gennbo on large files

gennbo on large files

Postby vitko » Mon Aug 17, 2015 12:37 pm

Dear staff,
I have problems in running gennbo on large molecules e.g. 98 atoms 4317 NBAS as written in the .47 file obtained from G09. The test files like ch3.47 go well. Small .47 files (NATOMS=32 NBAS=1360) constructed by me via G09 are ok too (I am interested in STERIC keyword). But if I try to run the large file with gennbo command, after 4-5 hours of running, it prints out NBOs but then ends without printing the steric analysis data. The so obtained .nbo file is not terminated normally.
The current settings of limits are:
- for src/nbo/include
maxmem=250000000
maxatm=1250
maxbas=12500
maxshl=5000
maxexp=12500
- for src/gennbo/include
maxmem=250000000
maxatm=1250
maxbas=12500
maxshl=5000

Could you comment upon this? Is there any "walltime" limit in the program that makes it terminate after a certain time the code has been run for? And, is there any way to extend what it seems to be (my system adm. wrote me this) an unbreakable limit of 2GB that makes impossible the extension of maxmem beyond 250 000 000 words?

Thank you very much in advance!!!
vitko
 
Posts: 2
Joined: Mon Aug 17, 2015 10:00 am

Re: gennbo on large files

Postby ericg » Mon Aug 17, 2015 4:22 pm

Do you have the most recent release of NBO6 (10-Aug-2015)? This release includes modifications to GenNBO that makes the parsing of FILE47 considerably more efficient. This might solve your issues.

Eric
ericg
 
Posts: 270
Joined: Sat Dec 29, 2012 9:31 am

Re: gennbo on large files

Postby vitko » Tue Aug 18, 2015 9:34 am

I am waiting the answer from the tech. support on the cluster I am sending my jobs to. Two days ago he said to me that he checked if there are no any new updates, so the version I am using should be updated. But I asked specifically if he updated using the 10-August-2015 one. In the while, I can provide more info on the output I got. The .nbo ends without any error message (like memory allocation problems I get with the default version i.e. not the memory extended one that I use now, as you see from the parameters above) just in the beginning of NLMO analysis (for this molecule this is the case, I dont know if with a smaller molecule the job will proceed further or not). Can it be that I use really huge basis sets in my G09 job (quadruple zeta with polarization and diffusion functions)? The .47 file says NBAS=4317 that is lower than the maxbas parameter. I dont know if there can be other problems with huge basis set I use. Also, If I use non-memory-extended version (the default distribution), then the error message is

"This calculation is attempting to allocate 18654367 words of scratch memory,
although GenNBO is currently configured with only 10000000 words. If you have
the NBO6 source distribution, you can reconfigure GenNBO by editing maxmem.fh
in nbo6/src/gennbo/include and rebuilding the executables."

If we believe that the number of words necessary for my job is 18654367, then also the current word number limit i.e. maxmem = 250 000 000 is ok.
Therefore I am out of clues where to look in order to solve the problem.
vitko
 
Posts: 2
Joined: Mon Aug 17, 2015 10:00 am

Re: gennbo on large files

Postby ericg » Tue Aug 18, 2015 11:59 am

I noticed that an old release (9-Mar-2015) is currently available on the NBO6 download page, not our most recent release. This should be corrected by tomorrow morning (19-Aug), so you should wait until then to download the most recent source.

I've run default NBO analysis (without STERIC and NLMO analysis) with as many as ~4700 basis functions. You are probably exploring new territory here with such large calculations, so you might expect some trial and error before a successful calculation is completed.

Eric
ericg
 
Posts: 270
Joined: Sat Dec 29, 2012 9:31 am


Return to NBO Installation

Who is online

Users browsing this forum: No registered users and 1 guest

cron