AMBER Archive (2004)

Subject: Re: AMBER: Normal modes of protein-ligand

From: Martin Lepsik (lepsik_at_minea.uochb.cas.cz)
Date: Thu Jan 29 2004 - 03:47:22 CST

Dear Holger,
thanks for your answers. I am still having problems, though, with
calculating vibrational entropies using nmode via mm_pbsa of 3000-atom
protein.

The problem is with the size of memory arrays (non-bonded pairs). How
can I make them bigger? In compilation (where?) or using more memory?

I'm sending you parts of nmode files, sizes.h and sizes2.h if you please
could have a look and suggest parameters for my case. (I am especially
not sure about nprec=2 and naxdup=800)

If it works, I prefer 64-bit Opterons of 4/8GB memory over Itanium2
Altix3000 which is busy all the time.

Looking forward to hearing from you. Thanks.

Best regards,

  Martin Lepsik

----------
nmode.out

Total memory required : 141798148 real words

Total memory avail : 265000000 real words

Total memory required : 213413 integer words

Total memory avail : 4000000 integer words

Maximum nonbond pairs 3786586

  Getting coordinates from file with title:

     Number of non-bonded pairs = 3812701
     Number of H-bonded pairs = 0
       Insufficient memory to hold
       nonbonded pair list
       Memory available for list = 3786586

           ***** ERROR STOP *****
-------------
sizes.h

c parameter (MAXMEMX=3500000)
       parameter (MAXMEMX=190000000)

       parameter (MAXMEMI=4000000)
c
c ----- More precisely,
c ----- bit length of an actual real array element
c ----- = nprec * (declared bit length of x)
c ----- provided it is allowed on the machine in use.
c
c ----- nshort should be 1 if all integers use the same
c ----- number of bits and 2 if some short integers use
c ----- half the number of bits needed for a normal integer.
c
c ML:changed to increase array size
       parameter (nprec = 2)
c parameter (nprec = 1)
       parameter (nshort = 1)
c
c ----- naxdup is the maximum number of dihedrals that can ever
c ----- be duplicated. 400 seems to be a good conserative number,
c ----- unless you have a really big molecule
c
c parameter (naxdup = 400)
c
c ML 2*
       parameter (naxdup = 800)
       real*8 x (MAXMEMX)
       integer ix(MAXMEMI)
       common/store_x/ x
       common/store_ix/ ix
-------------
sizes2.h

c parameter (MEMDRV=10000000)
c parameter (MAXATOM=2600)
c parameter (MAXINT=200)
c parameter (MAXVEC=500)
c ML dame to 5*vetsi
       parameter (MEMDRV=50000000)
       parameter (MAXATOM=13000)
       parameter (MAXINT=1000)
       parameter (MAXVEC=2500)
c
C ----- SET THE LIMITS OF SOME ARRAY BOUNDS -----
C
c parameter (MAXDIH = 5500)
c parameter (MAXDIA = 3500)
c parameter (MAXINB = 9000)
c parameter (MAXBON = 2600)
c parameter (MAXBNH = 2000)
c parameter (MAXANG = 3000)
c parameter (MAXANH = 3000)
c ML dame to 5*vetsi
       parameter (MAXDIH = 27500)
       parameter (MAXDIA = 17500)
       parameter (MAXINB = 45000)
       parameter (MAXBON = 13000)
       parameter (MAXBNH = 10000)
       parameter (MAXANG = 15000)
       parameter (MAXANH = 15000)
--------------------------
-----------
Martin Lepsik
Dept. of Molecular Modeling
Institute of Organic Chemistry & Biochemistry
Czech Academy of Sciences
phone: +420-220 183 540
fax:+420-220 183 292
e-mail:lepsik_at_uochb.cas.cz

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu