Thank you for the reply.

My questions are the following:

1- What do you meaning of mixing parameters?

2- Also What do you mean compiling with slow optimisations? how to faster them?

3- When are  we sure we can use parallel or serial? In other words what made 
you almost sure that we dont need mpi?

Thank you again and looking again to your reply.

EL-abed


 El-abed Haidar | Doctor of Philosophy (Science)
 Condensed Matter Theory (CMT) Group
 | School of Physics
 THE UNIVERSITY OF SYDNEY  | NSW | 2006

________________________________
From: [email protected] <[email protected]> on behalf of Nick 
Papior <[email protected]>
Sent: Thursday, 28 June 2018 4:42:42 PM
To: [email protected]
Subject: Re: [SIESTA-L] Calculating cohesive energy in the fastest but most 
efficient manner?

There could be several reasons:
1) if the number of SCF iterations is very large (say > 200) then your input 
parameters are bad, then check for mixing parameters etc.
2) You have compiled siesta with slow optimizations.
3) You probably don't need MPI for this small simulation, it may yield a worse 
performance.
4) You have a slow computer.

Den ons. 27. jun. 2018 kl. 22.11 skrev El-abed Haidar 
<[email protected]<mailto:[email protected]>>:

Hello again,

I am calculating the cohesive energy by finding the energy of a single atom. In 
my code, I am interested in calculating the cohesive energy for different 
cutoff energies (b in the code) while finding the optimised lattice constant (a 
in code).

The reason the code at the moment is only one value of cut off is that for only 
one run, the time taken is around one full day. I am not expecting that at all.

The lattice vectors are clearly larger than 10 A while using only one K point. 
Based on my parameters what is slowing down my code that much?

Thank you and looking forward to your reply.

EL-Abed


 El-abed Haidar | Doctor of Philosophy (Science)
 Condensed Matter Theory (CMT) Group
 | School of Physics
 THE UNIVERSITY OF SYDNEY  | NSW | 2006


--
Kind regards Nick

Responder a