[ibis-macro] Re: question on tap parameters in AMI
- From: "C. Kumar" <kumarchi@xxxxxxxxx>
- To: scott@xxxxxxxxxxxxx, ckumar@xxxxxxxxxxx
- Date: Thu, 11 Feb 2010 11:06:47 -0800 (PST)
scott:
but the ami model has provision for limits. Within the context if the model
accepts parameters it is still possible to do optimization outside the model.
I was just pointing out that there may not be any need to know whether it is
dfe,ffe, analog or any other combination The eda tool/user can optimize based
on eye, ber or any other eda tool/user understandable criteria.
________________________________
From: Scott McMorrow <scott@xxxxxxxxxxxxx>
To: ckumar@xxxxxxxxxxx
Cc: fangyi_rao@xxxxxxxxxxx; ambrishv@xxxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
Sent: Thu, February 11, 2010 1:57:42 PM
Subject: [ibis-macro] Re: question on tap parameters in AMI
Kumar,
A standard way of identifying which of various equalization types are
being used would be nice. Certainly classical FFE and DFE could be
indicated, allowing the EDA platform to perform intelligent
optimization or "dithering" of the tap parameters for sensitivity
studies outside the boundaries of what the silicon provides. I can see
this as valuable for a silicon vendor, during compliance verification,
and during robustness analysis of a system. (Personally I hate
optimizations that have the potential to be unstable with slight errors
that occur from those un-modeled, or improperly modeled, demons.
Experience dictates that any particular model is only an imperfect
approximation of the silicon. As a system designer, I like to stress
models outside of their normal limits.) A
silicon vendor that has a proprietary equalization scheme, or does not
want to reveal the details, could use a type of UNKNOWN. Future
methods, including analog methods, could also be easily added with the
same construct.
If the AMI model is truly a "black box" there is no need to pass
unknown parameters at all. By definition, a black box is a closed
system with well-defined inputs and outputs. User-defined taps, without
further documentation, would not be well-defined in my estimation.
Thus, if the EDA platform is going to pass Taps back and forth, more
standardized knowledge should be included in the description than is
found in model comment lines or help files. If a silicon vendor does
not want to reveal the equalization technique used by their "black
box", then there is no need to define taps at all, or pass them through
the EDA tool.
best regards,
Scott
--
Scott McMorrow
Teraspeed Consulting Group LLC
121 North River Drive
Narragansett, RI 02882
(401) 284-1827 Business
(401) 284-1840 Fax
http://www.teraspeed.com
Teraspeed® is the registered service mark of
Teraspeed Consulting Group LLC
ckumar@xxxxxxxxxxx wrote:
if the model allows for user defined inputs it can be done. Even in this case
>the eda tool need not know the details of model type or nature of its
>implementation. The whole idea of ami model is, it is a black box.
>
>Quoting fangyi_rao@xxxxxxxxxxx:
>
>Some customers want the eda tool to optimize these taps and set their
>>values in the ami Init call.
>>
>>Fangyi
>>
>>-----Original Message-----
>>From: ckumar@xxxxxxxxxxx [mailto:ckumar@xxxxxxxxxxx]
>>Sent: Thursday, February 11, 2010 10:16 AM
>>To: RAO,FANGYI (A-USA,ex1)
>>Cc: ambrishv@xxxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
>>Subject: Re: [ibis-macro] Re: question on tap parameters in AMI
>>
>>why should the eda tool know what type of filter or whatever it is?
>>
>>As far as the eda tool is concerned the only job it has to do is to pass
>>the
>>model instances of some arbitrarily named parameters whose values comply
>>with
>>ami definition.
>>
>>Quoting fangyi_rao@xxxxxxxxxxx:
>>
>>Hi, Ambrish;
>>>
>>>
>>>
>>>Thanks for the clarification. In your example a EDA tool will find two
>>>groups of tap parameter, RX_FFE_Taps and RX_DFE_Taps in the .ami file
>>>based on the parameter type. How does the tool know which group is for
>>>FFE and which is for DFE? Note that model specific parameter names can
>>>be arbitrary.
>>>
>>>
>>>
>>>Fangyi
>>>
>>>
>>>
>>>From: ibis-macro-bounce@xxxxxxxxxxxxx
>>>[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Ambrish Varma
>>>Sent: Wednesday, February 10, 2010 7:57 PM
>>>To: ibis-macro@xxxxxxxxxxxxx
>>>Subject: [ibis-macro] Re: question on tap parameters in AMI
>>>
>>>
>>>
>>>Hi Fangyi,
>>>
>>>You can give any name to the parameters themselves. For ex,
>>>
>>>The model_specific section of the RX model with a 5 tap FFE and 3 tap
>>>DFE can look like:
>>>
>>>(Model_Specific
>>>
>>>(RX_FFE_Taps
>>>
>>> (-2 (Usage Inout)(Type Tap) (Format Range 0.1 -0.1 0.2)(Default 0.1)
>>>
>>> (Description "Second Precursor Tap"))
>>>
>>> (-1 (Usage Inout)(Type Tap) (Format Range 0.2 -0.4 0.4)(Default 0.2)
>>>
>>> (Description "First Precursor Tap"))
>>>
>>> (0 (Usage Inout)(Type Tap) (Format Range 1 -1 2)(Default 1)
>>>
>>> (Description "Main Tap"))
>>>
>>> (1 (Usage Inout)(Type Tap) (Format Range 0.2 -0.4 0.4)(Default2 0.2)
>>>
>>> (Description "First Post cursor Tap"))
>>>
>>> (2 (Usage Inout)(Type Tap) (Format Range 0.1 -0.1 0.2)(Default 0.1)
>>>
>>> (Description "Second Post cursor Tap"))
>>>
>>>)
>>>
>>>(RX_DFE_Taps
>>>
>>> (-1 (Usage Inout)(Type Tap) (Format Range 0.2 -0.4 0.4)(Default
>>>
>>0.2)
>>
>> (Description "First Precursor Tap"))
>>>
>>> (0 (Usage Inout)(Type Tap) (Format Range 1 -1 2)(Default 1)
>>>
>>> (Description "Main Tap"))
>>>
>>> (1 (Usage Inout)(Type Tap) (Format Range 0.2 -0.4 0.4)(Default2
>>>
>>0.2)
>>
>> (Description "First Post cursor Tap"))
>>>
>>>)
>>>
>>>) | end model_specific
>>>
>>>
>>>
>>>Hope that helps,
>>>
>>>Ambrish.
>>>
>>>________________________________
>>>
>>>From: ibis-macro-bounce@xxxxxxxxxxxxx
>>>[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of
>>>fangyi_rao@xxxxxxxxxxx
>>>Sent: Wednesday, February 10, 2010 9:21 PM
>>>To: ibis-macro@xxxxxxxxxxxxx
>>>Subject: [ibis-macro] question on tap parameters in AMI
>>>
>>>
>>>
>>>Hi, Experts;
>>>
>>>
>>>
>>>I have a question on tap parameter definition in .ami file. The
>>>
>>current
>>
>>standard states that:
>>>
>>>
>>>
>>>If in addition the individual tap weights are each given a name
>>>
>>>which is their tap number (i.e., "-1" is the name of the
>>>
>>>first precursor tap, "0" is the name of the main tap, "1" is
>>>
>>>the name of the first postcursor tap, etc.) and the tap
>>>
>>>weights are declared to be of type Tap, then the EDA platform
>>>
>>>can assume that the individual parameters are tap weights in
>>>
>>>a tapped delay line
>>>
>>>
>>>
>>>This way we have to assume that Tx only has FFE and Rx only has DFE.
>>>What if Rx has both FFE and DFE? How can we tell which group of taps
>>>
>>is
>>
>>FFE and which is for DFE?
>>>
>>>
>>>
>>>Thanks in advance.
>>>
>>>
>>>
>>>Fangyi
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>>
>
>
>---------------------------------------------------------------------
>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
>
>
>
--
Scott McMorrow
Teraspeed Consulting Group LLC
121 North River Drive
Narragansett, RI 02882
(401) 284-1827 Business
(401) 284-1840 Fax
http://www.teraspeed.com
Teraspeed® is the registered service mark of
Teraspeed Consulting Group LLC
Other related posts:
