CCL:G: "Allocation inconsistency in PrSmSU." error in freq calculations

 Sent to CCL by: "Aleksey  Kuznetsov" [AlexKuznetsov2007_+_yandex.ru]
 Dear CCLers,
 I am doing geometry optimizations of ligated quantum dots using the HF/Lanl2dz
 approach as implemented in Gaussian 03, the quantum dots composition is either
 Cd33Se33(OPH3)21 or Cd33Se33(SH3)21, bith with 1341 basis functions and C1
 symmetry. When I tried to perform frequency calculations of non-completely
 optimized QDs (to use obtained force constants in order to speed up geometry
 optimizations), in both cases these calculations interrupted at some point
 producing the following similar error messages:
 ..
      3 vectors were produced by pass 16.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         9.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         9.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  PrismC:  NFx=      2048 NFxT=        10 NFxU=         8.
  Inv2:  IOpt= 1 Iter= 1 AM= 7.36D-16 Conv= 1.00D-12.
  Inverted reduced A of dimension   51 with in-core refinement.
  End of Minotr Frequency-dependent properties file   721 does not exist.
  Allocation inconsistency in PrSmSU.
  Error termination via Lnk1e in /home/dbchem/mapete/chem/g03-D.02/g03/l1101.exe
 at Sat Jan 15 03:09:14 2011.
  Job cpu time:  0 days  2 hours 56 minutes 59.6 seconds.
  File lengths (MBytes):  RWF=   4869 Int=      0 D2E=      0 Chk=    165 Scr=
 1
 I intentionally did again single-point calculations of these structures to make
 sure that chk-files were not lost or corrupted somehow. The memory amount
 assigned by the %mem keyword is either 320MW or 330MW. What could be the reason
 of this error and how it could be amended? Any help would be greatly
 appreciated.
 With best regards,
 Aleksey.