Arpad,
The whole intent of multiple calls to GetWave is to break very long
simulations into manageable chunks. A simulation might be 100
million bits long and be broken up into 100000 1000 bit chunks. The
EDA tool need only allocate somewhat more than 1000 double precision
numbers. The EDA tool re-uses the same 1000 word buffer on each call
to GetWave. The –1 is the way the DLL tells the EDA tool how many
clock tics were generated in this particular GetWave call. The EDA
tool then assembles the waveform and clock tics into a very long 100
million-bit waveform. The EDA tool really does not keep the
100-million bit waveform in memory at once, but can, for example,
build a statistical eye on the fly. The EDA tool allocates the
memory. There is no requirement for the EDA tool to initialize the
value of this memory.
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 *Muranyi, Arpad
*Sent:* Sunday, March 28, 2010 1:07 PM
*To:* ibis-macro@xxxxxxxxxxxxx
*Subject:* [ibis-macro] Re: FW: Question on clock_times
All,
Things are getting more and more interesting as the replies
come in... First, let me press the reset button because the
notations I see are getting a bit confusing.
Walter uses "na" by which I suppose he means something like
uninitialized memory or garbage.
In my original question I used "#" to indicate a valid usable
number, I just didn't want to specifically write improvised
numbers there. It seems that others may have interpreted
this differently in the discussion, so I just wanted to
mention this to be sure we are in agreement.
The other thing that was not mentioned in this discussion yet
is the indexing of the retuned values from GetWave. So is the
GetWave function supposed to put the numbers to the beginning
of the vector FOR EACH CALL, or keep track of where the
previous GetWave call finished in memory and continue from there
on? The reason I am asking this is because Walter's last reply
indicates the GetWave function supposed to return a -1 at the
end of each call. In my mind this may imply that the EDA tool
supposed to assemble the returned short vectors into a single
long continuous vector which doesn't have a -1 between the
chunks that it was assembled from, which doesn't seem to be
mentioned in the spec and was not my understanding at all.
In this illustration I am shortening the GetWave call length
to 5 so I can fit the example on a line better (if it gets
garbled with new lines by the mail system I will try again
in a different way).
memory index: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
call 1 value: 0 10p 20p 30p 40p -1
call 2 value: 50p 60p 70p 80p 90p -1
call 3 value: 100p 110p 120p 130p 140p -1
call 4 value: 150p 160p 170p 180p 190p -1
memory index: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
21 22 23 24
call 1 value: 0 10p 20p 30p 40p -1
call 2 value: 50p 60p 70p 80p -1
call 3 value: 90p 100p 110p 120p 130p 140p -1
call 4 value: 150p 160p 170p 180p 190p 200p -1
memory index: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
call 1 value: 0 10p 20p 30p 40p -1
call 2 value: 50p 60p 70p 80p 90p -1
call 3 value: 100p 110p 120p 130p 140p -1
call 4 value: 150p 160p 170p 180p 190p 200p -1
memory index: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
call 1 value: 0 10p 20p 30p 40p
call 2 value: 50p 60p 70p 80p 90p
call 3 value: 100p 110p 120p 130p 140p
call 4 value: 150p 160p 170p 180p 190p 200p -1
My understanding was this last example. I even remember some
discussions on how much memory the EDA tool was supposed to
allocate fir this vector and the answers I received then was
that the safest amount is the length of the entire waveform.
If each GetWave call would return the values starting at index
0 we wouldn't need to allocate that much memory. Also, in this
case I don't see why each GetWave call would need to return a
-1 at its end except for the last one.
Any comments?
Thanks,
Arpad
==================================================================
------------------------------------------------------------------------
*From:* Walter Katz [mailto:wkatz@xxxxxxxxxx]
*Sent:* Sunday, March 28, 2010 10:04 AM
*To:* Muranyi, Arpad; ibis-macro@xxxxxxxxxxxxx
*Subject:* RE: [ibis-macro] Re: FW: Question on clock_times
Arpad,
The following two cases are valid:
CASE 1:
After the 1-st call:
0 1 2 3 4 5 6 7 8 9 –1 na na na …
After the 2-nd call:
10 11 12 13 14 15 16 17 18 19 –1 na na na …
After the 3-rd call:
20 21 22 23 24 25 26 27 28 29 –1 na na na …
etc.
CASE 2:
After the 1-st call:
-1 # # # # # # # # # # // First call does not generate clocks
After the 2-nd call:
0 1 2 3 4 5 6 7 8 9 -1
After the 3-rd call:
10 11 12 13 14 15 16 17 18 19 -1 // time increment continued from
the last call
Each call to GetWave returns an array of clock times terminated by a
single –1. The data after the –1 is undefined, and can be anything
(including –1’s).
Any individual call to GetWave can return a –1 in the first element
of the array if either GetWave does not return clock ticks at all,
or if there were no clock ticks during that call to GetWave. The
later is possible but very highly unlikely. It is also required that
the times of the clock ticks monotonically increase and always
reference the time from 0, the beginning of the simulation.
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 *Muranyi, Arpad
*Sent:* Sunday, March 28, 2010 10:46 AM
*To:* ibis-macro@xxxxxxxxxxxxx
*Subject:* [ibis-macro] Re: FW: Question on clock_times
Todd,
Now you really confused me. In your previous email you said:
**"**Surprisingly – all three of Vladimir’s cases are technically
valid, according to the way the current spec is written. Clearly,
cases 2 and 3 don’t make sense (for different reasons), but there’s
nothing in the *current* spec that says they’re wrong."
Now you write:
"We’ve always taken that to mean that clock times maintain a
constant time zero reference across Getwave calls (i.e. not
referenced to the start of the block)."
The reason this confuses me is because all three cases from Vladimir
include a "-1" at the end of each GetWave call's clock_time vector.
If there is only a single time zero reference across all GetWave
calls, shouldn't there be a single "-1" at the end of the very last
GetWave call? So putting this all together, it seems that only the
first case would be correct, but even that only if we took out the
extra "-1"-s like this:
CASE 1:
After the 1-st call:
0 1 2 3 4 5 6 7 8 9
After the 2-nd call:
10 11 12 13 14 15 16 17 18 19
After the 3-rd call:
20 21 22 23 24 25 26 27 28 29 -1
etc.
Thanks,
Arpad
======================================================================
------------------------------------------------------------------------
*From:* ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Todd Westerhoff
*Sent:* Friday, March 26, 2010 4:41 PM
*To:* ibis-macro@xxxxxxxxxxxxx
*Subject:* [ibis-macro] Re: FW: Question on clock_times
Scott,
| 3.2.2.3 clock_times
| ===================
|
| Vector to return clock times. The clock times are referenced to
the start
| of the simulation (the first AMI_GetWave call). The time is always
| greater or equal to zero. The last clock is indicated by putting a
value
| of -1 at the end of clocks for the current wave sample. The
clock_time
| vector is allocated by the EDA platform and is guaranteed to be
greater
| than the number of clocks expected during the AMI_GetWave call.
The clock
| times are the times at which clock signal at the output of the clock
| recovery loop crosses the logic threshold. It is to be assumed
that the
| input data signal is sampled at exactly one half clock period after a
| clock time.
We’ve always taken the phrase:
| Vector to return clock times. The clock times are referenced to
the start
| of the simulation (the first AMI_GetWave call).
We’ve always taken that to mean that clock times maintain a constant
time zero reference across Getwave calls (i.e. not referenced to the
start of the block). You can argue that isn’t stated explicitly
enough, but that’s my understanding.
Suggestions on wording to make things clearer are welcome and
appreciated.
Todd.
________________________
Todd Westerhoff
VP, Software Products
SiSoft
6 Clock Tower Place, Suite 250
Maynard, MA 01754
(978) 461-0449 x24
twesterh@xxxxxxxxxx <mailto:twesterh@xxxxxxxxxx>
www.sisoft.com <http://www.sisoft.com>
------------------------------------------------------------------------
*From:* ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Scott McMorrow
*Sent:* Friday, March 26, 2010 5:34 PM
*To:* Wenyi.Jin@xxxxxxx
*Cc:* vladimir_dmitriev-zdorov@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
*Subject:* [ibis-macro] Re: FW: Question on clock_times
The specification is not clear concerning the meaning of the
clock_times vector.
I see two valid interpretations:
· Clock_times are relative to the start of the current block
· clock_times are relative to the start of the 1st block
I do not see this defined in the specification. If it is, then
please point me to the section.
--
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
Jin, Wenyi wrote:
for CASE 2 and CASE 3, how can clock_time begin from 0? It is
already 2^nd block.
If some call generate valid clock_time, some not, that means DLL has
to track
the information like block number to decide from which block begin
return clock time.
DLL can always return clock time even CDR is in lock-in stage, it is
the EDA tool that
Specified after how many UI or block, to make use of clock_time.
*From:* ibis-macro-bounce@xxxxxxxxxxxxx
<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of
*Dmitriev-Zdorov, Vladimir
*Sent:* Friday, March 26, 2010 2:08 PM
*To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
*Subject:* [ibis-macro] Re: FW: Question on clock_times
Please clarify if the following clock time patterns are legitimate
when performing several GetWave calls (for simplicity/brevity, let’s
assume that we have only 10 bits per call and will use integers
instead of small double values). Which of the following 3 cases are
allowed?
CASE 1:
After the 1-st call:
0 1 2 3 4 5 6 7 8 9 -1
After the 2-nd call:
10 11 12 13 14 15 16 17 18 19 -1
After the 3-rd call:
20 21 22 23 24 25 26 27 28 29 -1
etc.
CASE 2:
After the 1-st call:
-1 # # # # # # # # # # // First call does not generate clocks
After the 2-nd call:
0 1 2 3 4 5 6 7 8 9 -1
After the 3-rd call:
10 11 12 13 14 15 16 17 18 19 -1 // time increment continued from
the last call
etc.
CASE 3:
After the 1-st call:
-1 # # # # # # # # # #
After the 2-nd call:
0 1 2 3 4 5 6 7 8 9 -1
After the 3-rd call:
0 1 2 3 4 5 6 7 8 9 -1 // i.e. always start from zero
etc.
-----Original Message-----
*From:* ibis-macro-bounce@xxxxxxxxxxxxx
<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Muranyi, Arpad
*Sent:* Friday, March 26, 2010 2:38 PM
*To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
*Subject:* [ibis-macro] FW: Question on clock_times
Kumar,
So are you saying that the vector can begin
with a bunch of -1 and then have valid number
later in the vector? Lie this:
-1 -1 -1 -1 # # # # # # # # -1 -1 -1 -1
If this is true, is it also true that you can
have alternating -1 and good values in the
vector? Like this?
-1 -1 -1 -1 # # # # # # # # -1 -1 -1 -1 -1 -1 -1 -1 # # # # # # # #
-1 -1 -1 -1
Arpad
==============================================
------------------------------------------------------------------------
*From:* C. Kumar [mailto:kumarchi@xxxxxxxxx]
*Sent:* Friday, March 26, 2010 3:29 PM
*To:* Muranyi, Arpad
*Subject:* Re: [ibis-macro] Question on clock_times
arpad:
let me take shot at this
1. Yes -1 indicates end of clock vector
2. clock vector is the clocks associated with the particular wave
vector in the particular getwave call.
3. some models can have delayed start for returning a clock vector.
Once a legitimate clock vector is returned , the eda tool is
expected to use that information to sample the wave
returned by the getwave call. Other wise the eda tool may rely on
its own devices to sample the wave, (usually this is an 'ideal' clock)
4. some models may modify the wave form but may never return a clock
vector. (i.e -1 in the first position)
------------------------------------------------------------------------
*From:* "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
<mailto:Arpad_Muranyi@xxxxxxxxxx>
*To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
*Sent:* Fri, March 26, 2010 4:08:16 PM
*Subject:* [ibis-macro] Question on clock_times
Hello AMI experts,
I would like to ask a question about the clock_times vector.
The spec says that "The last clock in indicated by putting a
value of -1 at the and of clocks for the current wave sample".
We got a DLL from a vendor which seems to fill the vector
for the first few GetWave calls with -1 and then it starts
filling it with good numbers.
What is the interpretation of the specification? First,
does the above text refer to the entire vector, or each
little section independently that a single GetWave call
returns?
If it is for the entire vector, then having -1 in the first
few thousand locations would mean that everything else
after that is garbage, correct?
If it is per GetWave call, is the EDA tool expected to
assemble something from bits and pieces that are returned
by the various GetWave calls?
Thanks,
Arpad
===============================================================
---------------------------------------------------------------------
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
<mailto: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
