The ABINIT Users Mailing List ( CLOSED )

[sara yazdani <sarayazdani@yahoo.com>]

Hello all,

Taking the previous advice, I ran the input, but after finishing the first loop of ddk and receiving the first DEN, WFK and ... the run haulted.

It stopped on status number "207124" it is over three hours that I have been stopped! but it is using up 2.7GiB of memory and the 4 processors are 100% busy!!!! Could you let me know what I should do? Should I head for running such heavy job sequentially, or there's a way to solve it?

Regards,
Sara




--- On Tue, 9/12/08, matthieu verstraete <matthieu.jean.verstraete@gmail.com> wrote:

From: matthieu verstraete <matthieu.jean.verstraete@gmail.com>
Subject: Re: [abinit-forum] frustrating RF calculation
To: sarayazdani@yahoo.com
Cc: forum@abinit.org
Date: Tuesday, 9 December, 2008, 11:54 AM

Hello Sara,

a dirty fix is fairly simple: in src/18seqpar/loper3.F90, around line 1220, comment the allocation of gh1_rbz and replace it by

!      allocate(gh1_rbz(nkpt_rbz,dtset%mband,2,mpw1*nspinor))
      allocate(gh1_rbz(1,1,1,1))

Recompile and run. This will of course crash if you try to use Paul's developments, but should be fine for you (keep ieig2rf to its default value of 0).

Cheers

Matthieu

On Tue, Dec 9, 2008 at 11:13 AM, sara yazdani <sarayazdani@yahoo.com> wrote:

Hello Paul and abinit users,

Thank you for your reply. Since I am in short of time to present my results and I have done some other good calculations with abinit, please let me know what I should do instead to get better phonon frequencies. Since I used "rfasr=1 and 2" for all the calculations as well and the result was worse.

Regards,
Sara Yazdani



--- On Tue, 9/12/08, Paul Boulanger <paul.boulanger@umontreal.ca> wrote:

From: Paul Boulanger <paul.boulanger@umontreal.ca>
Subject: Re: [abinit-forum] frustrating RF calculation
To: forum@abinit.org, sarayazdani@yahoo.com
Date: Tuesday, 9 December, 2008, 1:58 AM

Hello Sara,

For the 8x8x8 grid, the memory allocation error is actually
 my fault, I
included in past versions an array that scales badly with the number of
K-points. This is fixed in my 5.6.5 public branch. I am terribly sorry if this
issue as bothered/hindered anybody else. I will discuss with Xavier to see what

is the best way to make it accessible for you and others before the release of
5.6.5.

Paul

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF) 
Dpto. Fisica de Materiales, 
U. del Pais Vasco,            
Centro Joxe Mari Korta, Av. de Tolosa, 72,   Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500