[CPMD-list] different results from P4 and Opteron

[Axel Kohlmeyer]

On Wed, 20 Apr 2005, Sombat Ketrat wrote:


dear sombat,

SK> Hi CPMD users,
SK> 
SK> I just recompiled the CPMD for Opteron system with pgf90 v.5.2-4 without 
SK> any error. First I check the binary with the testing jobs provided by 
SK> cpmd.org.

compiling a fortran code without any errors does not say a lot
about whether the result is running correctly. but this is what
test jobs are for. ;-)

SK> For the job with use VANDERBILT PP (1,4-benzoquinone), I got the 
SK> difference resulting energies and VANDERBILT AUGMENTATION CHARGES (as 
SK> show below) from p4 and Opteron.

in general you should not pay too much attention to that quantity 
as it is displayed in the cpmd output. it should be the same when
running identical jobs, though.

SK> Are there anything gone wrong with my binary file. Where is the souce of 
SK> the different results? I also attach the input filed output provided by 
SK> cpmd.org and my calculated outputs as well.

you seem to have a miscompiled opteron binary.
the pgi compilers are known for not always being fully reliable 
at full optimization levels. i just checked on a machine 
with pgi-5.2-1 installed, where i could reproduce your (wrong?) 
opteron numbers with optimization turned on and got your
p4 numbers with -O0.

using the intel compiler in 32- and 64-bit with full optimization
and different libraries, i always get your p4 numbers to the last digit.

regards,
	axel.

SK> 
SK> Thank you in advanced.
SK> Sombat
SK> 
SK> **
SK> 

-- 

=======================================================================
Dr. Axel Kohlmeyer   e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.