[Wien] compiler options - ifort 8.0 / mkl 7.2 on AMD Athlon XP2500+?

[Manish Singh]

Dear Wien2k users,

I have Wien2k code compiled using ifort8.0/MKL7.2 running on an AMD Athlon
XP2500+. I ran the test_case and it took ~ 1500 seconds!!! I guess my
compiler options were not optimal. I went through the entire mailing list
but couldn't find anything useful.

Has anyone compiled Wien2k on an AMD Athlon machine using ifort 8.0 and MKL
7.2? What are the optimal compiler options for ifort8.0/mkl7.2 on an AMD
Athlon platform (32 bit)? 

The machine specifications are:
AMD Athlon XP 2500+ (1.8 GHz)
512 mb RAM
2.5 GB swap

Fortran Compiler: ifort 8.0
Math libraries: MKL 7.2

The options I'm using currently are:

FOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML
FPOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML
LDFLAGS:-L../SRC_lib -L/opt/intel/mkl721/lib/32 -Vaxlib
DPARALLEL:'-DParallel'
R_LIBS:-lmkl_lapack -lmkl_ia32 -lguide -lpthread
 
Regards,
Manish

 

________________________________

From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Joey Davis
Sent: Monday, May 08, 2006 9:58 PM
To: wien at zeus.theochem.tuwien.ac.at
Subject: [Wien] "rigid-band" model of version WIEN2K_06.3.


Dear WIEN2K users:
When I used "rigid-band" model to calculate PbTe's Fermi Energy 
changing with doping, I found a problem. I selected -8.0 Ry as 
ENERGY to separate core and valence states. After SCF cycles 
with spin orbit coupling, I changed NE in case.in2 from 36.0 
to 35.999, and run "x lapw2 -c -so", I found the Fermi Energy 
is not changed. If I set NE=36.001, the Fermi Energy become 
higher than the original one.
 How can I use "rigid-band" model to calculate the Fermi Energy of NE=35.999
?

 My wien version is WIEN2k_06.3.
 
Best Wishes!
Joey Davis


__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com