[petsc-users] Correlation between da_refine and pg_mg_levels

Justin Chang jychang48 at gmail.com
Thu Mar 30 16:17:56 CDT 2017 

Also, this was the output before the error message:

Level 0 domain size (m)    1e+04 x    1e+04 x    1e+03, num elements 5 x 5
x 3 (75), size (m) 2000. x 2000. x 500.
Level -1 domain size (m)    1e+04 x    1e+04 x    1e+03, num elements 2 x 2
x 2 (8), size (m) 5000. x 5000. x 1000.
Level -2 domain size (m)    1e+04 x    1e+04 x    1e+03, num elements 1 x 1
x 1 (1), size (m) 10000. x 10000. x inf.

Which tells me '-da_refine 4' is not registering

On Thu, Mar 30, 2017 at 4:15 PM, Justin Chang <jychang48 at gmail.com> wrote:

> Okay I'll give it a shot.
>
> Somewhat unrelated, but I tried running this on Cori's Haswell node
> (loaded the module 'petsc/3.7.4-64'). But I get these errors:
>
> [0]PETSC ERROR: --------------------- Error Message
> --------------------------------------------------------------
> [0]PETSC ERROR: Argument out of range
> [0]PETSC ERROR: Partition in y direction is too fine! 0 1
> [0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html
> for trouble shooting.
> [0]PETSC ERROR: Petsc Release Version 3.7.4, Oct, 02, 2016
> [0]PETSC ERROR: /global/u1/j/jychang/Icesheet/./ex48 on a
> arch-cori-opt64-INTEL-3.7.4-64 named nid00020 by jychang Thu Mar 30
> 14:04:35 2017
> [0]PETSC ERROR: Configure options --known-sizeof-void-p=8
> --known-sizeof-short=2 --known-sizeof-int=4 --known-sizeof-long=8
> --known-sizeof-long-long=8 --known-sizeof-float=4 --known-sizeof-size_t=8
> --known-mpi-int64_t=1 --known-has-attribute-aligned=1
> --prefix=/global/common/cori/software/petsc/3.7.4-64/hsw/intel
> PETSC_ARCH=arch-cori-opt64-INTEL-3.7.4-64 --COPTFLAGS="-mkl -O2 -no-ipo
> -g -axMIC-AVX512,CORE-AVX2,AVX" --CXXOPTFLAGS="-mkl -O2 -no-ipo -g
> -axMIC-AVX512,CORE-AVX2,AVX" --FOPTFLAGS="-mkl -O2 -no-ipo -g
> -axMIC-AVX512,CORE-AVX2,AVX" --with-hdf5-dir=/opt/cray/pe/
> hdf5-parallel/1.8.16/INTEL/15.0 --with-hwloc-dir=/global/
> common/cori/software/hwloc/1.11.4/hsw --with-scalapack-include=/opt/
> intel/compilers_and_libraries_2017.1.132/linux/mkl/include
> --with-scalapack-lib= --LIBS="-mkl -L/global/common/cori/
> software/petsc/3.7.4-64/hsw/intel/lib -I/global/common/cori/
> software/petsc/3.7.4-64/hsw/intel/include -L/global/common/cori/software/xz/5.2.2/hsw/lib
> -I/global/common/cori/software/xz/5.2.2/hsw/include -L/global/common/cori/
> software/zlib/1.2.8/hsw/intel/lib -I/global/common/cori/
> software/zlib/1.2.8/hsw/intel/include -L/global/common/cori/software/libxml2/2.9.4/hsw/lib
> -I/global/common/cori/software/libxml2/2.9.4/hsw/include
> -L/global/common/cori/software/numactl/2.0.11/hsw/lib
> -I/global/common/cori/software/numactl/2.0.11/hsw/include
> -L/global/common/cori/software/hwloc/1.11.4/hsw/lib -I/global/common/cori/
> software/hwloc/1.11.4/hsw/include -L/global/common/cori/
> software/openssl/1.1.0a/hsw/lib -I/global/common/cori/
> software/openssl/1.1.0a/hsw/include -L/global/common/cori/
> software/subversion/1.9.4/hsw/lib -I/global/common/cori/
> software/subversion/1.9.4/hsw/include -lhwloc -lpciaccess -lxml2 -lz
> -llzma -Wl,--start-group /opt/intel/compilers_and_
> libraries_2017.1.132/linux/mkl/lib/intel64/libmkl_scalapack_lp64.a
> /opt/intel/compilers_and_libraries_2017.1.132/linux/mkl/lib/intel64/libmkl_core.a
> /opt/intel/compilers_and_libraries_2017.1.132/linux/
> mkl/lib/intel64/libmkl_intel_thread.a /opt/intel/compilers_and_
> libraries_2017.1.132/linux/mkl/lib/intel64/libmkl_blacs_intelmpi_lp64.a
> -Wl,--end-group -lstdc++" --download-parmetis --download-metis --with-ssl=0
> --with-batch --known-mpi-shared-libraries=0 --with-clib-autodetect=0
> --with-cxxlib-autodetect=0 --with-debugging=0
> --with-fortranlib-autodetect=0 --with-mpiexec=srun
> --with-shared-libraries=0 --with-x=0 --known-mpi-int64-t=0
> --known-bits-per-byte=8 --known-sdot-returns-double=0
> --known-snrm2-returns-double=0 --known-level1-dcache-assoc=0
> --known-level1-dcache-linesize=32 --known-level1-dcache-size=32768
> --known-memcmp-ok=1 --known-mpi-c-double-complex=1
> --known-mpi-long-double=1 --known-sizeof-MPI_Comm=4
> --known-sizeof-MPI_Fint=4 --known-sizeof-char=1 --known-sizeof-double=8
> CC=cc MPICC=cc CXX=CC MPICXX=CC FC=ftn F77=ftn F90=ftn MPIF90=ftn
> MPIF77=ftn CFLAGS=-axMIC-AVX512,CORE-AVX2,AVX CXXFLAGS=-axMIC-AVX512,CORE-AVX2,AVX
> FFLAGS=-axMIC-AVX512,CORE-AVX2,AVX CC=cc MPICC=cc CXX=CC MPICXX=CC FC=ftn
> F77=ftn F90=ftn MPIF90=ftn MPIF77=ftn CFLAGS=-fPIC FFLAGS=-fPIC
> LDFLAGS=-fPIE --download-hypre --with-64-bit-indices
> [0]PETSC ERROR: #1 DMSetUp_DA_3D() line 298 in
> /global/cscratch1/sd/swowner/sleak/petsc-3.7.4/src/dm/impls/da/da3.c
> [0]PETSC ERROR: #2 DMSetUp_DA() line 27 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/dm/impls/da/dareg.c
> [0]PETSC ERROR: #3 DMSetUp() line 744 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/dm/interface/dm.c
> [0]PETSC ERROR: #4 DMCoarsen_DA() line 1196 in
> /global/cscratch1/sd/swowner/sleak/petsc-3.7.4/src/dm/impls/da/da.c
> [0]PETSC ERROR: #5 DMCoarsen() line 2371 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/dm/interface/dm.c
> [0]PETSC ERROR: #6 PCSetUp_MG() line 616 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/ksp/pc/impls/mg/mg.c
> [0]PETSC ERROR: #7 PCSetUp() line 968 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/ksp/pc/interface/precon.c
> [0]PETSC ERROR: #8 KSPSetUp() line 390 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/ksp/ksp/interface/itfunc.c
> [0]PETSC ERROR: #9 KSPSolve() line 599 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/ksp/ksp/interface/itfunc.c
> [0]PETSC ERROR: #10 SNESSolve_NEWTONLS() line 230 in
> /global/cscratch1/sd/swowner/sleak/petsc-3.7.4/src/snes/impls/ls/ls.c
> [0]PETSC ERROR: #11 SNESSolve() line 4005 in /global/cscratch1/sd/swowner/
> sleak/petsc-3.7.4/src/snes/interface/snes.c
> [0]PETSC ERROR: #12 main() line 1548 in /global/homes/j/jychang/
> Icesheet/ex48.c
> [0]PETSC ERROR: PETSc Option Table entries:
> [0]PETSC ERROR: -da_refine 4
> [0]PETSC ERROR: -ksp_rtol 1e-7
> [0]PETSC ERROR: -M 5
> [0]PETSC ERROR: -N 5
> [0]PETSC ERROR: -P 3
> [0]PETSC ERROR: -pc_mg_levels 5
> [0]PETSC ERROR: -pc_type mg
> [0]PETSC ERROR: -thi_mat_type baij
> [0]PETSC ERROR: ----------------End of Error Message -------send entire
> error message to petsc-maint at mcs.anl.gov----------
> Rank 0 [Thu Mar 30 14:04:35 2017] [c0-0c0s5n0] application called
> MPI_Abort(MPI_COMM_WORLD, 63) - process 0
> srun: error: nid00020: task 0: Aborted
> srun: Terminating job step 4363145.1z
>
> it seems to me the PETSc from this module is not registering the
> '-da_refine' entry. This is strange because I have no issue with this on
> the latest petsc-dev version, anyone know about this error and/or why it
> happens?
>
> On Thu, Mar 30, 2017 at 3:39 PM, Matthew Knepley <knepley at gmail.com>
> wrote:
>
>> On Thu, Mar 30, 2017 at 3:38 PM, Justin Chang <jychang48 at gmail.com>
>> wrote:
>>
>>> Okay, got it. What are the options for setting GAMG as the coarse solver?
>>>
>>
>> -mg_coarse_pc_type gamg I think
>>
>>
>>> On Thu, Mar 30, 2017 at 3:37 PM, Matthew Knepley <knepley at gmail.com>
>>> wrote:
>>>
>>>> On Thu, Mar 30, 2017 at 3:04 PM, Justin Chang <jychang48 at gmail.com>
>>>> wrote:
>>>>
>>>>> Yeah based on my experiments it seems setting pc_mg_levels to
>>>>> $DAREFINE + 1 has decent performance.
>>>>>
>>>>> 1) is there ever a case where you'd want $MGLEVELS <= $DAREFINE? In
>>>>> some of the PETSc tutorial slides (e.g., http://www.mcs.anl.gov/
>>>>> petsc/documentation/tutorials/TutorialCEMRACS2016.pdf on slide
>>>>> 203/227) they say to use $MGLEVELS = 4 and $DAREFINE = 5, but when I ran
>>>>> this, it was almost twice as slow as if $MGLEVELS >= $DAREFINE
>>>>>
>>>>
>>>> Depending on how big the initial grid is, you may want this. There is a
>>>> balance between coarse grid and fine grid work.
>>>>
>>>>
>>>>> 2) So I understand that one cannot refine further than one grid point
>>>>> in each direction, but is there any benefit to having $MGLEVELS > $DAREFINE
>>>>> by a lot?
>>>>>
>>>>
>>>> Again, it depends on the size of the initial grid.
>>>>
>>>> On really large problems, you want to use GAMG as the coarse solver,
>>>> which will move the problem onto a smaller number of nodes
>>>> so that you can coarsen further.
>>>>
>>>>    Matt
>>>>
>>>>
>>>>> Thanks,
>>>>> Justin
>>>>>
>>>>> On Thu, Mar 30, 2017 at 2:35 PM, Barry Smith <bsmith at mcs.anl.gov>
>>>>> wrote:
>>>>>
>>>>>>
>>>>>>   -da_refine $DAREFINE  determines how large the final problem will
>>>>>> be.
>>>>>>
>>>>>>   By default if you don't supply pc_mg_levels then it uses $DAREFINE
>>>>>> + 1 as the number of levels of MG to use; for example -da_refine 1 would
>>>>>> result in 2 levels of multigrid.
>>>>>>
>>>>>>
>>>>>> > On Mar 30, 2017, at 2:17 PM, Justin Chang <jychang48 at gmail.com>
>>>>>> wrote:
>>>>>> >
>>>>>> > Hi all,
>>>>>> >
>>>>>> > Just a general conceptual question: say I am tinkering around with
>>>>>> SNES ex48.c and am running the program with these options:
>>>>>> >
>>>>>> > mpirun -n $NPROCS -pc_type mg -M $XSEED -N $YSEED -P $ZSEED
>>>>>> -thi_mat_type baij -da_refine $DAREFINE -pc_mg_levels $MGLEVELS
>>>>>> >
>>>>>> > I am not too familiar with mg, but it seems to me there is a very
>>>>>> strong correlation between $MGLEVELS and $DAREFINE as well as perhaps even
>>>>>> the initial coarse grid size (provided by $X/YZSEED).
>>>>>> >
>>>>>> > Is there a rule of thumb on how these parameters should be? I am
>>>>>> guessing it probably is also hardware/architectural dependent?
>>>>>> >
>>>>>> > Thanks,
>>>>>> > Justin
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> What most experimenters take for granted before they begin their
>>>> experiments is infinitely more interesting than any results to which their
>>>> experiments lead.
>>>> -- Norbert Wiener
>>>>
>>>
>>>
>>
>>
>> --
>> What most experimenters take for granted before they begin their
>> experiments is infinitely more interesting than any results to which their
>> experiments lead.
>> -- Norbert Wiener
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20170330/24d15d62/attachment.html>

More information about the petsc-users mailing list