[Re: Termination of application process]

Subject: Re: [lammps-users] Termination of application process
From: "Axel Kohlmeyer" <[email protected]>
Date: Fri, June 28, 2013 1:33 pm
To: "Saikat Basu" <[email protected]>
--------------------------------------------------------------------------

you are not making it easy to help you, on the contrary.

axel.

I am posting my input script here,

where is the data file? one cannot run this input without it.
also, best attach as gzipped files or tar.gz archive, since copying an
inline script and particularly removing the quotation characters is a
lot of (needless and tedious) effort.

remember, i said "you have to make it easy to help you". you are not
exactly doing that.

axel.

---------------------------- Original Message
----------------------------
Subject: Re: [lammps-users] Termination of application process
From: "Axel Kohlmeyer" <[email protected]>
Date: Fri, June 28, 2013 1:33 pm
To: "Saikat Basu" <[email protected]>
--------------------------------------------------------------------------

you are not making it easy to help you, on the contrary.

axel.

I am posting my input script here,

where is the data file? one cannot run this input without it.
also, best attach as gzipped files or tar.gz archive, since copying an
inline script and particularly removing the quotation characters is a
lot of (needless and tedious) effort.

  I am attaching both the input script as well as the data file.

Thanking you,

Best,
******************
Saikat Basu
Junior Research Fellow
Department of Solid State Physics
IACS,Kolkata
Kolkata-700032
Ph no- 03324734971(extn- 2114)

--
Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

Thanking you,

Best,
******************
Saikat Basu
Junior Research Fellow
Department of Solid State Physics
IACS,Kolkata
Kolkata-700032
Ph no- 03324734971(extn- 2114)

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
lammps-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/lammps-users

--
Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

Thanking you,

Best,

LAMMPS.tar.gz (885 KB)

where is the data file? one cannot run this input without it.
also, best attach as gzipped files or tar.gz archive, since copying an
inline script and particularly removing the quotation characters is a
lot of (needless and tedious) effort.

  I am attaching both the input script as well as the data file.

so it has taken you five (5!) emails to do what is explained in the
manual and has been said on this mailing list over and over again in
case you have a problem where you suspect that something is not
working correctly.

i ran the input with the current version of LAMMPS and it runs fine for me.

so you have have to provide more details how *exactly* this can be reproduced.
axel.

where is the data file? one cannot run this input without it.
also, best attach as gzipped files or tar.gz archive, since copying an
inline script and particularly removing the quotation characters is a
lot of (needless and tedious) effort.

  I am attaching both the input script as well as the data file.

so it has taken you five (5!) emails to do what is explained in the
manual and has been said on this mailing list over and over again in
case you have a problem where you suspect that something is not
working correctly.

I admit, there were a number of mistakes from my part in the previous
posts.

If you remember , in the very first mail(subject: Termination of
application process)I have attached my screenshot.
Before posting, I have verified it in the 'error section' in the LAMMPS
website. And as there were no such evidence to counter the error I had
to post it in the mail list.

In the reply ,you asked me to do the followings

"this report is far too generic and there is not enough information to

draw any conclusions.

you should try running the same input on a different machine with the
same (or newer) version of lammps.

you should try to determine, if the problem happens only in parallel
or also with the serial version

you should try to generate a core dump and then get a stack trace
using a debugger (i gave a detailed description of how to do this
several times on this mailing list, so it should be somewhere in the
mailing list archives).

if that all leads to no tangible evidence, you can post your
complete(!) input deck to the mailing list here and see if somebody
else can track it down."

After failing to which I again wrote in the list.

i ran the input with the current version of LAMMPS and it runs fine for
me.

Is it running for any number of processors ?
My point was to verify whether there is any error in my script for
which I couldn't be able to run it in more than 6 processors.

so you have have to provide more details how *exactly* this can be
reproduced.

I have to check again.

Thanking you,

Best,
******************
Saikat Basu
Junior Research Fellow
Department of Solid State Physics
IACS,Kolkata
Kolkata-700032
Ph no- 03324734971(extn- 2114)

--
Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

Thanking you,

Best,

i ran the input with the current version of LAMMPS and it runs fine for
me.

Is it running for any number of processors ?

i ran with 4. you never mentioned that was an issue.

My point was to verify whether there is any error in my script for
which I couldn't be able to run it in more than 6 processors.

and here is *another* piece of information that would have been
*extremely* useful to go with your files.
remember that by looking at the 5th e-mail, i have long forgotten what
you wrote in the first.

one thing that sticks out are those many crazy and pointless tweaks to
the neighbor list and communication settings. why do you need them?

axel.

i ran the input with the current version of LAMMPS and it runs fine for
me.

Is it running for any number of processors ?

i ran with 4. you never mentioned that was an issue.

  I had posted it today at 3:42 am (IST).

  "I have tried all sorts of possibilities that Axel opined in

the earlier post.

It's running in my Fujitsu cluster and also in CRAY. And even
in my work station its also running.

But the problem arises when I am trying to run it in more than 6
processors. In all cases the error message which I have posted before
is coming out with an additional error ' SEGMENTATION FAULT'.

But the examples code which have been provided in the LAMMPS example
directory are running for any number of processors.

Is there anything wrong with my input script ? "

Probably you might missed the section.
Yes, It was the issue. What is the problem which leads to the
failure to execute it.

and here is *another* piece of information that would have been
*extremely* useful to go with your files.
remember that by looking at the 5th e-mail, i have long forgotten what
you wrote in the first.

one thing that sticks out are those many crazy and pointless tweaks to
the neighbor list and communication settings. why do you need them?

Yes, I have lots of confusion in this part.
I have to go through the manual again and check with the existing scripts
I have in the example directory.

Thanking you,

Best,
******************
Saikat Basu
Junior Research Fellow
Department of Solid State Physics
IACS,Kolkata
Kolkata-700032
Ph no- 03324734971(extn- 2114)

--
Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

Thanking you,

Best,

i ran the input with the current version of LAMMPS and it runs fine for
me.

Is it running for any number of processors ?

i ran with 4. you never mentioned that was an issue.

  I had posted it today at 3:42 am (IST).

  "I have tried all sorts of possibilities that Axel opined in

the earlier post.

It's running in my Fujitsu cluster and also in CRAY. And even
in my work station its also running.

But the problem arises when I am trying to run it in more than 6
processors. In all cases the error message which I have posted before
is coming out with an additional error ' SEGMENTATION FAULT'.

But the examples code which have been provided in the LAMMPS example
directory are running for any number of processors.

Is there anything wrong with my input script ? "

Probably you might missed the section.
Yes, It was the issue. What is the problem which leads to the
failure to execute it.

and here is *another* piece of information that would have been
*extremely* useful to go with your files.
remember that by looking at the 5th e-mail, i have long forgotten what
you wrote in the first.

one thing that sticks out are those many crazy and pointless tweaks to
the neighbor list and communication settings. why do you need them?

Yes, I have lots of confusion in this part.
I have to go through the manual again and check with the existing scripts
I have in the example directory.

Thanking you,

Best,
******************
Saikat Basu
Junior Research Fellow
Department of Solid State Physics
IACS,Kolkata
Kolkata-700032
Ph no- 03324734971(extn- 2114)

--
Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

Thanking you,

Best,

Is there anything wrong with my input script ? "

Probably you might missed the section.

as i said, i *cannot* remember what was written earlier. i read and
answer a large number of e-mails on a daily basis. i cannot remember
every little detail.

*you* want the help, so it is *your* job to provide a *complete* and
*consistent* description. if that means to repeat things, than so be
it. *you* want somebody else to do something for you (and for free,
too), so it is your responsibility.

[...]

one thing that sticks out are those many crazy and pointless tweaks to
the neighbor list and communication settings. why do you need them?

Yes, I have lots of confusion in this part.
I have to go through the manual again and check with the existing scripts
I have in the example directory.

it has been explained many, many times, that the procedure to debug a
script is to take out everything and then put it back step by step and
evaluate each feature separately.

with the additional information, i could determine the location of the segfault.

which you could also have identified by following the method from
above, as it is triggered by using a large communication cutoff
in combination with dump image. you turn off dump image or use
"communicate single cutoff 2.0" and there is no segfault.

i repeat, this is something that is being explained on this very
mailing list many, many times. spend some time reading through the
conversations and you will learn a lot. and as a bonus, you will keep
people that can help you happy instead of frustrating them.

axel.