Dear Professors,
Now I want to compile vasp.4.6 on my 8 cpu workstation. I have download MPich.1.2.7. Now I want to known whether the version mpi is enough. Thanks!
complication on workstation
Moderators: Global Moderator, Moderator
complication on workstation
Last edited by lanjh on Fri Nov 16, 2007 4:23 am, edited 1 time in total.
-
- Administrator
- Posts: 2921
- Joined: Tue Aug 03, 2004 8:18 am
- License Nr.: 458
complication on workstation
yes
Last edited by admin on Fri Nov 16, 2007 2:35 pm, edited 1 time in total.
complication on workstation
Professor, I have posted in other place about the problem I am confronted with. My machine is a four-core two-pathway intel Xeon 5300 professors. There are in all 8 cpu. I have installed mpich1.2.5.2, but wrong results have be generated.
I have used -opt=-O3 also, the problem still occur.
With the lam-mpi, the same problem also appear. I guess that the compilation problem of VASP lead to this large error. Because there is no makefile suit for my machine, so I compile it with the makefile.linux_ifc_P4. Though I obtained the compiled executable vasp, but the notes like "ifort don't support -tp " still occur.
May be these notes don't affect the result obtained in serial version vasp. but generate wrong result in parallel calculation, am I right.
I don't know whether you have test vasp with this type of workstation. If not, I hope you can tell me how to configure on this platform~~~
Thanks
> entering main loop
> N E dE d eps
> ncg rms rms(c)
> DAV: 1 -0.200798705400E+02 -0.20080E+02 -0.44239E-02
> 160 0.157E+00BRMIX: very serious problems
> the old and the new charge density differ
> old charge density: 8.05531 new 8.03549
> 0.168E+00
> DAV: 2 -0.200634400081E+02 0.16431E-01 -0.82681E-04
> 200 0.294E-01BRMIX: very serious problems
> the old and the new charge density differ
> old charge density: 8.05098 new 8.03550
<span class='smallblacktext'>[ Edited Mon Nov 26 2007, 01:47AM ]</span>
I have used -opt=-O3 also, the problem still occur.
With the lam-mpi, the same problem also appear. I guess that the compilation problem of VASP lead to this large error. Because there is no makefile suit for my machine, so I compile it with the makefile.linux_ifc_P4. Though I obtained the compiled executable vasp, but the notes like "ifort don't support -tp " still occur.
May be these notes don't affect the result obtained in serial version vasp. but generate wrong result in parallel calculation, am I right.
I don't know whether you have test vasp with this type of workstation. If not, I hope you can tell me how to configure on this platform~~~
Thanks
> entering main loop
> N E dE d eps
> ncg rms rms(c)
> DAV: 1 -0.200798705400E+02 -0.20080E+02 -0.44239E-02
> 160 0.157E+00BRMIX: very serious problems
> the old and the new charge density differ
> old charge density: 8.05531 new 8.03549
> 0.168E+00
> DAV: 2 -0.200634400081E+02 0.16431E-01 -0.82681E-04
> 200 0.294E-01BRMIX: very serious problems
> the old and the new charge density differ
> old charge density: 8.05098 new 8.03550
<span class='smallblacktext'>[ Edited Mon Nov 26 2007, 01:47AM ]</span>
Last edited by lanjh on Mon Nov 26, 2007 12:44 am, edited 1 time in total.
-
- Administrator
- Posts: 2921
- Joined: Tue Aug 03, 2004 8:18 am
- License Nr.: 458
complication on workstation
please try if it helps to add a line
export OMP_NUM_THREADS=1
in your .bashrc file to suppress any additional parallelization which is not encoded in vasp itself.
export OMP_NUM_THREADS=1
in your .bashrc file to suppress any additional parallelization which is not encoded in vasp itself.
Last edited by admin on Tue Nov 27, 2007 11:38 am, edited 1 time in total.