a2y: difference by running a script and from command line

[feffeficus]

hi dear yambo guys!
i'm please to post here my first official question:

i'm running a2y and by running it from command line and from a script submitted to the queue I obtain different responses:


from command line
/home/fiori/CODES/yambo-3.2.1-r.448/bin/a2y -N -S -F out_KSS

a2y recognize the kss file but it runs over memory (and it's normal...it's a 21 GB kss.)

here the output;'
<---> [01] A(binit) 2 Y(ambo)
<---> Checking input file ...out_KSS
<---> DBs path set to :.
<---> KSS Header...forrtl: severe (41): insufficient virtual memory
Image PC Routine Line Source


by running from this script to a 8Gb memory queuewith this script:

#!/bin/sh
#PBS -S /bin/sh
#PBS -N pi666shfkss
#PBS -q mem_8
#PBS -l nodes=1:ppn=1
#PBS -r n
#PBS -o pbslog
#PBS -j oe
cd $PBS_O_WORKDIR

/home/fiori/CODES/yambo-3.2.1-r.448/bin/a2y -N -S -F out_KSS


it doesn't recognize the kss file to convert:
<---> [01] A(binit) 2 Y(ambo)
<---> Checking input file ...
<---> DBs path set to :.



can you help me?????
thanks a lot!

[andrea marini]

from command line
/home/fiori/CODES/yambo-3.2.1-r.448/bin/a2y -N -S -F out_KSS

a2y recognize the kss file but it runs over memory (and it's normal...it's a 21 GB kss.)

here the output;'
<---> [01] A(binit) 2 Y(ambo)
<---> Checking input file ...out_KSS
<---> DBs path set to :.
<---> KSS Header...forrtl: severe (41): insufficient virtual memory

Dear Feffe,

can you check carefully where a2y is stopping ? You're in a2y/a2y_db1.F and if it is a problem of wavefunctions size it should stop a little bit later, after, for example, the message ":: Atoms positions...". Put some stop by hand in a2y_db1 to check where the insufficient virtual memory error pop up.

by running from this script to a 8Gb memory queuewith this script:
....
it doesn't recognize the kss file to convert:
<---> [01] A(binit) 2 Y(ambo)
<---> Checking input file ...
<---> DBs path set to :.

This sounds strange, very strange. If a2y does not see the KSS file it should write

Code: Select all

 <---> [01] A(binit) 2 Y(ambo)
 <---> Checking input file ...failed

Instead in your case it does not write anything after "Checking input file ..." but it does not stop It could be related to a well known problem of some mpirun executables that does not allow p2y (and in general the C driver used by Yambo) to accept options. Simply a2y does not see the -F option but, again, in this case it should stop.

Feffe can you check in the code the value of the function

Code: Select all

a2y_KSS_file_name(inf,KSS_file_name) 

in a2y_i.F ? Because an empty KSS_file_name string should return a negative integer and, thus, the error message. In your case I cannot understand what is the value of a2y_KSS_file_name and what is the string KSS_file_name.


P.S.: Remeber to add your affiliation

[myrta gruning]

ciao Feffe

I do not think that the problem is due to the fact the program does not find/recognize the KSS file.
It seems to me more that the job is crashing for mem problems as well, just the error message is not written in the log (it often happens with crashing batch jobs that the error would not appear in the log). Did you inspect the standard err and out files? I think that an error message will appear there, not in the log.

Also, did you ever experience such a problem with a smaller kss file? If not I would suspect it is still a problem with the memory (21 Gb of kss file).

It could be related to a well known problem of some mpirun executables that does not allow p2y (and in general the C driver used by Yambo) to accept options. Simply a2y does not see the -F option but, again, in this case it should stop.

I do not think this is the case. Usually (if it is the same problem we are talking about) this problem appear when calling the program with mpirun (it is due to the mpiscript, and the program does accept options, but alas it takes as well those of the mpirun script), and this is not the case here. As you said the error message will be different.

cheers,
m

[feffeficus]

hi mirta!

thanks for the reply. but just to understand:

a2y is not sequential?

[myrta gruning]

hi mirta!

thanks for the reply. but just to understand:

a2y is not sequential?

I messed up things. The part relative to mpirun was for Andrea, and not relevant for your problem.

To answer, if you enable mpi in the configure, then a2y can be run with mpirun. I frankly don't know if this will be of any use (i.e. if tasks are effectively parallelized, actually it does not seem so).

Sorry for this aside discussion on the mpirun script and good luck with converting your huge kss file.

m