Walter;
You may not notice in my example the Rx Init returns h_AC*h_TEI + h_REI
instead of h_AC*h_TEI*h_REI. In this case the Tx portion can’t be removed
from the final impulse by deconvolution. We can’t recover h_REI by
deconvolution either. The flow will be broken if Tx has GetWave but Rx does
not.
Thanks,
Fangyi
From: Walter Katz [mailto:wkatz@xxxxxxxxxx]
Sent: Monday, October 12, 2009 4:39 PM
To: RAO,FANGYI (A-USA,ex1); ibis-macro@xxxxxxxxxxxxx
Subject: RE: [ibis-macro] Re: An AMI Overview
Fangyi,
By deconvolution ( <http://en.wikipedia.org/wiki/Deconvolution>
http://en.wikipedia.org/wiki/Deconvolution ;).
The input to Rx Init is h_AC*h_TEI, the output of Rx Init is
h_AC*h_TEI*h_REI. Deconvolution takes the Fourier Transform of the input and
output impulse responses, then divides the coefficients of the two Fourier
Transforms, and then does an inverse Fourier Transform to get h_REI. We do
this for this flow because we have no other choice. That is why I am
proposing to add to Rx Init models the ability to just return h_REI.
Walter
Walter Katz
303.449-2308
Mobile 720.333-1107
wkatz@xxxxxxxxxx
www.sisoft.com
-----Original Message-----
From: fangyi_rao@xxxxxxxxxxx [mailto:fangyi_rao@xxxxxxxxxxx]
Sent: Monday, October 12, 2009 6:55 PM
To: wkatz@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
Subject: RE: [ibis-macro] Re: An AMI Overview
Hi, Walter;
Thanks for your explanation. Regarding point 2, I still think if model
vendors are allowed to modify impulse any way they want, then following
scenario will be broken.
Tx has GetWave.
Tx Init modifies h_AC and returns h_AC_Tx=h_AC*h_TEI
Rx does NOT have GetWave.
h_AC_Tx is passed into Rx Init.
Rx Init modified h_AC_Tx and returns h_AC_Tx_Rx= h_AC*h_TEI + h_REI
Tx GetWave output is to be convolved with h_AC_Rx=h_AC + h_REI, which is the
combined impulse of channel and Rx.
Without knowing how Rx Init modifies the impulse, how can EDA tools remove
the h_TEI portion from h_AC_Tx_Rx to get h_AC_Rx?
Thanks,
Fangyi
From: Walter Katz [mailto:wkatz@xxxxxxxxxx]
Sent: Monday, October 12, 2009 3:04 AM
To: RAO,FANGYI (A-USA,ex1); ibis-macro@xxxxxxxxxxxxx
Subject: RE: [ibis-macro] Re: An AMI Overview
Fangyi,
1. Even if a model is non-LTI, it may have an LTI approximation. If the
Init call returns an LTI approximation, then the EDA tool can use
“Statistical” methods to do very fast analysis, realizing that the GetWave
time domain simulation will still need to be made to confirm the result. If
the Init call does not return an LTI approximation, then the EDA tool must
use time domain methods only to analyze the channel.
2. The current IBIS 5.0 specification has words like (section 2.1.6):
| 6. AMI_Init parses the configuration parameters, allocates dynamic
| memory, places the address of the start of the dynamic memory in
| the memory handle, computes the impulse response of the block and
| passes the modified impulse response to the EDA tool. The new
| impulse response is expected to represent the filtered response.
Since returning an impulse response assumes LTI (or an LTI approximation),
the model maker can do this calculation any way he chooses. One way to do
this calculation convolution. We spent much time determining if we should
use the word convolution, concatenation, combination, and ended up with the
word modified. It would probably be worth adding a paragraph to the document
that explicitly says that whenever convolution is used, it is simply to
represent the result of the function to be preformed, and the model maker or
EDA vendor can use any mathematical method deemed appropriate for
performance, accuracy or other algorithmic reasons.
Walter
Walter Katz
303.449-2308
Mobile 720.333-1107
wkatz@xxxxxxxxxx
www.sisoft.com
-----Original Message-----
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx]On Behalf Of fangyi_rao@xxxxxxxxxxx
Sent: Sunday, October 11, 2009 8:22 PM
To: wkatz@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: An AMI Overview
Hi, Walter;
Thanks for your clarification of AMI methodology. I’d like to see the
overview or the spec also states clearly about following two topics.
1. For non-LTI models that have GetWave, how does AMI_Init enable
model developers to also support statistical simulation using LTI
approximation?
2. Do we explicitly assume the modified impulse returned by AMI_Init
is the CONVOLUTION between input impulse and the equalizer? I know at least
one model vendor would not be happy about this assumption because their
modified impulse is not from convolution.
Regards,
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Walter Katz
Sent: Saturday, October 10, 2009 2:20 PM
To: IBIS-ATM
Subject: [ibis-macro] An AMI Overview
All,
We have all been getting into the nuts and bolts of AMI. I think we need to
step back and understand what AMI is all about. The following has been said
in many ways in many places, but I think it is appropriate to review it at
this time. I do not know if this belongs in the IBIS AMI specification, but
it sure would be nice.
Walter
AMI Overview
The analysis of high speed (>4 Gbps) offers some simplifications and some
challenges. The simplifications arise from the fact the driver (Tx) final
stage output, and the receiver (Rx) input along with the interconnect
between them can be treated as Linear and Time Invariant (LTI)
“Analog-Channel”. This allows various mathematical techniques to be applied
improving simulation performance by orders of magnitude. The complication
that has led to the development of the AMI standard arises from the need for
complex signal processing in both the Tx before the final stage driver and
particularly the Rx after the Rx receiver. The good news is that the signal
processing does not interact with the Tx/Interconnect/Rx Analog-Channel. The
signal processing that is going on is considered important Intellectual
Property (IP) of the IC vendors, and this along with the need for simulating
millions of bits and the IP considerations led to the AMI standard.
A SerDes-Channel consists of Tx signal processing, Analog-Channel and Rx
signal processing. The Tx AMI model represents the Tx signal processing, an
impulse response represents the Analog-Channel, and the Rx AMI model
represents the Rx signal processing.
The Analog-Channel is represented as an impulse response hAC(t). It is the
differential mode impulse response of the interconnects between the Tx and
the Rx and includes the reactive load (impedance) of the Tx final stage
driver and the Rx receiver. This impulse response can be created using many
mathematical techniques, including, but not limited to simulation and TDR
measurement. Traditional IBIS does not model differential LTI buffers, this
will be addressed when I introduce new AMI reserved parameters. This is also
related to the topic of Vladimir’s presentation this Tuesday.
The Tx AMI model and Rx AMI model may themselves be either LTI or non-LTI.
If they are LTI they can be represented accurately by an impulse response.
If they are not LTI, they can be approximated by an impulse response. AMI
models are delivered as executable code in the form of a Shared Object (SO)
or a Dynamically Linked Library (DLL), and an .ami ASCII file. All AMI DLL’s
have an AMI_Init entry, and an AMI_Close entry. This is all that is required
if the model is LTI. If a model is non-LTI then is must also have an
AMI_GetWave. If a model does have an AMI_GetWave then the model make is
telling the EDA tool that the model is non-LTI and that using the just
impulse response from the AMI_Init is not deemed sufficient to accurately
model the channel.
The .ami file tells the EDA tool if the model has an AMI_GetWave entry, and
sufficient information to configure the model, and to analyze the results
that the model generates.
When a model is LTI it does not need a GetWave entry. When LTI, the only
information that need be abstracted from the model is the impulse response
of the models equalization. The input to the Tx AMI_Init function is the
impulse response of the channel. The input to the Rx AMI_Init function is
the impulse response of the channel combined with the impulse response of
the Tx equalization obtained from the call to Tx AMI_Init.
When a model is non-LTI (or more importantly, the IC Vendor believes that
the non-LTI behavior is important) then it does require an AMI_GetWave
entry. The AMI_Init entry is used to return to the EDA tool an approximate
impulse response of the models equalization, and to pass on initialization
information to be used by the AMI_GetWave entry. The AMI_GetWave entry is
then called repeatedly with sequential blocks of waveforms. The waveforms
are that are input to Tx AMI_GetWave indicate the transition times of the
digital stimulus input to the Tx equalization circuitry. The waveforms that
are output of Tx AMI_GetWave are the waveforms that drive the
Analog-Channel. The waveform that is input to Rx AMI_GetWave is the waveform
that is the input to the Rx buffer. The output of the Rx AMI_GetWave is the
waveform at the Rx decision point, and optionally clock ticks indicating the
location of each recovered clock. The EDA tool processes the waveform at the
Rx decision point, and either uses the clock ticks or the Clock Recovery
Mean and Rj to generate bathtub curves, BER and other channel data.
A fundamental principle of AMI modeling is that every EDA platform (both
software and hardware) will give the same results when presented with the
same Analog-Channel impulse response, the same AMI model conditions, and the
same input stimulus pattern. Each EDA platform may differ on how its sets
the Tx and Rx AMI model conditions, the stimulus pattern, how it creates the
Analog-Channel impulse response, and how it processes the resulting outputs.
Walter Katz
303.449-2308
Mobile 720.333-1107
wkatz@xxxxxxxxxx
www.sisoft.com
--- End Message ---