Ambrish,
The BIRD explicitly states that
"Both default input and output values of DC_Offset are 0.0 Volt. If the
parameter is defined as Usage In, the output value of DC_Offset is accordingly
assumed to be 0.0 Volt."
Isn't that clear enough to derive the model behavior when DC_Offset Usage is
In? I would not expect a BIRD to spell out all possible scenarios.
Regards,
Fangyi
From: Ambrish Varma <ambrishv@xxxxxxxxxxx>
Sent: Wednesday, August 14, 2019 11:47 AM
To: Fangyi Rao <fangyi_rao@xxxxxxxxxxxx>; Arpad_Muranyi@xxxxxxxxxx; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx>
Subject: RE: DC_Offset BIRD - shifting the output waveform.
[EXTERNAL]
Hi Fangyi,
Your answer suggests that the BIRD does require the AMI Model maker to shift
the waveform and the specification does require the model to output the
physical waveform at the latch *when DC_Offset is Usage In*.
Nowhere is this mentioned in the BIRD - it is only implied.
In fact if you read the definition only - you are lead to believe that the "EDA
tools form the complete waveform at the Rx algorithmic model output"
I am not saying that the BIRD is wrong or it is hard or easy to do something in
the BIRD. I am saying it is misleading and confusing for the model maker who
does not live and breathe the IBIS/AMI spec.
Thanks,
Ambrish.
From: Fangyi Rao <fangyi_rao@xxxxxxxxxxxx<mailto:fangyi_rao@xxxxxxxxxxxx>>
Sent: Wednesday, August 14, 2019 2:35 PM
To: Ambrish Varma <ambrishv@xxxxxxxxxxx<mailto:ambrishv@xxxxxxxxxxx>>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: RE: DC_Offset BIRD - shifting the output waveform.
EXTERNAL MAIL
Then declare DC_Offset usage to be InOut and return the value of DC_Offset. How
difficult is that?
Fangyi
From: Ambrish Varma <ambrishv@xxxxxxxxxxx<mailto:ambrishv@xxxxxxxxxxx>>
Sent: Wednesday, August 14, 2019 11:19 AM
To: Fangyi Rao <fangyi_rao@xxxxxxxxxxxx<mailto:fangyi_rao@xxxxxxxxxxxx>>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: RE: DC_Offset BIRD - shifting the output waveform.
[EXTERNAL]
Fangyi,
When DC_Offset is Usage In, the AMI model needs to take care of the shifting of
the waveform. The AMI model is expected to output the physical waveform at the
latch as the EDA tool will only get 0.0 Volt to shift the waveform.
Is this not the case?
Thanks,
Ambrish.
From: Fangyi Rao <fangyi_rao@xxxxxxxxxxxx<mailto:fangyi_rao@xxxxxxxxxxxx>>
Sent: Wednesday, August 14, 2019 2:00 PM
To: Ambrish Varma <ambrishv@xxxxxxxxxxx<mailto:ambrishv@xxxxxxxxxxx>>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: RE: DC_Offset BIRD - shifting the output waveform.
EXTERNAL MAIL
Ambrish,
Why do you think EDA tool forming the final output waveform is misleading? You
mentioned that it's a very common scenario where AMI models do not shift the
waveform and expect/let the EDA tool handle the shift. Why do you think that is
not misleading but the EDA tool forming the final output waveform is? These two
are the same.
Your two following statements are contradicting to each other.
* "This is a very common scenario where the AMI model may ... not shift the
waveform and expect/let the EDA tool handle the shift"
* "the EDA tool forms the final waveform at the output. This is misleading
..."
Your understanding of this BIRD is incorrect. The BIRD does NOT require models
to perform waveform shift. It just requires that if the model wants the EDA
tool to shift the waveform the model needs to tell the tool how much to shift
by returning DC_Offset in AMI_Init. Please study the examples attached at the
end of the BIRD. Do you have an example where the model is forced to perform
waveform shift by this BIRD?
Regards,
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Ambrish Varma
Sent: Tuesday, August 13, 2019 8:07 PM
To: Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: [ibis-macro] Re: DC_Offset BIRD - shifting the output waveform.
[EXTERNAL]
Arpad, Fangyi,
I understand the implicit and nuanced meaning/behavior. The BIRD mentions that
the EDA tool forms the final waveform at the output. This is misleading and may
throw off model makers. I am going to hazard a guess and even say that a
majority of the models will not want to perform a waveform shift and this BIRD
will require them to make changes/add code that they may not want/need to make.
Thanks,
Ambrish.
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Muranyi, Arpad
Sent: Tuesday, August 13, 2019 9:23 PM
To: IBIS-ATM <ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: [ibis-macro] Re: DC_Offset BIRD - shifting the output waveform.
EXTERNAL MAIL
You might look at it this way too:
When DC_Offset is Usage In, we are telling the EDA too TO SHIFT the
waveform, using the default value of 0... :)
Thanks,
Arpad
=====================================================
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Fangyi Rao
Sent: Tuesday, August 13, 2019 8:16 PM
To: dmarc-noreply@xxxxxxxxxxxxx<mailto:dmarc-noreply@xxxxxxxxxxxxx>;
ambrishv@xxxxxxxxxxx<mailto:ambrishv@xxxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: [ibis-macro] Re: DC_Offset BIRD - shifting the output waveform.
I don't see anything confusing here. If you want the EDA tool to shift the
waveform, you must tell the tool explicitly how much to shift.
Neither do I see anything cumbersome for model maker. Returning the DC_Offset
value in AMI_Init is just one line of code.
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Fangyi Rao
Sent: Tuesday, August 13, 2019 6:06 PM
To: ambrishv@xxxxxxxxxxx<mailto:ambrishv@xxxxxxxxxxx>; IBIS-ATM
<ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: [ibis-macro] Re: DC_Offset BIRD - shifting the output waveform.
[EXTERNAL]
"and not shift the waveform and expect/let the EDA tool handle the shift."
Specifying the DC_Offset usage to be In is equivalent to telling the EDA tool
NOT to shift the waveform because the default output value of DC_Offset is 0.
If you want the EDA tool to shift the waveform, specify the DC_Offset usage to
be InOut and return the value in AMI_Init so that the EDA tool knows how much
to shift.
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Ambrish Varma
Sent: Tuesday, August 13, 2019 4:41 PM
To: IBIS-ATM <ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>>
Subject: [ibis-macro] DC_Offset BIRD - shifting the output waveform.
[EXTERNAL]
Hi All,
At the end of today's meeting, I raised a couple of question about finding out
1. Who creates the final waveform - the EDA tool or the AMI Model and
2. Whether the waveform output from the Rx AMI Getwave is the physical
waveform at the latch.
It appeared that the lines in the Definition
"The output value of DC_Offset returned by Rx AMI_Init is to be added to the Rx
AMI_GetWave output waveform by EDA tools to form the complete waveform at the
Rx algorithmic model output."
clarified this. A few people in the meeting agreed that the EDA tool is the one
that creates the final waveform and that the Getwave output is NOT the waveform
at the latch.
On closer reading though - there is a scenario where I am still not clear if
that is the case.
If the DC_Offset is In, the BIRD says that the output value is assumed to be
0.0 Volt, in that case, the EDA tool is not creating the final waveform and the
waveform output is the output of the latch.
If the AMI model does not want to shift the O/P waveform, it has to pass the
DC_Offset value as an Out parameter such that the EDA tool may then create the
physical waveform.
This is a very common scenario where the AMI model may use the DC_Offset to
determine its CTE code/taps etc and not shift the waveform and expect/let the
EDA tool handle the shift.
In this case - it must pass the DC_Offset as an Out parameter which will be
cumbersome for the model/model maker.
This is confusing and will invite errors and result in faulty models.
Comments?
Thanks,
Ambrish.