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

From: Ray Luo <rluo.uci.edu>
Date: Thu, 18 Feb 2016 12:36:53 -0800

Gerald,

Your fix also worked on my side! Thanks a lot!

I used "make -j 8 install" though I skipped python install in both the
serial and parallel tests.

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 11:23 AM, Ray Luo <rluo.uci.edu> wrote:
> 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 - 13:00:03 PST
Custom Search