Hi Sarah,
I believe that the jump in temperature seen at the beginning of the
simulation with ntt=2 was coming from a missing CPU to GPU copy. I believe
that it was purely cosmetic although don't have internet access right now
to look back through the bug report but my recollection was that this was
purely cosmetic in that it only effected the reported energies and not
forces on which the trajectory is propagated.
This arose because ntt=2 is rarely used with AMBER and so not extensively
tested. My suggestion though would be to rerun the previous calculations
if you can afford to. There are also some fixes for NPT they came out
after the release of AMBER 12 so it certainly wouldn't help to rerun with
the new code. At the very least rerun one of the previous simulations and
see if you see any differences in your result analysis that can't be
attributed to convergence difference in the sampling.
All the best
Ross
On 6/10/13 11:34 AM, "Sarah Graham" <sarahgra.umich.edu> wrote:
>Hi,
>
>We applied the newest bugfixes and it seems to be working fine now, thank
>you. I had already run some simulations before I noticed these
>temperature spikes. I was wondering what caused these temperature spikes
>and do you think it would be necessary to redo all the previous
>simulations
>where this occurred? There seemed to be no problems in the resulting
>trajectories that I can see, they look fine and all of the rmsd's are
>fine.
>Thanks for your help,
>
>Sarah Graham
>
>On Mon, Jun 10, 2013 at 10:12 AM, Ross Walker <ross.rosswalker.co.uk>
>wrote:
>
>> Hi Sarah,
>>
>> Can you clarify whether you see this with AMBER 12 with the latest bug
>> fixes please. That is your output file should contain the following:
>>
>> |--------------------- INFORMATION ----------------------
>> | GPU (CUDA) Version of PMEMD in use: NVIDIA GPU IN USE.
>> | Version 12.3
>> |
>> | 04/24/2013
>> |
>>
>>
>>
>> All the best
>> Ross
>>
>>
>>
>> On 6/10/13 5:26 AM, "Sarah Graham" <sarahgra.umich.edu> wrote:
>>
>> >Sorry if this is a repeat, my emails don't seem to be going through.
>> >
>> >Hi,
>> >
>> >I have been running MD simulations on a gpu and am having issues when
>> >restarting. I have ntt=2 and when restarting on gpu the first step
>>has a
>> >temperature of about 2000 instead of 300. I have irest=1 and ntx=5 so
>> >velocity information should be read from the restart file. The restart
>> >files are there and have information in them. When I run the same
>> >simulations on cpu using the exact same input files the temperature
>>does
>> >not spike like this. Has anyone seen this before or know what might be
>> >causing it? I am using Amber11 but I have also tried Amber12 and I
>>still
>> >have the same issue.
>> >
>> >Thanks for your help,
>> >
>> >Sarah Graham
>> >
>> >
>> >input file
>> >
>> >Production run
>> > &cntrl
>> > imin = 0,
>> > irest = 1, ntx = 5,
>> > ntwe = 1000, ntwx = 1000, ntwr = 1000, ntpr = 1000, ntwv = 1000,
>> > ntt = 2, temp0 = 300.0, tempi = 300.0,
>> > dielc = 1, cut = 10.0,
>> > ntb = 2, ntp = 1, ntc = 2, ntf = 2,
>> > ig = -1,
>> > nstlim = 1250000, dt = 0.0020,
>> > /
>> >
>> >example output
>> >
>> >|-----------------------------
>> >----------------------
>> >Setting new random velocities at step 1000
>> >check COM velocity, temp: 0.004194 0.03(Removed)
>> >
>> > NSTEP = 1000 TIME(PS) = 69332.000 TEMP(K) = 1860.02 PRESS =
>> >-154.1
>> > Etot = 28134.4824 EKtot = 174025.4076 EPtot =
>> >-145890.9252
>> > BOND = 696.4149 ANGLE = 1824.4578 DIHED =
>> >2475.8020
>> > 1-4 NB = 833.1366 1-4 EEL = 8244.7069 VDWAALS =
>> >17977.2913
>> > EELEC = -177942.7347 EHBOND = 0.0000 RESTRAINT =
>> >0.0000
>> > EKCMT = 12792.8104 VIRIAL = 14330.6250 VOLUME =
>> >462044.0055
>> > Density =
>> >1.0133
>> >
>>
>>>------------------------------------------------------------------------
>>>--
>> >----
>> >
>> >Setting new random velocities at step 2000
>> >check COM velocity, temp: 0.000951 0.00(Removed)
>> >
>> > NSTEP = 2000 TIME(PS) = 69334.000 TEMP(K) = 300.63 PRESS =
>> >-54.6
>> > Etot = -117857.7631 EKtot = 28126.8179 EPtot =
>> >-145984.5810
>> > BOND = 734.8103 ANGLE = 1823.5534 DIHED =
>> >2467.0220
>> > 1-4 NB = 852.7348 1-4 EEL = 8202.9194 VDWAALS =
>> >17875.8311
>> > EELEC = -177941.4521 EHBOND = 0.0000 RESTRAINT =
>> >0.0000
>> >
>> >203,2 1%
>> >_______________________________________________
>> >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
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 12 2013 - 13:30:04 PDT
