Hi Jason,
I’d like to follow up on this… Is there a way in post-processing to recalculate the pressure?
I set the pressure with the Berendsen barostat first before switching to MC, so I imagine that it’s where I’d like it to be on average (I realize it fluctuates a lot), but it’d still be nice to check that it hasn’t diverged too much due to unforeseen events…
Is it not possible to asynchronously compute the virial for the energy output using pmemd.cuda maybe by throwing it to the CPU similar to writing files? This way the simulation can proceed, keeping it’s efficient speed, yet outputting this property similar to the Berendsen barostat… I guess this would also be a good question for Ross Walker since it involves the GPU version of pmemd.
Best,
Jesper
> On May 31, 2014, at 8:51 AM, Jason Swails <jason.swails.gmail.com> wrote:
>
>
> O
>
> n Sat, May 31, 2014 at 11:10 AM, Asaminew Haile <ashaethio.gmail.com> wrote:
>
>>
>> Dear Amber Users
>> I’m running MD with pmemd.cuda with barostat=2 option for pressure
>> control, but in my energy output pressure is
>>
>> printed out as 0.0 as shown bellow
>>
>>> NSTEP = 100000 TIME(PS) = 12670.000 TEMP(K) = 302.51 PRESS =
>> 0.0
>>
>>>
>>
>>> NSTEP = 200000 TIME(PS) = 12870.000 TEMP(K) = 299.32 PRESS =
>> 0.0
>>
>>>
>>
>>> NSTEP = 300000 TIME(PS) = 13070.000 TEMP(K) = 301.04 PRESS =
>> 0.0
>>
>>>
>>
>>> NSTEP = 400000 TIME(PS) = 13270.000 TEMP(K) = 301.09 PRESS =
>> 0.0
>>
>>>
>>
>>> and also damped the following kind of warning information into the output
>>
>>> | Attempting MC barostat change: Failed
>>
>>>
>>
>>> | Attempting MC barostat change: Succeeded
>>
>>>
>>
>>> | Attempting MC barostat change: Failed
>>
>>>
>>
>>> | Attempting MC barostat change: Failed
>>
>>>
>>
>>> | Attempting MC barostat change: Succeeded
>>
>>>
>>
>>> | Attempting MC barostat change: Failed
>>
>>>
>>
>>> | Attempting MC barostat change: Failed
>>
>>>
>>
>>> | Attempting MC barostat change: Succeeded
>>
>>>
>>
>>> Any idea
>>
>
> This is because pmemd.cuda does not compute the virial when using the MC
> Barostat. Computing the virial is expensive (results in ~10-20%
> performance reduction on GPUs compared to _not_ computing the virial). One
> of the great things about the MC barostat is that the virial is not needed
> (which is why ntp=1 or 2 with barostat=2 gives approximately the same
> performance as NVT, whereas barostat=1 is significantly slower).
>
> The MC barostat in sander _does_ compute the virial and report the pressure
> if those values are important to you.
>
> Hope this helps,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> 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 Wed Nov 05 2014 - 10:30:02 PST
