Skip to Content.
Sympa Menu

forum - [abinit-forum] abinit result stability.

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

[abinit-forum] abinit result stability.


Chronological Thread 
  • From: Samuel Sarholz <sarholz@rz.rwth-aachen.de>
  • To: forum@abinit.org
  • Subject: [abinit-forum] abinit result stability.
  • Date: Thu, 14 Aug 2008 10:52:11 +0200

Hi,

we are having trouble with abinit (abinis) results. They seem very unstable and differ a lot depending on the compiler used.

I used the 5.5.4 abinit version and 3 different compilers.
Intel 10.1 (10.1.015)
GNU 4.3.1
Sun Compiler (latest express release Sun Ceres Fortran 95 8.3 Linux_i386 2008/07/10)

I compiled with different optimization levels, but disabled floating point optimization that can change results.
However the results change with different optimization options and compilers.

With the Intel compiler I can use -O2 or abinit will crash with a segfault.

We have similar problems with abinit version 5.4.4, however I didn't test that extensively as I wanted to try the newest version first.

Does anyone here have similar problems?

Here some examples:
Intel -g :
-Cartesian components of stress tensor (GPa) [Pressure= 5.5580E-02 GPa]
- sigma(1 1)= -4.31249889E-02 sigma(3 2)= 1.35869431E-04
- sigma(2 2)= -7.84461342E-02 sigma(3 1)= -1.06868342E-03
- sigma(3 3)= -4.51687135E-02 sigma(2 1)= -3.42468306E-04

Intel -O2
-Cartesian components of stress tensor (GPa) [Pressure= -1.1747E-01 GPa]
- sigma(1 1)= 1.39267967E-01 sigma(3 2)= 1.21530532E-04
- sigma(2 2)= 7.15355586E-02 sigma(3 1)= -6.46687805E-04
- sigma(3 3)= 1.41609539E-01 sigma(2 1)= -3.21397785E-02

gfortan -O3 -mtune=opteron
-Cartesian components of stress tensor (GPa) [Pressure= -8.3018E-02 GPa]
- sigma(1 1)= 1.45328827E-01 sigma(3 2)= 7.75753140E-04
- sigma(2 2)= 1.01924105E-01 sigma(3 1)= -1.44278970E-03
- sigma(3 3)= 1.80179531E-03 sigma(2 1)= -1.98720137E-02

Sun f90 -fast -fsimple=0 -xtarget=opteron
-Cartesian components of stress tensor (GPa) [Pressure= -5.3268E-02 GPa]
- sigma(1 1)= 1.21039524E-01 sigma(3 2)= 9.50071267E-05
- sigma(2 2)= 7.07466991E-02 sigma(3 1)= 2.01962256E-04
- sigma(3 3)= -3.19837156E-02 sigma(2 1)= -1.56241767E-02

Other numbers of the final result do not differ a much e.g.:
xcart 3.8071129825E-03 3.6439365147E+00 3.6707808181E+00
xcart 3.0493453246E-03 3.6452959405E+00 3.6695516772E+00
xcart 3.0077708691E-03 3.6441605930E+00 3.6686544611E+00
xcart 2.8544487449E-03 3.6423896089E+00 3.6679849447E+00


best regards,
Samuel

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature




Archive powered by MHonArc 2.6.16.

Top of Page