[petsc-users] Use block Jacobi preconditioner with SNES
Ali Reza Khaz'ali
arkhazali at cc.iut.ac.ir
Tue Aug 28 04:34:08 CDT 2018
> Actually you do not need my new branch to achieve what you desired. All you need in your main program is something like
>
> ierr = SNESCreate(PETSC_COMM_WORLD,&snes);CHKERRQ(ierr);
> ierr = SNESGetKSP(snes,&ksp);CHKERRQ(ierr);
> ierr = KSPGetPC(ksp,&pc);CHKERRQ(ierr);
> ierr = PCSetType(pc,PCBJACOBI);CHKERRQ(ierr);
> ierr = PCBJacobiSetTotalBlocks(pc,3,lens);CHKERRQ(ierr); /* here you set your block sizes to whatever you need */
>
> Then simply do not call PCBJacobiGetSubKSP() but use the options database to set the inner solver with -sub_pc_type lu -sub_ksp_type preonly
>
> I have updated the branch to move the PCBJacobiSetTotalBlocks() to the main program but left the callback in there for setting the inner solver types (though as I just said you don't need to use the callback since you can control the solver from the options database). The callback is needed, if, for example, you wished to use a different solver on different blocks (which is not your case).
>
> Barry
>
> PETSc developers - do you think we should put the callback functionality into PETSc? It allows doing things that are otherwise not doable but is rather ugly (perhaps too specialized)?
>
>
>
It works! Thanks a lot. Here is log of a 30x30x10 system (18000 blocks,
with GMRes solver). I like to have variable sized block preconditioners
and solvers in PETSc. Their application is more than it may first
appear. If it is possible, I would like to contribute to PETSc code, to
build a variable sized block Jacobi and block ILU(k) at the first step
(If I can, of course). Where can I start?
type: newtonls
maximum iterations=2000, maximum function evaluations=2000
tolerances: relative=0.0001, absolute=1e-05, solution=1e-05
total number of linear solver iterations=3
total number of function evaluations=2
norm schedule ALWAYS
SNESLineSearch Object: 1 MPI processes
type: bt
interpolation: cubic
alpha=1.000000e-04
maxstep=1.000000e+08, minlambda=1.000000e-12
tolerances: relative=1.000000e-08, absolute=1.000000e-15,
lambda=1.000000e-08
maximum iterations=40
KSP Object: 1 MPI processes
type: gmres
restart=30, using Classical (unmodified) Gram-Schmidt
Orthogonalization with no iterative refinement
happy breakdown tolerance 1e-30
maximum iterations=5000, initial guess is zero
tolerances: relative=1e-05, absolute=1e-06, divergence=10000.
left preconditioning
using PRECONDITIONED norm type for convergence test
PC Object: 1 MPI processes
type: bjacobi
number of blocks = 18000
Local solve is same for all blocks, in the following KSP and PC
objects:
KSP Object: (sub_) 1 MPI processes
type: preonly
maximum iterations=10000, initial guess is zero
tolerances: relative=1e-05, absolute=1e-50, divergence=10000.
left preconditioning
using NONE norm type for convergence test
PC Object: (sub_) 1 MPI processes
type: lu
out-of-place factorization
tolerance for zero pivot 2.22045e-14
matrix ordering: nd
factor fill ratio given 0., needed 0.
Factored matrix follows:
Mat Object: 1 MPI processes
type: mkl_pardiso
rows=6, cols=6
package used to perform factorization: mkl_pardiso
total: nonzeros=26, allocated nonzeros=26
total number of mallocs used during MatSetValues calls =0
MKL_PARDISO run parameters:
MKL_PARDISO phase: 33
MKL_PARDISO iparm[1]: 1
MKL_PARDISO iparm[2]: 2
MKL_PARDISO iparm[3]: 1
MKL_PARDISO iparm[4]: 0
MKL_PARDISO iparm[5]: 0
MKL_PARDISO iparm[6]: 0
MKL_PARDISO iparm[7]: 0
MKL_PARDISO iparm[8]: 0
MKL_PARDISO iparm[9]: 0
MKL_PARDISO iparm[10]: 13
MKL_PARDISO iparm[11]: 1
MKL_PARDISO iparm[12]: 0
MKL_PARDISO iparm[13]: 1
MKL_PARDISO iparm[14]: 0
MKL_PARDISO iparm[15]: 144
MKL_PARDISO iparm[16]: 144
MKL_PARDISO iparm[17]: 0
MKL_PARDISO iparm[18]: 37
MKL_PARDISO iparm[19]: 0
MKL_PARDISO iparm[20]: 0
MKL_PARDISO iparm[21]: 0
MKL_PARDISO iparm[22]: 0
MKL_PARDISO iparm[23]: 0
MKL_PARDISO iparm[24]: 0
MKL_PARDISO iparm[25]: 0
MKL_PARDISO iparm[26]: 0
MKL_PARDISO iparm[27]: 0
MKL_PARDISO iparm[28]: 0
MKL_PARDISO iparm[29]: 0
MKL_PARDISO iparm[30]: 0
MKL_PARDISO iparm[31]: 0
MKL_PARDISO iparm[32]: 0
MKL_PARDISO iparm[33]: 0
MKL_PARDISO iparm[34]: -1
MKL_PARDISO iparm[35]: 1
MKL_PARDISO iparm[36]: 0
MKL_PARDISO iparm[37]: 0
MKL_PARDISO iparm[38]: 0
MKL_PARDISO iparm[39]: 0
MKL_PARDISO iparm[40]: 0
MKL_PARDISO iparm[41]: 0
MKL_PARDISO iparm[42]: 0
MKL_PARDISO iparm[43]: 0
MKL_PARDISO iparm[44]: 0
MKL_PARDISO iparm[45]: 0
MKL_PARDISO iparm[46]: 0
MKL_PARDISO iparm[47]: 0
MKL_PARDISO iparm[48]: 0
MKL_PARDISO iparm[49]: 0
MKL_PARDISO iparm[50]: 0
MKL_PARDISO iparm[51]: 0
MKL_PARDISO iparm[52]: 0
MKL_PARDISO iparm[53]: 0
MKL_PARDISO iparm[54]: 0
MKL_PARDISO iparm[55]: 0
MKL_PARDISO iparm[56]: 0
MKL_PARDISO iparm[57]: -1
MKL_PARDISO iparm[58]: 0
MKL_PARDISO iparm[59]: 0
MKL_PARDISO iparm[60]: 0
MKL_PARDISO iparm[61]: 144
MKL_PARDISO iparm[62]: 145
MKL_PARDISO iparm[63]: 21
MKL_PARDISO iparm[64]: 0
MKL_PARDISO maxfct: 1
MKL_PARDISO mnum: 1
MKL_PARDISO mtype: 11
MKL_PARDISO n: 6
MKL_PARDISO nrhs: 1
MKL_PARDISO msglvl: 0
linear system matrix = precond matrix:
Mat Object: 1 MPI processes
type: seqaij
rows=6, cols=6
total: nonzeros=26, allocated nonzeros=26
total number of mallocs used during MatSetValues calls =0
using I-node routines: found 4 nodes, limit used is 5
linear system matrix = precond matrix:
Mat Object: 1 MPI processes
type: seqaij
rows=108000, cols=108000
total: nonzeros=2868000, allocated nonzeros=8640000
total number of mallocs used during MatSetValues calls =0
not using I-node routines
************************************************************************************************************************
*** WIDEN YOUR WINDOW TO 120 CHARACTERS. Use 'enscript -r
-fCourier9' to print this document ***
************************************************************************************************************************
---------------------------------------------- PETSc Performance
Summary: ----------------------------------------------
E:\Documents\Visual Studio 2015\Projects\compsim\x64\Release\compsim.exe
on a named ALIREZA-PC with 1 processor, by AliReza Tue Aug 28 13:57:09 2018
Using Petsc Development GIT revision: v3.9.3-1238-gce82fdcfd6 GIT Date:
2018-08-27 15:47:19 -0500
Max Max/Min Avg Total
Time (sec): 1.353e+02 1.000 1.353e+02
Objects: 1.980e+05 1.000 1.980e+05
Flop: 2.867e+07 1.000 2.867e+07 2.867e+07
Flop/sec: 2.119e+05 1.000 2.119e+05 2.119e+05
MPI Messages: 0.000e+00 0.000 0.000e+00 0.000e+00
MPI Message Lengths: 0.000e+00 0.000 0.000e+00 0.000e+00
MPI Reductions: 0.000e+00 0.000
Flop counting convention: 1 flop = 1 real number operation of type
(multiply/divide/add/subtract)
e.g., VecAXPY() for real vectors of length
N --> 2N flop
and VecAXPY() for complex vectors of length
N --> 8N flop
Summary of Stages: ----- Time ------ ----- Flop ------ --- Messages
--- -- Message Lengths -- -- Reductions --
Avg %Total Avg %Total Count
%Total Avg %Total Count %Total
0: Main Stage: 1.3529e+02 100.0% 2.8668e+07 100.0% 0.000e+00
0.0% 0.000e+00 0.0% 0.000e+00 0.0%
------------------------------------------------------------------------------------------------------------------------
See the 'Profiling' chapter of the users' manual for details on
interpreting output.
Phase summary info:
Count: number of times phase was executed
Time and Flop: Max - maximum over all processors
Ratio - ratio of maximum to minimum over all processors
Mess: number of messages sent
AvgLen: average message length (bytes)
Reduct: number of global reductions
Global: entire computation
Stage: stages of a computation. Set stages with PetscLogStagePush()
and PetscLogStagePop().
%T - percent time in this phase %F - percent flop in this
phase
%M - percent messages in this phase %L - percent message
lengths in this phase
%R - percent reductions in this phase
Total Mflop/s: 10e-6 * (sum of flop over all processors)/(max time
over all processors)
------------------------------------------------------------------------------------------------------------------------
Event Count Time (sec)
Flop --- Global --- --- Stage ---- Total
Max Ratio Max Ratio Max Ratio Mess AvgLen
Reduct %T %F %M %L %R %T %F %M %L %R Mflop/s
------------------------------------------------------------------------------------------------------------------------
--- Event Stage 0: Main Stage
BuildTwoSidedF 2 1.0 1.2701e-04 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
SNESSolve 1 1.0 1.1583e+02 1.0 2.87e+07 1.0 0.0e+00 0.0e+00
0.0e+00 86100 0 0 0 86100 0 0 0 0
SNESFunctionEval 2 1.0 5.4101e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 4 0 0 0 0 4 0 0 0 0 0
SNESJacobianEval 1 1.0 9.3770e+01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 69 0 0 0 0 69 0 0 0 0 0
SNESLineSearch 1 1.0 3.1033e+00 1.0 6.82e+06 1.0 0.0e+00 0.0e+00
0.0e+00 2 24 0 0 0 2 24 0 0 0 2
VecDot 1 1.0 1.8688e-04 1.0 2.16e+05 1.0 0.0e+00 0.0e+00
0.0e+00 0 1 0 0 0 0 1 0 0 0 1156
VecMDot 3 1.0 9.9299e-04 1.0 1.30e+06 1.0 0.0e+00 0.0e+00
0.0e+00 0 5 0 0 0 0 5 0 0 0 1305
VecNorm 7 1.0 6.0845e-03 1.0 1.51e+06 1.0 0.0e+00 0.0e+00
0.0e+00 0 5 0 0 0 0 5 0 0 0 248
VecScale 4 1.0 1.4437e+00 1.0 4.32e+05 1.0 0.0e+00 0.0e+00
0.0e+00 1 2 0 0 0 1 2 0 0 0 0
VecCopy 3 1.0 1.6059e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 1 0 0 0 0 1 0 0 0 0 0
VecSet 90002 1.0 1.3843e-02 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
VecAXPY 1 1.0 3.1733e-01 1.0 2.16e+05 1.0 0.0e+00 0.0e+00
0.0e+00 0 1 0 0 0 0 1 0 0 0 1
VecWAXPY 1 1.0 2.2665e-04 1.0 1.08e+05 1.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 477
VecMAXPY 4 1.0 8.6085e-04 1.0 1.94e+06 1.0 0.0e+00 0.0e+00
0.0e+00 0 7 0 0 0 0 7 0 0 0 2258
VecAssemblyBegin 2 1.0 1.6379e-04 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
VecAssemblyEnd 2 1.0 1.4112e-05 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
VecReduceArith 2 1.0 3.1304e-04 1.0 4.32e+05 1.0 0.0e+00 0.0e+00
0.0e+00 0 2 0 0 0 0 2 0 0 0 1380
VecReduceComm 1 1.0 2.1382e-06 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
VecNormalize 4 1.0 1.4441e+00 1.0 1.30e+06 1.0 0.0e+00 0.0e+00
0.0e+00 1 5 0 0 0 1 5 0 0 0 1
MatMult 4 1.0 2.0402e-02 1.0 2.25e+07 1.0 0.0e+00 0.0e+00
0.0e+00 0 79 0 0 0 0 79 0 0 0 1103
MatSolve 72000 1.0 5.3514e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 4 0 0 0 0 4 0 0 0 0 0
MatLUFactorSym 18000 1.0 1.9405e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 1 0 0 0 0 1 0 0 0 0 0
MatLUFactorNum 18000 1.0 1.8373e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatAssemblyBegin 18002 1.0 1.0409e-03 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatAssemblyEnd 18002 1.0 3.3879e-02 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatGetRowIJ 18000 1.0 3.1819e-02 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatCreateSubMats 1 1.0 3.7015e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatGetOrdering 18000 1.0 3.0787e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatZeroEntries 1 1.0 2.7952e-02 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
MatView 3 1.0 2.9153e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
KSPSetUp 18001 1.0 7.5898e-03 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 0 0 0 0 0 0 0 0 0 0 0
KSPSolve 1 1.0 1.6244e+01 1.0 2.16e+07 1.0 0.0e+00 0.0e+00
0.0e+00 12 75 0 0 0 12 75 0 0 0 1
KSPGMRESOrthog 3 1.0 8.4669e-02 1.0 2.59e+06 1.0 0.0e+00 0.0e+00
0.0e+00 0 9 0 0 0 0 9 0 0 0 31
PCSetUp 18001 1.0 3.3536e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 2 0 0 0 0 2 0 0 0 0 0
PCSetUpOnBlocks 1 1.0 2.5973e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 2 0 0 0 0 2 0 0 0 0 0
PCApply 4 1.0 6.2752e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 5 0 0 0 0 5 0 0 0 0 0
PCApplyOnBlocks 72000 1.0 5.9278e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00
0.0e+00 4 0 0 0 0 4 0 0 0 0 0
------------------------------------------------------------------------------------------------------------------------
Memory usage is given in bytes:
Object Type Creations Destructions Memory Descendants' Mem.
Reports information only for process 0.
--- Event Stage 0: Main Stage
SNES 1 0 0 0.
DMSNES 1 0 0 0.
SNESLineSearch 1 0 0 0.
Vector 36020 0 0 0.
Matrix 36001 0 0 0.
Distributed Mesh 2 0 0 0.
Index Set 90000 36000 29088000 0.
Star Forest Graph 4 0 0 0.
Discrete System 2 0 0 0.
Krylov Solver 18001 0 0 0.
DMKSP interface 1 0 0 0.
Preconditioner 18001 0 0 0.
Viewer 1 0 0 0.
========================================================================================================================
Average time to get PetscTime(): 1.28294e-07
#PETSc Option Table entries:
-ksp_atol 1e-6
-ksp_rtol 1e-5
-snes_rtol 1e-4
-sub_ksp_type preonly
-sub_pc_factor_mat_solver_type mkl_pardiso
-sub_pc_type lu
#End of PETSc Option Table entries
Compiled without FORTRAN kernels
Compiled with full precision matrices (default)
sizeof(short) 2 sizeof(int) 4 sizeof(long) 4 sizeof(void*) 8
sizeof(PetscScalar) 8 sizeof(PetscInt) 4
Configure options: --prefix=/home/alireza/PetscGit
--with-mkl_pardiso-dir=/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mkl
--with-hypre-incl
ude=/cygdrive/E/hypre-2.11.2/Builds/Bins/include
--with-hypre-lib=/cygdrive/E/hypre-2.11.2/Builds/Bins/lib/HYPRE.lib
--with-ml-include=/cygdrive/E/Trilinos-master/Bins/in
clude --with-ml-lib=/cygdrive/E/Trilinos-master/Bins/lib/ml.lib
ظ€ôwith-openmp --with-cc="win32fe icl" --with-fc="win32fe ifort"
--with-mpi-include=/cygdrive/E/Program_Fi
les_x86/IntelSWTools/compilers_and_libraries/windows/mpi/intel64/include
--with-mpi-lib=/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mpi/int
el64/lib/impi.lib
--with-mpi-mpiexec=/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mpi/intel64/bin/mpiexec.exe
--with-debugging=0 --with-blas
-lib=/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mkl/lib/intel64_win/mkl_rt.lib
--with-lapack-lib=/cygdrive/E/Program_Files_x86/IntelSWTool
s/compilers_and_libraries/windows/mkl/lib/intel64_win/mkl_rt.lib
-CFLAGS="-O2 -MT -wd4996 -Qopenmp" -CXXFLAGS="-O2 -MT -wd4996 -Qopenmp"
-FFLAGS="-MT -O2 -Qopenmp"
-----------------------------------------
Libraries compiled on 2018-08-27 22:42:15 on AliReza-PC
Machine characteristics: CYGWIN_NT-6.1-2.10.0-0.325-5-3-x86_64-64bit
Using PETSc directory: /home/alireza/PetscGit
Using PETSc arch:
-----------------------------------------
Using C compiler: /home/alireza/PETSc/lib/petsc/bin/win32fe/win32fe icl
-O2 -MT -wd4996 -Qopenmp
Using Fortran compiler:
/home/alireza/PETSc/lib/petsc/bin/win32fe/win32fe ifort -MT -O2 -Qopenmp
-fpp
-----------------------------------------
Using include paths: -I/home/alireza/PetscGit/include
-I/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mkl/include
-I/cygdrive/E/hypre-2.11.2/
Builds/Bins/include -I/cygdrive/E/Trilinos-master/Bins/include
-I/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mpi/intel64/include
-----------------------------------------
Using C linker: /home/alireza/PETSc/lib/petsc/bin/win32fe/win32fe icl
Using Fortran linker: /home/alireza/PETSc/lib/petsc/bin/win32fe/win32fe
ifort
Using libraries: -L/home/alireza/PetscGit/lib
-L/home/alireza/PetscGit/lib -lpetsc
/cygdrive/E/hypre-2.11.2/Builds/Bins/lib/HYPRE.lib
/cygdrive/E/Trilinos-master/Bins/lib
/ml.lib
/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mkl/lib/intel64_win/mkl_rt.lib
/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and
_libraries/windows/mkl/lib/intel64_win/mkl_rt.lib
/cygdrive/E/Program_Files_x86/IntelSWTools/compilers_and_libraries/windows/mpi/intel64/lib/impi.lib
Gdi32.lib User32.lib
Advapi32.lib Kernel32.lib Ws2_32.lib
-----------------------------------------
More information about the petsc-users
mailing list