Dear Vinícius,
I have checked that the code is using 4 different GPUs, one for each
pH-REMD replica. Please, find attached below the information shown by the
mdout file.
|------------------- GPU DEVICE INFO --------------------
|
| Task ID: 0
| CUDA_VISIBLE_DEVICES: 0,1,2,3
| CUDA Capable Devices Detected: 4
| CUDA Device ID in use: 0
| CUDA Device Name: Tesla V100-PCIE-16GB
| CUDA Device Global Mem Size: 16160 MB
| CUDA Device Num Multiprocessors: 80
| CUDA Device Core Freq: 1.38 GHz
|
|--------------------------------------------------------
|---------------- GPU PEER TO PEER INFO -----------------
|
| Peer to Peer support: ENABLED
|
|--------------------------------------------------------
| Conditional Compilation Defines Used:
| MPI
| PUBFFT
| BINTRAJ
| CUDA
| EMIL
Bests,
Laura
On Fri, Feb 25, 2022 at 10:55 AM Laura Milena Pedraza-González <
la.pedrazagonzalez.gmail.com> wrote:
> Dear Vinícius,
>
> In fact, I am using a machine with 8 GPUs (Tesla V100-PCIE-16GB). Do you
> know if any specific setup on the GPUs is needed before running pH-REMD?
>
> Thanks and regards,
>
> Laura
>
> On Fri, Feb 25, 2022 at 12:18 AM Vinicius Wilian Cruzeiro <
> vwcruz.stanford.edu> wrote:
>
>> Hello Laura,
>>
>> Are you running this pH-REMD test in a machine with 4 GPUs? REMD requires
>> one GPU per replica.
>>
>> Best,
>>
>> Vinícius Wilian D. Cruzeiro, PhD
>> Postdoctoral Researcher
>> Department of Chemistry
>> Stanford University
>> Twitter: .vwcruzeiro
>> ________________________________
>> From: Laura Milena Pedraza-González <la.pedrazagonzalez.gmail.com>
>> Sent: Thursday, February 24, 2022 6:33 AM
>> To: amber.ambermd.org <amber.ambermd.org>
>> Subject: [AMBER] pH replica exchange MD (pH-REMD) in explicit solvent,
>> using CUDA MPI in AMBER 18
>>
>> Dear Amber users,
>>
>> I am interested in doing constant pH replica exchange MD (pH-REMD) in
>> explicit solvent, using CUDA MPI in AMBER 18. As far as I know, such
>> methodology has been implemented and recently employed (e.g., J. Phys.
>> Chem. B 2021, 125, 9168−9185). In this regard, I ran the test
>> amber18/test/cnstph_remd/Explicit_pHREM/, that works correctly when using
>> both sander.MPI and pmemd.MPI. However, when using pmemd.cuda.MPI (i.e.,
>> mpirun -np 4 pmemd.cuda.MPI -ng 4 -groupfile groupfile -rem 4 -remlog
>> rem.log), the calculation crashes after the first protonation state change
>> attempt (e.g., at NSTEP 4 with ntcnstph=5), without printing any error
>> different to the standard:
>>
>> ".Primary job terminated normally, but 1 process returned
>>
>> a non-zero exit code. Per user-direction, the job has been aborted.
>> --------------------------------------------------------------------------
>> --------------------------------------------------------------------------
>> mpirun detected that one or more processes exited with non-zero status,
>> thus causing
>> the job to be terminated. The first process to do so was:
>>
>> Process name: [[42377,1],0]
>> Exit code: 174"
>>
>> On the other hand, I tested my pmemd.cuda.MPI AMBER installation by
>> running
>> the test of temperatura REMD (
>> https://ambermd.org/tutorials/advanced/tutorial7/index.php) and it worked
>> smoothly. Moreover, I have some experience computing normal CpHMD with
>> pmemd.cuda and the methodology works well in my installation.
>> Do you have any idea about which would be the problem associated with
>> pmemd.cuda.MPI in pH-REMD?
>>
>> Thanks in advance for your help.
>>
>> Best regards,
>>
>> Laura
>>
>> --
>> *Dr. Laura Pedraza-González*
>> Postdoctoral Researcher
>> MoLECoLab (Modeling Light & Environment in Complex Systems)
>> Department of Chemistry and Industrial Chemistry
>> University of Pisa
>> Via Moruzzi 13, 56124, Pisa, ITALY
>> Tel. +39-331-2760784
>>
>> https://molecolab.dcci.unipi.it/people/laura-pedraza.html
>> _______________________________________________
>> 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
>>
>
>
> --
> *Dr. Laura Pedraza-González*
> Postdoctoral Researcher
> MoLECoLab (Modeling Light & Environment in Complex Systems)
> Department of Chemistry and Industrial Chemistry
> University of Pisa
> Via Moruzzi 13, 56124, Pisa, ITALY
> Tel. +39-331-2760784
>
> https://molecolab.dcci.unipi.it/people/laura-pedraza.html
>
--
*Dr. Laura Pedraza-González*
Postdoctoral Researcher
MoLECoLab (Modeling Light & Environment in Complex Systems)
Department of Chemistry and Industrial Chemistry
University of Pisa
Via Moruzzi 13, 56124, Pisa, ITALY
Tel. +39-331-2760784
https://molecolab.dcci.unipi.it/people/laura-pedraza.html
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 25 2022 - 04:00:02 PST
