Reversing bugfix.patch9 results in the removal of all files and
folders under $AMBERHOME/test/cuda/nmropt/pme .
I have given the command :
./patch_amber.py --reverse-patch=.patches/Amber_Applied_Patches/bugfix.9
The result is:
Error or warning during patching process for
.patches/Amber_Applied_Patches/bugfix.9:
The next patch, when reversed, would delete the file
test/cuda/nmropt/pme/distance_COM/distCOM_pbc.RST,
which does not exist! Skipping patch.
1 out of 1 hunk ignored
Unreversed patch detected! Skipping patch.
1 out of 1 hunk ignored -- saving rejects to file
test/nmropt/pme/distance_COM/Run.distCOM_pbc.rej
But actually I was kind of hoping that I would be able to run the
simulation using sander or pmemd. I am not expecting to be able to run
cuda accelerated now. So I have done a fresh install of amber12,
patched it again and tried everything all over again. The fact is that
I cannot even run the COM dissociation of my system with sander. The
mdout file ends with:
5. REFERENCE ATOM COORDINATES
defa
----- READING GROUP 1; TITLE:
/
rfree: End of file on unit 5
Here is my mdin file:
# Distance_COM of chains A and B
&cntrl
imin = 0,
irest = 1,
ntx = 5,
ntb = 2, pres0 = 1.0, ntp = 3, csurften = 3, gamma_ten = 17,
ninterface = 2 taup = 2.0,
cut = 8.0,
ntr = 1,
ntc = 2,
ntf = 2,
temp0 = 310.0,
ntt = 3,
gamma_ln = 1.0,
nstlim = 5000000, dt = 0.002,
ntpr = 1000, ntwx = 1000, ntwr = 10000,
ig = 71277, ioutfm = 1, iwrap = 1
nmropt=1,
/
&wt type='DUMPFREQ', istep1=2 /
&wt type='END' /
DISANG=dist.RST
DUMPAVE=distCOM_pbc_vs_t
LISTIN=POUT
LISTOUT=POUT
/
And my dist.RST file:
# Distance restraint for chain A and B (center of mass)
&rst iat=-1,-1, r1=0., r2=11., r3=11., r4=35., rk2 = 30., rk3 = 30.,
igr1=1,10,15,25,33, igr2=34,43,48,58,66, /
&end
I have tried to do as you told me to. I took the example .RST file
above given by Mo Chen and modified to fit my needs. The mdin file was
also taken from the test examples and modified.
The command line I gave was:
sander -O -i mdin -p r2r2_lipid.prmtop -c restrt.npt_rest14 -ref
restrt.npt_rest14 -o mdout.11.md -inf mdinfo.11.md -x mdcrd.11.md -r
restrt.11.md
as for pmemd:
pmemd -O -i mdin -c restrt.npt_rest14 -p r2r2_lipid.prmtop -ref
restrt.npt_rest14 -suffix 11.md
And the error is:
At line 757 of file file_io.F90 (unit = 5, file = 'mdin')
Fortran runtime error: End of file
Since Unit 5 on the mdout file refers to the reading of the reference
file, I have tried to change the reference file. The result is the
same. When I try to run the command without the -ref flag, I get an
error message saying that the ref file is missing. I have looked at
the examples given in test/nmropt/pme/distance_COM/ or
test/cuda/nmropt/pme/distance_COM/, and from the mdout file, I could
not spot the use of a refc file. It seems that the command given by
Run.distCOM_pbc does not specify a reference file. Is that correct? So
how come I am asked for one?
Thank you
Fabrício
2012/9/19 Jason Swails <jason.swails.gmail.com>:
> On Wed, Sep 19, 2012 at 1:26 AM, Fabrício Bracht <bracht.iq.ufrj.br> wrote:
>
>> Hello.
>> Before modifying my files, I decided to see if I could run the test
>> simulations located at
>> $AMBERHOME/test/cuda/nmropt/pme/jar_distance_COM/ . I ran the command
>> " ./Run.jar_pbc " and the simulation failed. The error message in the
>> mdout file was:
>>
>> CUDA (GPU): Implementation does not currently support the use of
>> Jarzynski simulations.
>> Require jar == 0 (default).
>>
>> Input errors occurred. Terminating execution.
>>
>> The example from $AMBERHOME/test/cuda/nmropt/pme/distance_COM/ also
>> fails. The error message is:
>>
>> CUDA (GPU): Implementation does not currently support the use of COM
>> simulations.
>> Require iat(2) >= 0
>>
>> I have already updated amber12 and ./patch_amber.py says that there
>> are no updates to apply. Am I doing something wrong?
>>
>
> In addition to adding support for Kepler, bugfix.9 also accelerated the
> NMROPT calculations. However, it did not accelerate all of them due to
> time constraints, so some functionality (like COM distances and steered MD)
> was actually removed by bugfix.9.
>
> You can always reverse bugfix.9 using patch_amber.py if you *need*
> CUDA-accelerated COM and/or steered MD simulations, but I would suggest
> against it. (Keep in mind that you would lose Kepler support doing this).
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
> _______________________________________________
> 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 Sep 19 2012 - 10:30:03 PDT
