[gmx-users] blue gene running error

[Mark Abraham]

On 28/04/2012 4:14 PM, Albert wrote:
> hello Mar:
>   thanks a lot for kind reply.
>   From the link you you mentioned, it seems that this problem comes 
> from the MD system itself. However, it goes well in my workstation. 
> Moreover, I visualized and analyzed the results from my workstation 
> running, everything goes well. I don't find any problem with it.
>
> But I don't know why it doesn't work in the blue gene computer.

Sometimes a marginally stable calculation will work or not based on 
apparently insignificant things, like whether it's in parallel or not. 
Numerical integration can be messy. If you prepare the system slightly 
better and/or equilibrate more gently, then it will run under BlueGene, 
from your evidence of what I assume to be "lucky" good behaviour on your 
workstation.

Mark

>
> THX
>
> ALbert
>
>
>
> On 04/28/2012 07:36 AM, Mark Abraham wrote:
>> On 28/04/2012 2:04 PM, Albert wrote:
>>> hello:
>>>   I am running NPT on a blue gene cluster, but the jobs always 
>>> failed with following messages. However, everything goes well if I 
>>> run it on my local cluster:
>>
>> Systems with marginally stable initial conditions can do this. See 
>> http://www.gromacs.org/Documentation/Errors#A_charge_group_moved_too_far_between_two_domain_decomposition_steps.
>>
>> Mark
>>
>>
>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120428/506ee007/attachment.html>