Re: CCL:Computation Resource Requirement.
- From: Kadir Diri <kadir~at~visual1.chem.pitt.edu>
- Subject: Re: CCL:Computation Resource Requirement.
- Date: Wed, 08 Dec 2004 14:06:21 -0500
Hi! What you can do to get an idea about the timings on a single cpu is
the following: do a smaller calculation (smaller basis set and/or
smaller molecule). Then scale the timings accordingly for larger basis
sets and more atoms. In Hartree-Fock, the 2 electron integrals will
scale as N^4 but the integral transformation will scale as N^5.
As to the case with multiple CPU`s, it depends on many factors, for
example on how good the code that you use is. Also, the speed will vary
from one architecture to another. It is also important
how the code
works, i.e. whether it uses shared memory on a multiple
cpu machine or
whether it runs across many nodes. There are probably many other
factors, like the speed and the configuration of the network, so it is
not very easy to answer that part of your question, but you will be
to figure out the answer after a few test calculations.
Pijush Ghosh wrote:
being new to the field of computational quantum mechanics, i
have a basic question on the requirement of the resources for this
type of computation. Can anybody cite any reference where i can get
some relation between number of processors and time required with
that of number of atoms in a system , basis set used, etc.
In my system of Hydroxyapatite- polymer I have the following details:
Basis set - 6-31+G**
Method - HF (Hartree Fock)
can somebody give idea on processor-hour required. All my
simulations are run on NCSA machines.
All suggesions and information appreciated in advance.
-= This is automatically added to each message by the mailing script =-
To send e-mail to subscribers of CCL put the string CCL: on your
and send your message to: CHEMISTRY~at~ccl.net
Send your subscription/unsubscription requests to:
CHEMISTRY-REQUEST~at~ccl.net HOME Page: http://www.ccl.net | Jobs Page:
If your mail is bouncing from CCL.NET domain send it to the maintainer:
Jan Labanowski, jlabanow~at~nd.edu (read about it on CCL Home Page)