[Users] Using gmake - Re: ETK in NERSC machines (Hopper/Carver) does not work properly

Ian Hinder ian.hinder at aei.mpg.de
Wed Mar 14 05:25:33 CDT 2012


On 14 Mar 2012, at 10:48, Jose Fiestas Iquira wrote:

> As mentioned, using the configuration files located in simfactory/mdb/optionslists/ helped. And compiling with gmake worked.
> 
> The only thing it confuses me is that after
> 
> gmake binary-config thornlist='binary,th'

It should be THORNLIST not thornlist, and it should be binary.th not binary,th.  You don't need the single quotes around the thornlist name, as it does not contain any shell-special characters.

In an ideal world, Cactus would complain that you had specified an unrecognised argument (thornlist).  Unfortunately, the current "make"-based system makes that impossible.  There is an open ticket which would solve this problem if it was addressed: https://trac.einsteintoolkit.org/ticket/332.

> I always need to copy the binary,th in configs/binary/
> 
> and then
> 
> gmake binary
> 
> otherwise, the compiler uses all thorns in the arrangements directory, and not only the listed in binary.th. It should not, right?

Since cactus thinks you have not given it any thornlist, it defaults to a thornlist containing all available thorns.

May I ask, why you are not using simfactory?


> 
> Best,
> Jose
> 
> 
> On Tue, Mar 13, 2012 at 1:25 AM, Ian Hinder <ian.hinder at aei.mpg.de> wrote:
> 
> On 13 Mar 2012, at 07:57, Jose Fiestas Iquira wrote:
> 
> > Hello Ian, all,
> >
> > I am trying now to use gmake to create an executable using McLachlan for a
> > simulation of binary coalescence.
> >
> > When I do:
> > gmake grbinary
> >
> > I obtain:
> > .....
> > CST finished.
> > /global/u1/f/fiestas/CACTUS/Cactus/arrangements/Carpet/CarpetLib/src/make.configuration.defn:5:
> > *** Configuration error: The Carpet thorns require MPI.  Please configure
> > with MPI, or remove the Carpet thorns from the ThornList..  Stop.
> > gmake: *** [grbinary] Error 2
> > ..........
> >
> > I checked the configuration and I found following line while
> > doing: gmake grbinary-config thornlist='thornlists/grbinary.th'
> >
> > ............
> > Configuring without MPI
> > ............
> >
> > So, I tried:
> >
> > export mpi=native
> > gmake grbinary-config
> >
> > and
> >
> > - setting mpi=native in a config. file and using it
> > gmake grbinary-config options='grbinary-config.txt'
> >
> > Initially looks OK:
> >
> > fiestas at hopper09:~/CACTUS/Cactus> gmake grbinarypgi-config
> > options='grbinary-config-pgi'
> > Cactus - version: 4.0.0
> > Reconfiguring grbinarypgi.
> > Adding configuration options from 'grbinary-config-pgi'...
> >  Setting mpi to 'native'
> >  Setting CC to 'mpicc'
> >  Setting CXX to 'mpicxx'
> > End of options from 'grbinary-config-pgi'.
> > ....
> >
> > but later I get again the message:
> >
> > ...........
> > Configuring without MPI
> > ........
> >
> > Am I doing something wrong?
> >
> > Btw, if I configure with my own thronlist and add it as above, it keeps
> > reading all thorns in the arrangements directory while compiling. So, I
> > copied my thornlist in configs/grbinary/ to force the compilation using my
> > thornlist. Is it normal?
> 
> Hi Jose,
> 
> You have not specified an optionlist.  There are a large number of variables which usually need to be set on HPC machines (compiler locations, flags, library paths, MPI paths etc).  I strongly recommend that you use simfactory for this, as I believe that most of the work has already been done for the NERSC machines.  If there is a problem with the configuration of simfactory on these machines, then we can help to find the problem and fix it.  Building up an optionlist and submit script etc from scratch will take a lot of time if you aren't very familiar with the machine and Cactus.
> 
> 
> 
> >
> > Thanks,
> > Jose
> >
> >
> > On Mon, Mar 12, 2012 at 4:24 AM, Ian Hinder <ian.hinder at aei.mpg.de> wrote:
> >
> >>
> >> On 10 Mar 2012, at 01:41, Jose Fiestas Iquira wrote:
> >>
> >>> Dear colleagues,
> >>>
> >>> I was trying the last days to install an own application based on ETK in
> >>> NERSC machines (Hopper/Carver), having some problems during the
> >> compilation.
> >>> Before trying to find the real problem, I started from scratch and
> >> realize
> >>> I can not even checkout ETK properly.
> >>> Does somebody tried it in machines different as LONI machines since the
> >>> last release? Would appreciate any experience about it.
> >>> Below my steps:
> >>>
> >>>
> >>> 1. GetComponents <Einstein Toolkit>  (release version)
> >>> curl -O https://raw.github.com/gridaphobe/CRL/master/GetComponents
> >>> ..............................
> >>> *183 components checked out.*
> >>> *  0 components updated.*
> >>> *
> >>> *
> >>> *  Unable to process simfactory2*
> >>> *  Unable to process LSUThorns/QuasiLocalMeasures*
> >>> *  Unable to process LSUThorns/SummationByParts*
> >>> *  Unable to process LSUThorns/Vectors*
> >>
> >> These are all hosted on svn.cct.lsu.edu, which we have found to be very
> >> unreliable.  It might be that the machine (or connection) was down when you
> >> tried it.  Can you try it again?
> >>
> >> We might want to think about moving these repositories to a more reliable
> >> server, for example svn.einsteintoolkit.org.
> >>
> >> Alternatively, it might be that there is a routing problem between NERSC
> >> and CCT.  Can you try
> >>
> >>       ping svn.cct.lsu.edu
> >>
> >> from one of the affected machines?
> >>
> >> PS: your WaveDemo build fails because the wavedemo thornlist is not up to
> >> date.  It needs to be modified to include the Vectors thorn (see the ET
> >> thornlist for an example).
> >>
> >> The Cactus thornlists are out of date - see
> >>
> >>       https://trac.einsteintoolkit.org/ticket/663
> >>
> >> and
> >>
> >>       https://trac.einsteintoolkit.org/ticket/641
> >>
> >>> *
> >>> *
> >>> *Summary of Warnings:*
> >>> *
> >>> *
> >>> *Could not checkout module simfactory2*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/QuasiLocalMeasures*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/SummationByParts*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/Vectors*
> >>> *
> >>> *
> >>>
> >>> *  Time Elapsed: 26 minutes, 19 seconds*
> >>> ..............................................
> >>>
> >>> 2. GetComponent  <WaveDemo>
> >>> ./GetComponents
> >>> http://cactuscode.org/documentation/tutorials/wavetoydemo/WaveDemo.th
> >>>
> >>> * Went OK*
> >>> *
> >>> *
> >>> 3.  cd Cactus
> >>> gmake WaveDemo-config
> >>> gmake WaveDemo
> >>>
> >>>
> >>> *------------------------------------------------------*
> >>> *Warnings were generated during execution of the CST*
> >>> *------------------------------------------------------*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 1:*
> >>> *  -> Duplicate thornname LocalReduce in CactusBase and CactusNumerical*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 2:*
> >>> *  -> Thorn 'CarpetLib' requires the capability 'Vectors'.*
> >>> *     Please add a thorn that provides 'Vectors' to your ThornList or
> >>> remove 'CarpetLib' from it !*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 3:*
> >>> *  -> Thorn 'ML_BSSN' requires the capability 'Vectors'.*
> >>> *     Please add a thorn that provides 'Vectors' to your ThornList or
> >>> remove 'ML_BSSN' from it !*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 4:*
> >>> *  -> Thorn 'ML_BSSN_O2' requires the capability 'Vectors'.*
> >>> *     Please add a thorn that provides 'Vectors' to your ThornList or
> >>> remove 'ML_BSSN_O2' from it !*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 5:*
> >>> *  -> ID_BIN_BH (thorn Meudon_Bin_BH) inherits from SummationByParts*
> >>> *     No thorn in your current ThornList implements SummationByParts*
> >>> *     Either remove Meudon_Bin_BH, or add a thorn to your*
> >>> *      ThornList implementing SummationByParts*
> >>> *     Available thorns in arrangements directory implementing
> >>> SummationByParts:*
> >>> *        LSUThorns/SummationByParts*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 6:*
> >>> *  -> MEUDON_BIN_NS (thorn Meudon_Bin_NS) inherits from SummationByParts*
> >>> *     No thorn in your current ThornList implements SummationByParts*
> >>> *     Either remove Meudon_Bin_NS, or add a thorn to your*
> >>> *      ThornList implementing SummationByParts*
> >>> *     Available thorns in arrangements directory implementing
> >>> SummationByParts:*
> >>> *        LSUThorns/SummationByParts*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 7:*
> >>> *  -> Configuration script for thorn BLAS returned exit code 1*
> >>> *     Error message: 'Error while building BLAS. Aborting.'*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *CST error 8:*
> >>> *  -> Configuration script for thorn LAPACK returned exit code 1*
> >>> *     Error message: 'Error while building LAPACK. Aborting.'*
> >>> *
> >>> *
> >>> *------------------------------------------------------*
> >>> *
> >>> *
> >>> *gmake[1]: ***
> >>>
> >> [/project/projectdirs/snfactry/hshukla/CACTUS/Cactus/configs/WaveDemo/config-data/make.thornlist]
> >>> Error 1*
> >>> *gmake: *** [WaveDemo] Error 2*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *
> >>> 4.  Copied LSUThorns by hand and repeated step 3 and got the same errors.
> >>> *
> >>>
> >>> *183 components checked out.*
> >>> *  0 components updated.*
> >>> *
> >>> *
> >>> *  Unable to process simfactory2*
> >>> *  Unable to process LSUThorns/QuasiLocalMeasures*
> >>> *  Unable to process LSUThorns/SummationByParts*
> >>> *  Unable to process LSUThorns/Vectors*
> >>> *
> >>> *
> >>> *Summary of Warnings:*
> >>> *
> >>> *
> >>> *Could not checkout module simfactory2*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/QuasiLocalMeasures*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/SummationByParts*
> >>> *
> >>> *
> >>> *
> >>> *
> >>> *Could not checkout module LSUThorns/Vectors*
> >>> *
> >>> *
> >>>
> >>> *  Time Elapsed: 26 minutes, 19 seconds*
> >>>
> >>> Best,
> >>> Jose
> >>> _______________________________________________
> >>> Users mailing list
> >>> Users at einsteintoolkit.org
> >>> http://lists.einsteintoolkit.org/mailman/listinfo/users
> >>
> >> --
> >> Ian Hinder
> >> http://numrel.aei.mpg.de/people/hinder
> >>
> >>
> 
> --
> Ian Hinder
> http://numrel.aei.mpg.de/people/hinder
> 
> 

-- 
Ian Hinder
http://numrel.aei.mpg.de/people/hinder



More information about the Users mailing list