[ns] R: Re: R: which trace format is more accurate?

[Marco Fiore]

Sasan,

don't have a clue on that.
I fear the only way is to analyze 
the trace files
by hand to spot the differences. Maybe printing
the 
sending times (using an awk script) can
speed up the process.

Let us 
know if you can solve the problem.

Regards,

Marco Fiore

>----
Messaggio originale----
>Da: ss1 at student.cs.ucc.ie
>Data: 19-apr-2006 
12.47 PM
>A: "Marco Fiore"<fl0wer at virgilio.it>
>Cc: <ss1 at student.cs.ucc.
ie>, <ns-users at isi.edu>
>Ogg: Re: R: [ns] which trace format is more 
accurate?
>
>Marco,
>
>This is the exact point,
>
>the difference is 
about 300 to 400 for 100 sec sim time
>
>I reran the same simulation 
and all I did was to change the trace format,
>but I see such a 
difference in number of sent
>
>$1~/s/ && /AGT/  { sent ++; }
>
>any 
comments?
>
>both running on the same machine. all the equipment and 
environment
>settings are the same. you might be able to reproduce it 
easily.
>
>Thanks,
>Sasan
>
>
>> Sasan,
>>
>> I see your point.
>> How 
much do differ results you got in the two
>> cases?
>> By exact same 
simulation you mean that you simply
>> changed the
>> trace format and 
nothing else (included
>> the machine you working on)?
>>
>> Regards,
>>
>> Marco Fiore
>>
>>>----Messaggio originale----
>>>Da: ss1 at student.
cs.
>> ucc.ie
>>>Data: 15-apr-2006 9.33 PM
>>>A: "Marco Fiore"
<fl0wer at virgilio.it>
>>>Cc: <ss1 at student.cs.ucc.ie>, <ns-users at isi.edu>
>>>Ogg: Re: R: [ns]
>> which trace format is more accurate?
>>>
>>>Marco,
>>>
>>>I agree that I need
>> separate awk scripts for 
different trace formats but
>>>the particular
>> example that I gave 
below is straight forward
>>>
>>>$1~/s/ && /AGT/  {
>> sent ++; }
>>>
>>>and should return total number of sents message in a
>> simulation 
and that
>>>still returns different values (unless this is not
>> the 
correct way to
>>>calculate the total sent messages)
>>>
>>>what do you
>> think?
>>>
>>>Thanks,
>>>Sasan
>>>
>>>
>>>
>>>> Sasan,
>>>>
>>>> 
both versions of
>> tracefile report the same information.
>>>> The
>>>> difference only lies
>> in the format (actually, the
>>>> new 
trace format is
>>>> somehow more
>> complete).
>>>> So they are 
accurate in the same way, but the
>>>> point
>>>>
>> is that you can't 
use the same awk script on both!
>>>> You have to
>>>>
>> write two 
versions of you awk script,
>>>> and then you'll get the same
>>>> 
results.
>>>>
>>>> Regards,
>>>>
>>>> Marco Fiore
>>>>
>>>>>----
Messaggio
>> originale----
>>>>>Da:
>>>> ss1 at student.cs.ucc.ie
>>>>>Data: 15-apr-2006 8.06
>> AM
>>>>>A: <ns-users at ISI.EDU>
>>>>>Ogg: 
[ns] which trace format is more
>> accurate?
>>>>>
>>>>>
>>>>>Hi All,
>>>>>
>>>>>I am
>>>> generating both trace
>> formats (old and new) for 
the "exact" same
>>>>>simulation, and for
>> instance I count total 
number of messages sent as
>>>>>below (in my awk
>> file)
>>>>>
>>>>>$1~/s/ && /AGT/  { sent ++; }
>>>>>
>>>>>but I get
>>>> two
>> 
different results
>>>>>
>>>>>now I wonder, which one of the trace files 
are
>>>> accurate? which one should
>>>>>I pick? the version is 2.29.
>>>>>
>>>>>I'll
>>>> appreciate your advice.
>>>>>
>>>>>Thanks,
>>>>>Sasan
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>