Re: [AMBER-Developers] Parallel make failing

From: Gerald Monard <Gerald.Monard.univ-lorraine.fr>
Date: Thu, 26 May 2016 22:18:03 +0200

Ray, Dan (+any volunteer): can you try (and test) that patch?

I can do the same for LIBPBSA dependencies...;-)

Gerald.

On 05/26/2016 09:41 PM, Gerald Monard wrote:
> Ray,
>
> Yes, my mistake: pb_force.o has a dependency for sa_driver.o.
> But it's weirder(?) than this: if I track the compilation, a problem
> appears for libFpbsa.a (pbsa library for sander). I can see this (in
> serial order):
>
> gfortran [...] -DSANDER [...] -I../sander \ -o sa_driver.SANDER.o
> sa_driver.F90
> gfortran [...] -DSANDER [...] -I../sander \ -o pb_force.SANDER.o
> pb_force.F90
> gfortran [...] -DSANDER [...] -I../sander \ -o pb_p3m.SANDER.o pb_p3m.F90
> gfortran [...] -DSANDER [...] -I../sander \ -o interpX.SANDER.o interpX.F90
> gfortran [...] [...] -o pb_force.o pb_force.F90
> gfortran [...] [...] -o pb_direct.o pb_direct.F90
>
> It means that pb_force is compiled twice (thus the error for the
> parallel make: pb_direct needs pb_force that needs sa_driver which will
> create the solvent_accessibility module; while pb_force.SANDER.o needs
> sa_driver.SANDER.o that will create the solvent_accessibility module!).
>
> Origin: makedepend has a definition for pb_direct.SANDER.o, but Makefile
> compiles pb_direct.o as requested by SANDERPBSAOBJS.
> Two options
> 1) It should be said pb_direct.SANDER.o in the Makefile (to be
> consistent with makedepend)
> 2) makedepend should not contain the $SANDER{"pb_direct"} directive (to
> be consistent with Makefile)
>
> I'm checking the discrepancy between Makefile and makedepend and will
> commit a patch asap.
>
> Ray: you know more than me what is needed: to patch Makefile or to patch
> makedepend? (I guess it depends where #ifdef SANDER are present)
>
> Gerald.
>
> On 05/26/2016 09:21 PM, Ray Luo wrote:
>> Gerald,
>>
>> Somehow in my "depend", as generated by "makedepend", "pb_force.o"
>> lists "sa_driver.o" as one of the depends in all three cases, i.e. for
>> PBSA, SANDER, and LIBPBSA ... I suppose "makedepend" is okay?
>>
>> All the best,
>> Ray
>> --
>> Ray Luo, Ph.D.
>> Professor
>> Biochemistry, Molecular Biophysics, Chemical Physics,
>> Chemical and Biomedical Engineering
>> University of California, Irvine, CA 92697-3900
>>
>>
>> On Thu, May 26, 2016 at 11:31 AM, Gerald Monard
>> <Gerald.Monard.univ-lorraine.fr> wrote:
>>> Hello,
>>>
>>> FYI, based on Dan's error report: pbsa/pb_force.F90 needs the
>>> solvent_accessibility module from sa_driver.F90, but makedepend does not
>>> recognize this dependency. So if you're lucky the first one get compiled
>>> before the second one (the files are in "correct" order in the
>>> Makefile), if not, then you're doomed for a compile error.
>>>
>>> Gerald.
>>>
>>> On 05/26/2016 07:37 PM, Gerald Monard wrote:
>>>> Hello,
>>>>
>>>> This issue was raised some weeks/months ago (see thread
>>>> [AMBER-Developers] Sander parallel build broken by PBSA updates). If
>>>> it's still the same problem, it's a dependency problem in PBSA
>>>> (makedepend not making its work correctly). You can't see it with make
>>>> -j1 and it's a random problem with make -jX (X>1).
>>>> I (partially) solved it last time. I'll check into it asap.
>>>>
>>>> Gerald.
>>>>
>>>> On 05/26/2016 07:23 PM, Ray Luo wrote:
>>>>> First thing first, I always "git clean -xfd" before rebuilding. I'll try
>>>>> different thead values to see whether I can reproduce the problem.
>>>>>
>>>>> Also all object files for sander are separately defined and built so this
>>>>> could be something higher up.
>>>>>
>>>>> Ray
>>>>> On May 26, 2016 10:04 AM, "Daniel Roe" <daniel.r.roe.gmail.com> wrote:
>>>>>
>>>>> On Thu, May 26, 2016, Ray Luo wrote:
>>>>>> Okay, I've just confirmed again that I can
>>>>>> make -j 32 install
>>>>>> if I turned off python building. I don't know too much how the python
>>>>>> part is built ... Please take a look of the dependence relations
>>>>>> related to python building.
>>>>>
>>>>> Nope - still fails for me with '--skip-python'. It's almost certainly
>>>>> an issue with pbsa/libpbsa dependencies since that is where the build
>>>>> always fails for me ('make -j4'):
>>>>>
>>>>> ```
>>>>> mpif90 -DBINTRAJ -DMPI -c -O3 -mtune=native -fPIC -ffree-form
>>>>> -I/home/droe/Amber/GIT/amber/include
>>>>> -I/home/droe/Amber/GIT/amber/include -o pb_force.o pb_force.F90
>>>>> pb_force.F90:292:77:
>>>>>
>>>>> sa_init, sa_driver, sa_free,
>>>>> sa_free_mb, &
>>>>>
>>>>> 1
>>>>> Error: Symbol ‘saslave_init’ referenced at (1) not found in module
>>>>> ‘solvent_accessibility’
>>>>> Makefile:226: recipe for target 'pb_force.o' failed
>>>>> make[3]: *** [pb_force.o] Error 1
>>>>> ```
>>>>>
>>>>> -Dan
>>>>>
>>>>> --
>>>>> -------------------------
>>>>> Daniel R. Roe, PhD
>>>>> Department of Medicinal Chemistry
>>>>> University of Utah
>>>>> 30 South 2000 East, Room 307
>>>>> Salt Lake City, UT 84112-5820
>>>>> http://home.chpc.utah.edu/~cheatham/
>>>>> (801) 587-9652
>>>>> (801) 585-6208 (Fax)
>>>>>
>>>>> _______________________________________________
>>>>> AMBER-Developers mailing list
>>>>> AMBER-Developers.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>>>> _______________________________________________
>>>>> AMBER-Developers mailing list
>>>>> AMBER-Developers.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>>>>
>>>>
>>>
>>> --
>>> ____________________________________________________________________________
>>>
>>> Prof. Gerald MONARD
>>> SRSMC, Université de Lorraine, CNRS
>>> Boulevard des Aiguillettes B.P. 70239
>>> F-54506 Vandoeuvre-les-Nancy, FRANCE
>>>
>>> e-mail : Gerald.Monard.univ-lorraine.fr
>>> tel. : +33 (0)383.684.381
>>> fax : +33 (0)383.684.371
>>> web : http://www.monard.info
>>>
>>> ____________________________________________________________________________
>>>
>>>
>>> _______________________________________________
>>> AMBER-Developers mailing list
>>> AMBER-Developers.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>
>> _______________________________________________
>> AMBER-Developers mailing list
>> AMBER-Developers.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>
>

-- 
____________________________________________________________________________
  Prof. Gerald MONARD
  SRSMC, Université de Lorraine, CNRS
  Boulevard des Aiguillettes B.P. 70239
  F-54506 Vandoeuvre-les-Nancy, FRANCE
  e-mail : Gerald.Monard.univ-lorraine.fr
  tel.   : +33 (0)383.684.381
  fax    : +33 (0)383.684.371
  web    : http://www.monard.info
____________________________________________________________________________



_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers

Received on Thu May 26 2016 - 13:30:03 PDT
Custom Search