Re: No Subject Given By The Author
On Sun, 17 Dec 2000, Sergey Noskov wrote:
> I have a big problems with memory allocation in CHARMM27b1 bothly on SGI
> and GNU platfoorms. Unfortunately, I need to make an aproximate normal
> mode analysis for medium sized protein ( 5362 atom in all hydrogen force
> field). Options like REDU CMP or DIMB diagonalization does not help in
> this program. CHARMM error output include a message like :
>
> expanded heap on ***** words
> problem with storage
> MALLOC set to 0.0
This indicates that CHARMM asked for some extra memory, which the OS
could not provide. Since normal modes require a 3N*3N matrix, the
memory needed in Mbytes is
(8*(3*N)**2)/(1024*1024) for N=5362 = 1974
or almost 2 Gbytes of memory. Your system (IRIX or Linux) needs to have
a fairly large amount of physical RAM and more that 2 Gbyte of swap
space configured in order to do this calculation.
--
Rick Venable =====\ |=| "Eschew Obfuscation"
FDA/CBER Biophysics Lab |____/ |=|
Bethesda, MD U.S.A. | \ / |=| ( Not an official statement or
Rick_Venable;at;nih.gov | \ / |=| position of the FDA; for
that,
rvenable;at;speakeasy.org \/ |=| see http://www.fda.gov )