[Wien] Two questions
Jorissen Kevin
Kevin.Jorissen at ua.ac.be
Wed Jan 12 19:58:16 CET 2005
nmatmax 3500 is not enough for 64 atom cells. Stefaan's userguide on DFT and LAPW contains a strategy to estimate the number of basis functions needed.
Do grep :RKM case.scf : that will give you the effective rkmax, the best the program could do for your nmatmax. It's probably less than what you put into case.in1. (It's mainly rkmax that determines memory needs, not gmax).
However, you cannot increase nmatmax to very large numbers : the maximal size for that is limited by the available memory (if you put half of the matrix in the swap space, your calculation will become dramatically slower) and the number of memory adresses available ("32 bit limit").
I typically use nmatmax = 10 000 (or a little more) for super cells, corresponding to about 1.5 Gb of memory needed; that may be more than you need for the 64 atoms.
512 Mb RAM is, generally speaking, not enough for a 64 atom system. I'd advise you to upgrade to 1Gb or (better) 2 Gb. (You can estimate the memory needed : 8b per number, and nmatmax*nmatmax numbers in the hamiltonian. If you have a complex calculation, you need 16b per number. Some additional memory will be needed for other variables.)
Have you tried T. Andersen's suggestion to recompile lapw1 without any optimisation (use -O0)?
Kevin Jorissen
EMAT - Electron Microscopy for Materials Science (http://webhost.ua.ac.be/emat/)
Dept. of Physics
UA - Universiteit Antwerpen
Groenenborgerlaan 171
B-2020 Antwerpen
Belgium
tel +32 3 2653249
fax + 32 3 2653257
e-mail kevin.jorissen at ua.ac.be
________________________________
Van: wien-admin at zeus.theochem.tuwien.ac.at namens mkhsabra at aec.org.sy
Verzonden: wo 12-1-2005 17:20
Aan: wien at zeus.theochem.tuwien.ac.at
Onderwerp: Re: [Wien] Two questions
Dear Dr. Andersen;
Thanks a lot again for your help.
I have run Lapw's for many systems with 16 and 32 atoms per unit cell. But
for cells with 48 and 64 atom it does not.
I add swap, it was about 1 Gb and now more than 2 Gm , RAM is 512 Mb.
I tried Gmax 8 and 9 and K points about 20 , also it did not work.
I do not have the Intel compiler now, I note that NMATMAX parameter is 3500,
is it enough for cell with 48 or 64 atoms.
Than you
M. K. Sabra
> Dear Mr. Sabra,
>
> your problem seems to be lapw1, not lapw2... the "segmentation fault"...
> so either you are running out of memory (add swap space or RAM, or
> decrase Gmax), or you have a buggy compiler that causes addressing
> outside of the valid memory segments (big arrays), and you might try to
> solve it by relaxing the compiler optimizations (-O3 instead of -O5).
>
> Does your lapw1's run at all?
> Or does it happen already in the step where the k-points are divided?
>
> Try in a command-line to execute the command "x lapw1 -p" and see what
> comes out... if it works, try "x lapw2 -p"...
>
> However, I am wondering why the script doesn't stop before lapw2 then...
>
> Best regards,
> Torsten Andersen.
>
> mkhsabra at aec.org.sy wrote:
> > Dear Dr. Andersen;
> >
> > Thank you for your help.
> >
> > Stdout file:
> >
> > cp: canot stat '.in.tmp' no such file or directory
> > rm: canot remove = = == = = = == ==
> >
> > Dayfile:
> >
> > Lapwo -p
> > lapw1 -p
> > parllab1(5) segmantation fault
> > .. .. ...
> > .. ....
> >
> > lapw2 crashed
> >
> > lapw2.error file :
> >
> > testerror: Error in parallel lapw2.
> >
> > It works for cells with 16 or 32 atoms.
> >
> > Best wishes
> >
> > M. K. Sabra
> >
> >
> >>> Dear All wien users;
> >>>
> >>> 1- I wish you all a very happy new year.
> >>>
> >>> 2- I am trying to build a supercell (2x2x2) of InP. I have got 64 atoms
> in
> >>
> >>>the *.struct file. But also I have got error:
> >>> testerror: parallel lapw2 error.
> >>> How can I solve it. where the Gmax 9 ant no. k points is 27.
> >>
> >>Well, what is at the end of the dayfile? And in the other error files
> >>from lapw2?
> >>
> >>It could mean that communication is too slow between the computers, and
> >>it could also just be a statistical error... what happens if you run
> >>lapw2 again?
> >>
> >>
> >>> 3- The 64 atoms ar generated from 10 inequvelant atoms. Do the
> >>
> >>calculations
> >>
> >>>consider only the inequvelant atoms or all atoms in the struct file.
> >>>
> >>
> >>Do they have different (and negative) numbers? Otherwise sgroup will
> >>reduce the struct file again...
> >>
> >>A calculation always consider all atoms in the struct file, once
> >>init_lapw has been successfully run.
> >>
> >>
> >>
> >>> please help
> >>>
> >>> best wihes
> >>>
> >>> M. K. Sabra
> >>>
> >>>
> >>>
> >>>-------------------------------------------------
> >>>This mail sent through AECS mail server.
> >>>
> >>>_______________________________________________
> >>>Wien mailing list
> >>>Wien at zeus.theochem.tuwien.ac.at
> >>>http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >>>
> >>
> >>--
> >>Dr. Torsten Andersen TA-web: http://deep.at/myspace/
> >>AG Hübner, Department of Physics, Kaiserslautern University
> >>http://cmt.physik.uni-kl.de http://www.physik.uni-kl.de/
> >>
> >>Symposium on Excited-state properties of solids, Mannheim 2005:
> >>See: http://cmt.physik.uni-kl.de/XSM05/
> >>
> >>_______________________________________________
> >>Wien mailing list
> >>Wien at zeus.theochem.tuwien.ac.at
> >>http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >>
> >
> >
> >
> >
> >
> > -------------------------------------------------
> > This mail sent through AECS mail server.
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
>
> --
> Dr. Torsten Andersen TA-web: http://deep.at/myspace/
> AG Hübner, Department of Physics, Kaiserslautern University
> http://cmt.physik.uni-kl.de http://www.physik.uni-kl.de/
>
> Symposium on Excited-state properties of solids, Mannheim 2005:
> See: http://cmt.physik.uni-kl.de/XSM05/
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
-------------------------------------------------
This mail sent through AECS mail server.
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 12069 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20050112/ab3b21c9/attachment.bin
More information about the Wien
mailing list