Hi Oz,
Can you verify that it did indeed update properly - as in you now see
version 12.3.1 in your mdout file.
Also can you try the GNU compilers please. We are investigating issues
with a potential race condition triggered by the intel compilers right now
and they make no difference to the performance of the GPU code anyway. I'd
also suggest trying just the single GPU (if you've got 4 GPUs in a node
it's better to run 4 runs from different random seeds and get more
sampling) than it is to run a single 4 GPU job.
All the best
Ross
On 11/21/13 5:29 PM, "O Z" <roza2006.hotmail.fr> wrote:
>Hi,
>
>The update to Amber12.21 does not work. I tried it in different ways but
>the errors are always the same (please find attached two examples of the
>scripts I tried and the error message I got).
>I contacted the people in charge of the supercomputer we use, they did
>the update as follow:
>
>"
># ./configure -cuda -mpi -intel
>
>
>
>
>
># make install
>
>
>
>
>
>
>
>
>
>
>
>With intel/13.1 and cuda/5.5.22 loaded. And it installs just fine.
>"
>
>I even tried using:
>
>module load cuda/5.5.22
>module load intel/13.1
>
>instead of:
>
>module load cuda/4.2.9
>module load intel/12.1 (as in the two scripts attached)
>
>But it gives the same error.
>Thanks in advance for your help.
>
>Regards.
>OZ
>
>
>
>> Date: Thu, 14 Nov 2013 08:04:00 -0700
>> Subject: Re: [AMBER] Temperature raises abnormally during accelerated MD
>> From: daniel.r.roe.gmail.com
>> To: roza2006.hotmail.fr
>>
>> Hi,
>>
>> On Wed, Nov 13, 2013 at 11:24 PM, O Z <roza2006.hotmail.fr> wrote:
>> > The Amber update did not make any change. I decided to run my
>>simulations
>> > without using MPI (pmemd.cuda_SPDP instead of pmemd.cuda_SPDP.MPI),
>>which
>> > works the temperature is stable.
>>
>> Are you sure you are running with the updated code? This line:
>>
>> > module load amber/12.9
>>
>> Makes me think that you're running amber only up to bugfix 9. The
>> current bugfix level is 21, and you need at least up to bugfix 19 to
>> have the most recent CUDA code. Your code is not up to date unless you
>> are seeing this in your MDOUT:
>>
>> |--------------------- INFORMATION ----------------------
>> | GPU (CUDA) Version of PMEMD in use: NVIDIA GPU IN USE.
>> | Version 12.3.1
>>
>> -Dan
>>
>> >
>> > mpirun -1sided -lsf -np 4 pmemd.cuda_SPDP -O -o MutPC_aMD_1-1_bis.out
>>-i
>> > mdrun_RandSeed.in -p Mut_Wat.parm7 -c MutPC_aMD_1-1.rst -ref
>> > MutPC_aMD_1-1.rst -r MutPC_aMD_1-1_bis.rst -x MutPC_aMD_1-1_bis.mdcrd
>>-amd
>> > MutPC_aMD_1-1_bis.log
>> >
>> >
>> >
>> >
>> >
>> >
>> >> Date: Tue, 5 Nov 2013 08:19:26 -0700
>> >
>> >> From: daniel.r.roe.gmail.com
>> >> To: amber.ambermd.org
>> >> Subject: Re: [AMBER] Temperature raises abnormally during
>>accelerated MD
>> >>
>> >> Hi,
>> >>
>> >> I received your output files and I note that your problems appear to
>> >> start when you start using pmemd.cuda.MPI with 2 nodes
>> >> (WTPC_aMD_1-2.out), and that you are running an old version of the
>> >> CUDA code (12.1), which is over a year out of date. There have been
>> >> major improvements and fixes to the CUDA code since then; the current
>> >> version is 12.3.1. Try updating your Amber installation by re-running
>> >> configure and re-compiling, then restart the 'WTPC_aMD_1-2.out' run
>> >> and see if the temperature jump still occurs.
>> >>
>> >> -Dan
>> >>
>> >>
>> >> On Mon, Nov 4, 2013 at 12:41 AM, O Z <roza2006.hotmail.fr> wrote:
>> >> > Dear Amber users,
>> >> >
>> >> > I have been running Accelerated MD on my system. I set up my aMD
>> >> > parameters as explained on the Amber manual:
>> >> >
>> >> > from 10ns MD results (trajectory from 50 to 60 ns).
>> >> >
>> >> >
>> >> >
>> >> > ave Dihedral: 2095.80
>> >> >
>> >> >
>> >> > ave EPtot: -194481.05
>> >> >
>> >> >
>> >> > total atoms: 60616
>> >> >
>> >> >
>> >> > protein residues: 199 (3.5*199=696.5)
>> >> >
>> >> >
>> >> > AlphaD: 1/5 * 696.5 = 139.3
>> >> >
>> >> >
>> >> > EthreshD: 696.5 + 2095 = 2791.5
>> >> >
>> >> >
>> >> > alphaP: 60616 * (1/5) = 12123.2
>> >> >
>> >> >
>> >> > EthresP: -194481.05 + 12123.2 = -191091.1
>> >> >
>> >> > I used these parameters as follow:
>> >> >
>> >> > &cntrl
>> >> >
>> >> >
>> >> > imin=0,
>> >> >
>> >> >
>> >> > ntx=5,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > ntpr=25000, ntwx=25000, ntwe=25000, ntwprt=0,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > ntf=2, ntb=2, cut=12.0, ntp=1, pres0=1.0, comp=44.6, taup=1.0,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > iwrap=1, nscm=250,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > nstlim=25000000, dt=0.002,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > ntt=3, temp0=298.0, tautp=1.0, gamma_ln=3, ig=-1,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > ntc=2, irest=1,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > iamd=3,EthreshD=2791.5,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > alphaD=139.3,EthreshP=-191091.0549,
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > alphaP=12123.2
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > /
>> >> >
>> >> >
>> >> > I am running aMD at 298K but in the output file I noticed
>> >> > that the temperature raises a lot from 298K to about 320K (please
>>see
>> >> > file attached).
>> >> >
>> >> >
>> >> > Is that due to the boost function? Or is it a normal behaviour,
>>even if
>> >> > I use the NPT ensemble?
>> >> >
>> >> > Many thanks for your help.
>> >> > Regards.
>> >> > OZ
>> >> >
>> >> > _______________________________________________
>> >> > AMBER mailing list
>> >> > AMBER.ambermd.org
>> >> > http://lists.ambermd.org/mailman/listinfo/amber
>> >> >
>> >>
>> >>
>> >>
>> >> --
>> >> -------------------------
>> >> Daniel R. Roe, PhD
>> >> Department of Medicinal Chemistry
>> >> University of Utah
>> >> 30 South 2000 East, Room 201
>> >> Salt Lake City, UT 84112-5820
>> >> http://home.chpc.utah.edu/~cheatham/
>> >> (801) 587-9652
>> >> (801) 585-6208 (Fax)
>> >>
>> >> _______________________________________________
>> >> AMBER mailing list
>> >> AMBER.ambermd.org
>> >> http://lists.ambermd.org/mailman/listinfo/amber
>>
>>
>>
>> --
>> -------------------------
>> Daniel R. Roe, PhD
>> Department of Medicinal Chemistry
>> University of Utah
>> 30 South 2000 East, Room 201
>> Salt Lake City, UT 84112-5820
>> http://home.chpc.utah.edu/~cheatham/
>> (801) 587-9652
>> (801) 585-6208 (Fax)
> _______________________________________________
>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 Nov 21 2013 - 18:30:03 PST
