The ABINIT Users Mailing List ( CLOSED )

[prendergast2@llnl.gov]

Hello Abinitioners!

I recently downloaded the 4.2.3 and 4.1.4 linux pc binaries for abinip, 
compiled with the Intel Fortran Compiler.

Running on a Linux cluster, I noticed that all processors attempt to write 
output to the same output file simultaneously. This has produced problems 
especially as the number of processors increases. Mostly the code crashes 
with fortran file input/output errors - different processors trying to access 
the same files or nonexistent files.

I don't understand why this happens since the paral_use info file suggests 
that the mode of output is different. That is, one main processor handles 
output and separate log files are created by each other processor with the 
tmp prefix. Is this the case or not? Or am I using the binary incorrectly?

Generally, I run the code using a variant of prun called srun

srun -n #procs abinip < abinit.i > abinit.log

Any idea what's going wrong for me?
Apologies in advance if I've done something stupid.

David.