RE: [gmx-users] parallelization error? gromacs-4.0.2

2008-11-20 Thread Berk Hess

Hi,

Do you have anisotropic pressure coupling turned on?

Could you send me the tpr file?

Berk

Date: Thu, 20 Nov 2008 14:47:53 +
From: [EMAIL PROTECTED]
To: gmx-users@gromacs.org
Subject: [gmx-users] parallelization error? gromacs-4.0.2

Hello,

I tried from the beginning to test gromacs-4.0.2 with a monoclinic system on 8 
processors (one two quad core machine). The skew errors seem to be gone, yet 
other errors appeared.

Now after a successful md, taking the output and trying to do annealing I get 
the following error:

Fatal error:
Domain decomposition has not been implemented for box vectors that have 
non-zero components in directions that do not use dom
ain decomposition: ncells = 2 1 4, box vector[3] = -1.070924 -0.07 6.415503

In my input files I do not have a vector as the bolded one above.

I tried to run the same input on my server which has two processors and up to 
know that I'm writing this email the run is running without any errors at all.

I have compiled gromacs-4.0.2 with fftw-3.2. Any files needed or any other 
information will be at
 your disposal.

Thank you,
Nikos



_
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/search before posting!
Please don't post (un)subscribe requests to the list. Use the 
www interface or send it to [EMAIL PROTECTED]
Can't post? Read http://www.gromacs.org/mailing_lists/users.php

RE: [gmx-users] parallelization error? gromacs-4.0.2

2008-11-20 Thread Berk Hess

Hi,

Ah, so you have anistropic pressure coupling on.
I forgot to put in a check for that when chosing the domain
decomposition grid.

Changing line 200 of src/mdlib/domdec_setup.c from:
if (box[j][i] != 0)
to
if (box[j][i] != 0 || ir-deform[j][i] != 0 ||
(ir-epc != epcNO  ir-compress[j][i] != 0))
should fix the problem.
If will commit this fix for 4.0.3.

(PS: you can also set -dd nx ny nz by hand, where for anisotropic
pressure coupling you should not have nx=1 or ny=1 when ny!=1 or nz!=1)

Berk

From: [EMAIL PROTECTED]
To: gmx-users@gromacs.org
Subject: RE: [gmx-users] parallelization error? gromacs-4.0.2
Date: Thu, 20 Nov 2008 22:05:52 +0100








Hi,

Do you have anisotropic pressure coupling turned on?

Could you send me the tpr file?

Berk

Date: Thu, 20 Nov 2008 14:47:53 +
From: [EMAIL PROTECTED]
To: gmx-users@gromacs.org
Subject: [gmx-users] parallelization error? gromacs-4.0.2

Hello,

I tried from the beginning to test gromacs-4.0.2 with a monoclinic system on 8 
processors (one two quad core machine). The skew errors seem to be gone, yet 
other errors appeared.

Now after a successful md, taking the output and trying to do annealing I get 
the following error:

Fatal error:
Domain decomposition has not been implemented for box vectors that have 
non-zero components in directions that do not use dom
ain decomposition: ncells = 2 1 4, box vector[3] = -1.070924 -0.07 6.415503

In my input files I do not have a vector as the bolded one above.

I tried to run the same input on my server which has two processors and up to 
know that I'm writing this email the run is running without any errors at all.

I have compiled gromacs-4.0.2 with fftw-3.2. Any files needed or any other 
information will be at
 your disposal.

Thank you,
Nikos



Express yourself instantly with MSN Messenger! MSN Messenger
_
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/search before posting!
Please don't post (un)subscribe requests to the list. Use the 
www interface or send it to [EMAIL PROTECTED]
Can't post? Read http://www.gromacs.org/mailing_lists/users.php