The skinnb errors are exactly the same as reported by many others here. No point reproducing it again.
Mine is a pmemd.cuda.MPI simulation across 8 GPUs. Yes it is related to the box size which changes drastically if small taup values are used. As you increase the taup value, the box size becomes stable, and skinnb error goes away. Either that or you restart the simulation from where it stopped, and depending on the taup value used, it either stops again or goes to completion.
My experience indicated that taup had a drastic effect for NPT and NPgT simulations which run into skinnb errors, and by adjusting this variable alone, I was able to 'cure' the skinnb error. I just wanted to know if this approach is right or wrong. Another way is to increase 'cut' to 14 instead of 10. However this incurs a severe computational penalty.
I set gamma_ten ZERO based on this paper - Fritz Jahnig, Biophysical Journal Vol 71, 1348-1349, Sep 1996. 'What is the surface tension of a lipid bilayer membrane?' Any comments ? I could change this back to 15 and rerun the simulations.
Thanks
On 26 Sep, 2013, at 11:19 AM, Ross Walker <ross.rosswalker.co.uk> wrote:
> What are the skinnb errors you are seeing? - Is this a GPU simulation
> which is quiting? If so that is entirely related to the box size and the
> value of taup is not going to affect it - your system should not be
> oscillating in size enough to cause skinnb failure on the GPU after it has
> equilibrated. Note gamma_ten should NOT be zero for constant surface
> tension. Something on the order of 15.0 or so is probably better here.
>
>
> My guess would be that this is your problem and not the value of taup.
>
> All the best
> Ross
>
> On 9/25/13 7:46 PM, "Wang Chern Hoe (Dr)" <WangCH.ntu.edu.sg> wrote:
>
>> I am running a simulation with the following parameters
>>
>> S.
>> ntp=3, taup=20, cut=10, csurften=3, gamma_ten=0.0, ninterface=2
>> S.
>>
>>
>> I wish to understand if it is alright to increase taup to 100 or even
>> 1000 to make the simulation run without skinnb problems.
>>
>> ________________________________
>>
>> CONFIDENTIALITY:This email is intended solely for the person(s) named and
>> may be confidential and/or privileged.If you are not the intended
>> recipient,please delete it,notify us and do not copy,use,or disclose its
>> content.
>>
>> Towards A Sustainable Earth:Print Only When Necessary.Thank you.
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 26 2013 - 00:00:03 PDT
