Re: [AMBER-Developers] Sander parallel build broken by PBSA updates

From: Ray Luo <rluo.uci.edu>
Date: Thu, 18 Feb 2016 11:23:28 -0800

Thanks, I'm downloading your fix ...

Also, the python-related installation seems to mess up the dependence
issue whether I use single or multi-thread make ... I don't know why
my changes in pbsa and sander have caused this to break.

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, Feb 18, 2016 at 11:08 AM, Gerald Monard
<Gerald.Monard.univ-lorraine.fr> wrote:
> Hi,
>
> I think that I found a fix. The problem was in the makedepend program in
> pbsa/ that had not been updated. That should work now. Can anybody test?
>
> Best,
>
> Gerald.
>
> On 02/18/2016 07:01 PM, Ross Walker wrote:
>> I second this - I tried to disable PBSA building for now but of course it is linked into Sander etc so we'd disabling Sander as well.
>>
>> I'd suggest spending a few hours trying to fix it and if you can't the checkin should be reversed until this is fixed.
>>
>> Serial builds really aren't a viable option these days - try doing a serial build on a Xeon Phi and you'll understand. :-(
>>
>> All the best
>> Ross
>>
>>> On Feb 18, 2016, at 9:13 AM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
>>>
>>> On Wed, Feb 17, 2016 at 6:04 PM, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>> Ross,
>>>>
>>>> Please pull again to see whether it fixed the build problem on your
>>>> side. Looks like multi-thread make does not work ... Too many
>>>> dependence issues ...
>>>
>>> The multi-threaded build definitely doesn't work for me when I hit pbsa stuff:
>>>
>>> $ make -j6 install
>>> ...
>>> make[2]: Entering directory '/home/droe/Amber/GIT/amber/AmberTools/src/pbsa'
>>> gcc -c -O3 -mtune=native  -fPIC  ...
>>> ...
>>> Makefile:238: recipe for target 'pb_p3m.LIBPBSA.o' failed
>>> Makefile:238: recipe for target 'pb_fdfrc.LIBPBSA.o' failed
>>>
>>> Amber is pretty large nowadays, and not being able to build in
>>> parallel is a real drawback. We can currently get around this by
>>> adding a '.NOTPARALLEL' target to the pbsa Makefile (or forcing -j1),
>>> but ideally I think the multi-threaded build should be fixed.
>>>
>>> -Dan
>>>
>>>>
>>>> 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 Wed, Feb 17, 2016 at 12:59 PM, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>>> Okay, I think I've fixed it. Don't know why the same makefile works in
>>>>> amber15 but doesn't work in amber16. I'm testing all the
>>>>> sander/pbsa/nab combs to make sure all can build without interruption
>>>>> and also pass the tests.
>>>>>
>>>>> Ray
>>>>> --
>>>>> Ray Luo, Ph.D.
>>>>> Professor
>>>>> Biochemistry, Molecular Biophysics, Chemical Physics,
>>>>> Chemical and Biomedical Engineering
>>>>> University of California, Irvine, CA 92697-3900
>>>>>
>>>>>
>>>>> On Tue, Feb 16, 2016 at 10:34 PM, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>>>> Finally I can reproduce your problem in building the MPI sander ...
>>>>>>
>>>>>> Will get it fixed tomorrow morning ...
>>>>>>
>>>>>> Ray
>>>>>> --
>>>>>> Ray Luo, Ph.D.
>>>>>> Professor
>>>>>> Biochemistry, Molecular Biophysics, Chemical Physics,
>>>>>> Chemical and Biomedical Engineering
>>>>>> University of California, Irvine, CA 92697-3900
>>>>>>
>>>>>>
>>>>>> On Tue, Feb 16, 2016 at 10:17 PM, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>>>>> Hi Ross,
>>>>>>>
>>>>>>> Looks like that there is no extra untracked file at least in all the
>>>>>>> src folders, though there were indeed two .swp files in the test
>>>>>>> cases. So I think lack of "git clean -fdx" is not the problem. I've
>>>>>>> finished the serial build and is running "test.sander.BASIC". Looking
>>>>>>> good, so far.
>>>>>>>
>>>>>>> I'll do the MPI build next after another "git clean -fdx" ...
>>>>>>>
>>>>>>> However, I think one possible reason could be the use of multi-thread
>>>>>>> in make. There may be inter-dependent issues involved.
>>>>>>>
>>>>>>> I'll do some experiment with the MPI build to see whether this is the cause.
>>>>>>>
>>>>>>> Ray
>>>>>>> --
>>>>>>> Ray Luo, Ph.D.
>>>>>>> Professor
>>>>>>> Biochemistry, Molecular Biophysics, Chemical Physics,
>>>>>>> Chemical and Biomedical Engineering
>>>>>>> University of California, Irvine, CA 92697-3900
>>>>>>>
>>>>>>>
>>>>>>> On Tue, Feb 16, 2016 at 8:24 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
>>>>>>>> Hi Ray,
>>>>>>>>
>>>>>>>> It's not about whether you compiled in that folder - it is whether you maybe edited files in that folder that you never committed / pushed. Try doing:
>>>>>>>>
>>>>>>>> git clean -f -d -x
>>>>>>>> git status
>>>>>>>>
>>>>>>>> And see what it reports.
>>>>>>>>
>>>>>>>> All the best
>>>>>>>> Ross
>>>>>>>>
>>>>>>>>> On Feb 16, 2016, at 23:04, Ray Luo <rluo.uci.edu> wrote:
>>>>>>>>>
>>>>>>>>> Hi Ross,
>>>>>>>>>
>>>>>>>>> Yes, I think my folder is clean since I have never compiled anything
>>>>>>>>> in my amber git folder. Every time after syncing with the master
>>>>>>>>> branch, I copy the whole master branch to a new folder for compiling
>>>>>>>>> and testing.
>>>>>>>>>
>>>>>>>>> I'm about to test this evening's synced version, but will do "git
>>>>>>>>> clean -fxd" before configuring/compiling. Maybe this is why.
>>>>>>>>>
>>>>>>>>> By the way, could you add my uci email to the mailing list as well,
>>>>>>>>> i.e. "ray.luo.uci.edu"?
>>>>>>>>>
>>>>>>>>> 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 Tue, Feb 16, 2016 at 7:51 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
>>>>>>>>>> Hi Ray,
>>>>>>>>>>
>>>>>>>>>> If I do this - on my OSX laptop (GCC 4.9.3), MPICH version 3.1.4
>>>>>>>>>>
>>>>>>>>>> git pull
>>>>>>>>>> git clean -f -d -x
>>>>>>>>>> git status
>>>>>>>>>> On branch master
>>>>>>>>>> Your branch is up-to-date with 'origin/master'.
>>>>>>>>>> nothing to commit, working directory clean
>>>>>>>>>>
>>>>>>>>>> ./configure -mpi gnu
>>>>>>>>>> make install
>>>>>>>>>>
>>>>>>>>>> ...
>>>>>>>>>> ...
>>>>>>>>>> dnrm2.F90:113.24:
>>>>>>>>>>
>>>>>>>>>>      ASSIGN 110 TO NEXT
>>>>>>>>>>                        1
>>>>>>>>>> Warning: Deleted feature: ASSIGN statement at (1)
>>>>>>>>>> mpif90  -DBINTRAJ -DEMIL -DMPI    -c -O3 -mtune=native -fPIC  -ffree-form  -I/Users/rcw/amber/amber/include -I/Users/rcw/amber/amber/include    -o dcopy.o dcopy.F90
>>>>>>>>>> mpif90  -DBINTRAJ -DEMIL -DMPI    -c -O3 -mtune=native -fPIC  -ffree-form  -I/Users/rcw/amber/amber/include -I/Users/rcw/amber/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'
>>>>>>>>>> make[3]: *** [pb_force.o] Error 1
>>>>>>>>>> make[2]: *** [libpbsa] Error 2
>>>>>>>>>> make[1]: *** [parallel] Error 2
>>>>>>>>>> make: *** [install] Error 2
>>>>>>>>>>
>>>>>>>>>> Same problem. Are you sure you have the latest tree and it is clean and doesn't have any modified files locally or files you forgot to add?
>>>>>>>>>>
>>>>>>>>>> All the best
>>>>>>>>>> Ross
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> On Feb 16, 2016, at 15:56, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Hi Ross,
>>>>>>>>>>>
>>>>>>>>>>> So far I can "configure -mpi gnu", "make install", and can also pass
>>>>>>>>>>> "make sander.BASIC.MPI". I'm running "make test.parallel.MM". I'm
>>>>>>>>>>> using "mpirun -np 4" for the mpi jobs. This is on Rocks 6.1/Centos
>>>>>>>>>>> 6.1. Which step was the problem?
>>>>>>>>>>>
>>>>>>>>>>> However I disabled "pbsa_mbfocus" in AmberTools/test/ since I'm in the
>>>>>>>>>>> process of removing this feature to accommodate incoming CUDA. I also
>>>>>>>>>>> need to update the test cases for sander/nab/mmpbsa ... which will be
>>>>>>>>>>> partially in today for sander/nab. I'm trying to get mmpbsa.py to work
>>>>>>>>>>> with the new python environment.
>>>>>>>>>>>
>>>>>>>>>>> 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 Tue, Feb 16, 2016 at 10:21 AM, Ray Luo, Ph.D. <ray.luo.uci.edu> wrote:
>>>>>>>>>>>> Hi Ross,
>>>>>>>>>>>>
>>>>>>>>>>>> I'm looking into this ... There was a major overhaul of the code in
>>>>>>>>>>>> addition to new features in the last check in.
>>>>>>>>>>>>
>>>>>>>>>>>> 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
>>>>>>>>>>
>>>>>>>>
>>>>
>>>> _______________________________________________
>>>> AMBER-Developers mailing list
>>>> AMBER-Developers.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>>
>>>
>>>
>>> --
>>> -------------------------
>>> 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
Received on Thu Feb 18 2016 - 11:30:07 PST
Custom Search