Author Topic: Frequency calculations crush for larger molecules  (Read 6283 times)

avd

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Frequency calculations crush for larger molecules
« on: February 23, 2011, 05:07:16 pm »
I'm trying to evaluate IR frequencies of a molecular complex of about 40 atoms at DFT+dispersion level with B3Lyp functional and def2-TVZP basis set (I useTimoleX 3,01 interfact for setting of the paratmeters. ) It ran smoothly for smaller systems (10-12 atoms) with the same settings but now it is constantly crushing at the end of the calculation process.  Typical error message is following:
" ridft ended normally
PGFIO/stdio: Invalid argument
PGFIO-F-/formatted write/unit=9/error code returned by host stdio - 22.

 File name = control    formatted, sequential access   record = 0

 In source file plumm.f, at line number 82"

Changing basis to 6-311++G** didn't help. Geometry optimizations worked fine.
Does anyone encounter similar problems?

thank you in advance

antti_karttunen

  • Sr. Member
  • ****
  • Posts: 200
  • Karma: +1/-0
Re: Frequency calculations crush for larger molecules
« Reply #1 on: February 23, 2011, 05:36:21 pm »
Hi,

How much free disk space do you have on the computer you are using for the calculation? Aoforce requires quite a lot of scratch space for larger systems.

Regards,
Antti

avd

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
Re: Frequency calculations crush for larger molecules
« Reply #2 on: February 24, 2011, 10:19:00 am »
How much free disk space do you have on the computer you are using for the calculation?

about 160 Gbytes diskspace are free in every of two disk partitions, and the computer is quite new in works nice and stable. How large can be the scratch space?

antti_karttunen

  • Sr. Member
  • ****
  • Posts: 200
  • Karma: +1/-0
Re: Frequency calculations crush for larger molecules
« Reply #3 on: February 24, 2011, 11:15:28 am »
The total disk usage depends on the size and the symmetry of the system, so it's difficult to say. I guess 160 GB should be enough for your system, though. In fact, now I realized that the error occurs during a write to the control file and not any scratch file. Maybe it's better to contact the Turbomole support in your case.