The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

Hello all

I'm running phonon calculations of my habitual Mac OSX G5/xlf platform 
(~30 cpu). For gamma everything runs fine, but all the other kpoints 
segfault or crash. abinit v 5.4.4 and latest 5.5devel

1) If I run with everything in memory (normal), it tries to allocate a 3Gb 
array, and crashes. The log claims it only needs 700 Mb, but I suspect the 
abinit estimator is wrong (because gamma works fine with 20x less kpoints)

2) If I run with mkmem=mk1mem=mkqmem=0, it stops during i/o on the 
temporary files (first tmpPb_WFGS_P-00XX, then tmpPb_WFQ_P-00XX if I 
remove mkmem=0) with the message:

-P-0000  rwwf: ERROR -
-P-0000   Reading option of rwwf. Trying to read
-P-0000   the (npw,nspinor,nband) record of a wf file, unit=   3
-P-0000   gave iostat=  -1. Your file is likely not correct.
-P-0000   Action: check your input wf file.

But this file has just been written by abinit, and the filesystem is fine 
and almost empty! There is nothing else to see in the log.


3) Running abinis with mkmem=mk1mem=mkqmem=0 gives the same 3Gb allocation 
crash as above, so the two might not be linked, in which case the 3Gb is a 
memory leak we have to debug.


This could be a compiler bug: has anyone had the same problems, or found a 
platform where this type of calculation works? In particular the mk*mem=0 
for response function calculations?

I will try to reproduce this with a smaller run, but so far nothing.

thanks and have a nice weekend!

Matthieu


--
================================================================
Dr. Matthieu Verstraete                 mailto:mjv500@york.ac.uk
Dept. of Physics, University of York,     tel: +44 1904 43 22 08
Heslington, YO10 5DD York, United Kingdom fax: +44 1904 43 22 14ndtset 3
jdtset  3 4 5

#prtvol 10
mkmem 0
mk1mem 0
mkqmem 0

ecut  16.0
nspinor 2
# pspso is implicit from pseudopot file

acell 3*6.6250714
angdeg 60.0 60.0 60.0
xred 0.0 0.0 0.0

nqpt   1
qpt                 5.00000000E-01  0.00000000E+00  0.00000000E+00

rfatpol3  1 1 rfdir3    1 0 0 
rfatpol4  1 1 rfdir4    0 1 0 
rfatpol5  1 1 rfdir5    0 0 1 
tolvrs   1.0e-9
getwfk   1
prtwf   0
kptopt  3
rfphon 1
prtgkk 1

###########################################

ngkpt 32 32 32

occopt 7
tsmear 0.001
nband 10
prteig 1
nshiftk 1
shiftk 0.0 0.0 0.0
natom 1
ntypat 1
typat 1
znucl 82
nstep 1000
#chkexit 0