[Wien] newbie on the list and AMDOpteron64

Florent Calvayrac Florent.Calvayrac at univ-lemans.fr
Mon Sep 19 11:52:19 CEST 2005


Dear list

As a new member of the list (since Prof.Silly, the usual correspondant 
for this site, was promoted
to Montpellier), I want to share my experience with our new cluster
made of AMDOpteron64 nodes.

Wien works as well with AMD performance libraries as with Intel MKL ;
however, I could not manage to have lapw1 terminate correctly
whatever the compile options with Intel Fortran 9.0 (fce)
until  I changed the following line in TAPEWF routine :


DOUBLE PRECISION SX,SY,SZ,WEIGHT

to

REAL *8 SX,SY,SZ,WEIGHT

(the same for SXHELP).


indeed, in lapw2, those values  are reread as REAL*8



For information, my options (with MKL) are :


current:FOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -i_dynamic 
-mcmodel=medium
current:LDFLAGS:-L/opt/intel/fce/9.0/lib/ -L/opt/intel/mkl/8.0/lib/em64t 
-Vaxlib -i_dynamic -mcmodel=medium -lcxa -pthread -lunwind
current:DPARALLEL:'-DParallel'
current:R_LIBS:-lmkl_lapack -lmkl_em64t -lguide


and compiler

/opt/intel/fce/9.0/bin/ifort


The mcmodel and i_dynamic options are here to help use more than 2Gb of 
memory and
I use dynamic libraries on the nodes.





-- 
Florent Calvayrac  | Tel : 06 64 31 43 86  | Tel  : 02 43 83 26 26
Directeur du SC Informatique Ressources Num.  de l'Universite du Maine
Lab. de Physique de l'Etat Condense UMR-CNRS 6087 
Inst. de Rech. en Ingenierie Molec. et Matx Fonctionnels  FR CNRS 2575 



More information about the Wien mailing list