Re: [AMBER] AMBER pmemd and pmemd.cuda, not restart files after minimization - fort.16 file instead

From: Marko Wehle <Marko.Wehle.mpikg.mpg.de>
Date: Tue, 5 Feb 2013 12:07:47 +0100

Dear Dan and Ross,
the input looks like the following:

For pmemd on CPU

full system minimization
 &cntrl
  imin = 1,
  maxcyc = 10000,
  ncyc = 5000,
  ntb = 1,
  ntr = 0,
  ntp = 0,
  ntf = 1,
  ntc = 1,
  cut = 10.0
 /

For pmemd on GPU

full system minimization
 &cntrl
  imin = 1,
  maxcyc = 10000,
  ncyc = 10000,
  ntb = 1,
  ntr = 0,
  ntp = 0,
  ntf = 1,
  ntc = 1,
  cut = 10.0
 /

ncyc is here 10000, because the minimization does not work further otherwise.

Command line Options:

PROGRAM="pmemd.MPI"
#run minimization...
PROGRAM_OPTIONS="-O -i "$name"_runSanderMIN.mdin -o "$name"_runSanderMIN.mdout -p "$name"_prmTop.prmtop -c "$name"_prmCoord.crd -r "$name"_MINCoord.rst -inf "$name"_runSanderMIN.mdinfo"
mpirun -v -d -x LD_LIBRARY_PATH -x PATH `which ${PROGRAM}` ${PROGRAM_OPTIONS}




The requested informations, by Ross, are:

AMBER11 GPU:

|--------------------- INFORMATION ----------------------
| GPU (CUDA) Version of PMEMD in use: NVIDIA GPU IN USE.
| Version 2.3
|
| 12/14/2011
|
|
| Implementation by:
| Ross C. Walker (SDSC)
| Scott Le Grand (nVIDIA)
| Duncan Poole (nVIDIA)
|
| CAUTION: The CUDA code is currently experimental.
| You use it at your own risk. Be sure to
| check ALL results carefully.
|
| Precision model in use:
| [SPDP] - Hybrid Single/Double Precision (Default).
|
|--------------------------------------------------------
 
|------------------- GPU DEVICE INFO --------------------
|
| CUDA Capable Devices Detected: 1
| CUDA Device ID in use: 0
| CUDA Device Name: Tesla M2050
| CUDA Device Global Mem Size: 2687 MB
| CUDA Device Num Multiprocessors: 14
| CUDA Device Core Freq: 1.15 GHz
|
|--------------------------------------------------------


and AMBER12 GPU:

|--------------------- INFORMATION ----------------------
| GPU (CUDA) Version of PMEMD in use: NVIDIA GPU IN USE.
| Version 12.2
|
| 01/10/2013
|
| Implementation by:
| Ross C. Walker (SDSC)
| Scott Le Grand (nVIDIA)
| Duncan Poole (nVIDIA)
|
| CAUTION: The CUDA code is currently experimental.
| You use it at your own risk. Be sure to
| check ALL results carefully.
|
| Precision model in use:
| [SPFP] - Mixed Single/Double/Fixed Point Precision.
| (Default)
|
|--------------------------------------------------------
 
|------------------- GPU DEVICE INFO --------------------
|
| CUDA Capable Devices Detected: 1
| CUDA Device ID in use: 0
| CUDA Device Name: Tesla M2050
| CUDA Device Global Mem Size: 2687 MB
| CUDA Device Num Multiprocessors: 14
| CUDA Device Core Freq: 1.15 GHz
|
|--------------------------------------------------------
 

Unfortunately, those kind of informations are not available for the CPU Versions of AMBER11 and AMBER12.


Thank you a lot for your support.
Sincerely yours,
Marko



>
>Dear Amber Developers and Users,
>unfortunatly I've did not get any better behavior of Amber12 pmemd.cuda
>after minimzation steps.
>Here a list of working steps, to let you know what our protocol for
>installation is:
>
>###########################################
>
>tar xvfj Amber12.tar.bz2
>tar xvfj AmberTools12.tar.bz2
>
>cd $AMBERHOME
>./patch_amber.py --update-tree
>./patch_amber.py --update-tree
>./patch_amber.py --update-tree
>
># ./patch_amber.py --patch-level
>> Latest patch applied to AmberTools12: 33
>> Latest patch applied to Amber12: 14
>
>cd $AMBERHOME
>./configure gnu
>make -j 10 install
>
>#############################################
>
>Like described in the AMBER Bugfix section of the AMBER website, we have
>to run the patch script several times.
>Sincerely yours,
>Marko
>
>
>-----Urspr?ngliche Nachricht-----
>Von: Marko Wehle
>Gesendet: Mi 30.01.2013 21:14
>An: amber.ambermd.org
>Betreff: Re: AMBER pmemd and pmemd.cuda, not restart files after
>minimization - fort.16 file instead
>>
>> Hello Amber Developers and Users,
>> I'm experiencing a behavior of the AMBER11 pmemd and pmemd.cuda version,
>> as well in AMBER12 pmemd.cuda, that is new to me.
>> They write out an empty restart file after minimization and put the
>>actual
>> coordinates in an fort.16 called file.
>> Simulations do fine.
>> After a search on the web I found the following entry:
>>
>> http://archive.ambermd.org/201207/0216.html
>>
>> that exactly describes this behavior.
>> The suggestion for solving the problem was recompiling AMBER11 with all
>> patches applied.
>> Unfortunately I could not find any further comments regarding this
>>issue.
>>
>
>bugfix.23 for Amber 11 and bugfix.7 for Amber 12 should fix this issue...
> You can use ./patch_amber.py --patch-level in Amber 12 to see which
>patches have been applied.
>
>
>> But what is irritating here, we've got this particular problem only
>>after
>> we applied all the patches while recompiling AMBER11 and AMBER12 last
>>week.
>> It concerns the pmemd.cuda in AMBER12 only, but in AMBER11 both pmemd
>>and
>> pmemd.cuda.
>> Before recompiling I did not observe this issues. The compile routine
>>and
>> the installation routine went fine so far and no errors were reported.
>>
>
>bugfix.1 for Amber 12 and bugfix.22 for Amber 11 actually introduced this
>issue, but I haven't seen them occur since the previous patches I
>mentioned
>were released.
>
>A temporary workaround is, of course, to use the fort.16 file instead (but
>that won't work particularly well in an automated workflow).
>
>Good luck,
>Jason
>
>_______________________________________________
>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 Tue Feb 05 2013 - 03:30:02 PST
Custom Search