Skip to Content.
Sympa Menu

forum - [abinit-forum] Segmentation fault while writing KSS file (Pre-yambo issue)

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

[abinit-forum] Segmentation fault while writing KSS file (Pre-yambo issue)


Chronological Thread 
  • From: "Fernando Vila" <f.d.vila@gmail.com>
  • To: forum@abinit.org
  • Subject: [abinit-forum] Segmentation fault while writing KSS file (Pre-yambo issue)
  • Date: Mon, 6 Oct 2008 11:20:14 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:mime-version:content-type; b=arnl+PU2MWdYBWA262DchM7pdhbqMixgAGa/flqTXYHgX/pgTwsy5LJ1E2Vju1bAQH XsDOnRE7JuY4yxNQdYPrFtJk3iBC7rL2bEBYoLV6rGYumj4EuyNPKjXdJSytSEbxuBNi cVAXHcMSjaXBIa7TzvXLeHMHbg/jTA9NQ82ck=

Dear all:

  I recently decided to give yambo a try but I run into some problems at the abinit stage. I took the LiF test run included in the yambo distribution and run it for 4x4x4, 6x6x6 and 8x8x8 k-point grids. The former two work correctly, while the 8x8x8 grid breaks down with a segfault while writing the KSS file.

  Since I don't like to just complain, I took the effort to locate where is that the segfault occurs: It happens in the following line of outkss.F90 (indicated with the arrow, line 1436 in my 5.4.4 distribution):

!MG CHECK THIS PART !MG cg(2,mpw*nspinor*mband*mkmem*nsppol)=planewave coefficients of wavefunctions.
!MG the increment of k_index is : k_index=k_index+npw_k*nband_k*nspinor
    else
     do ib=1,nbandksseff
      do ig=1,npwkss
       if(trsl(ig)>npw_k) cycle
       wfg(:,ig,ib)=cg(:,trsl(ig)+(ib-1)*npw_k*nspinor+k_index)                                                          <<<<<---------
      end do
     end do
    end if
   else

  I was wondering if either the abinit or yambo developers (which probably look at this forum :-) have seen this before. Here is a list of the things I tried to fix the problem which didn't work (most of them gleaned from a search of this forum):

1) Made stack unlimited.
2) Tried versions compiled with both intel and pgi fortran compilers.
3) Tried different values of mkmem (this does make a slight difference, but it still crashes).
4) Tried using different number of processors. This changes the place (in memory, i.e. the k-point number) where the segfault happens.

I'm attaching the input and output files for a typical run (using 30 processors). The abinit version used was 5.4.4.

Thanks in advance for any suggestions,

                                                                          Fer.

Attachment: 3D_LiF.files
Description: Binary data

Attachment: 3D_LiF.in
Description: Binary data

Attachment: 3D_LiF.log.gz
Description: GNU Zip compressed data

Attachment: 3D_LiF.out.gz
Description: GNU Zip compressed data



  • [abinit-forum] Segmentation fault while writing KSS file (Pre-yambo issue), Fernando Vila, 10/06/2008

Archive powered by MHonArc 2.6.15.

Top of Page