memory release problem on CCSORT


[ Molcas user's WWWBoard ]

Posted by Hirotoshi Mori on May 23, 2008 at 02:53:42:

Dear MOLCAS developers and users,

This is Hirotoshi Mori at Ochanomizu University, Japan.
Today, I'm posting this message since I have a question relating with memory on MOLCAS7sp1.

Now, I'm doing CCSD(T) calculation of some small molecules (7atom molecule) with c.a. 200 basis functions, and their isomers with different symmetry.
When I perform the calculation with C2v or Cs symmetry, I did not have any problem.
However with higher symmetric D2h group, I got the next error message.
(I used MOLCASMEM=2048 in every calculations.)


=== (The error message I got) ===
MA error: MA_free_heap: invalid guard(s) for memhandle 6 (name: 'CCSORT')
MA failed to release the memory block for further use.

Memory allocation map
---------------------

FldNam VarTyp ID Length iPos
- - - - - - - - - - - - - - - - - - - - - - - - -
IP_DUM REAL 0 1 43408578
IP_IDUM INTE 1 1 43408591
FIRAS REAL 2 1048576 43408604
FI REAL 3 1048576 44457192
FOKA REAL 4 524800 45505780
FOKB REAL 5 524800 46030592
CCSORT REAL 6 6283899 46555404
- - - - - - - - - - - - - - - - - - - - - - - - -
The parameters of the last call to GetMem were:
Name =CCSORT
FldNam=CCSORT
VarTyp=REAL
Key =FREE
Length= 6283899
iPos = 46555404
real 35.10
user 4.89
sys 9.32
--- Stop Module: ccsdt at Fri May 23 06:10:46 2008 /rc= 143 ---
--- Stop Module: ccsdt at Fri May 23 06:10:46 2008 /rc= 143 ---
=== (End of the error) ===


From the above message, I could understand the problem was related with memory release process, but I don't have any idea to avoid this kind of problem on MOLCAS7sp1.
Interestingly, with less symmetry, like C2v instead higher D2h, with exactly same molecular structure, I could perform CCSD(T) calculation without any problem.
Can the problem be avoided with higher symmetry? If someone knows the way, could you please let me know?

Yours sincerely,
Hirotoshi MORI
---
Hirotoshi MORI, Ph.D.
Ocha-dai Academic Production
Ochanomizu University
2-1-1 Ohtsuka, Bunkyo-ku, Tokyo 112-8610, Japan

E-mail: qc_forest@mac.com
mori.hirotoshi@ocha.ac.jp
Tel: +81-3-5978-5068




Follow Ups:



Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:


[ Follow Ups ] [ Post Followup ] [ Molcas user's WWWBoard ]