[petsc-dev] Valgrind invalid read in GAMG
Mark F. Adams
mfadams at lbl.gov
Fri Aug 9 13:18:40 CDT 2013
This is happening when the prolongation matrix is repartitioned. The number of active processors is being reduced. MatGetSubmatrix is called just a few lines before this with a square matrix, doing the same basic repartitioning. So this MatGetSubmatrix results in empty local matrices (active processors are being reduced) and seems to be OK with square matrices. Hope this helps.
On Aug 9, 2013, at 10:24 AM, John Mousel <john.mousel at gmail.com> wrote:
> Matt,
>
> Here is the output of ksp_view_binary.
>
> John
>
> binaryoutput
>
>
>
> On Fri, Aug 9, 2013 at 9:14 AM, Matthew Knepley <knepley at gmail.com> wrote:
> On Fri, Aug 9, 2013 at 8:49 AM, John Mousel <john.mousel at gmail.com> wrote:
> I'm getting the following invalid read when I use GAMG. I've included the output of KSPView at the bottom.
>
> This seems to be a more basic problem in MatGetSubmatrix(). Could you give us the matrix? It could be output
> using -ksp_view_binary.
>
> Thanks,
>
> Matt
>
> ==18312== Memcheck, a memory error detector
> ==18312== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
> ==18312== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info
> ==18312== Command: ../src/CYLINDER_EXE -ELAFINT_case_name cylinder -lspaint_ksp_type bcgsl -lspaint_pc_type gamg -lspaint_pc_gamg_type agg -lspaint_pc_gamg_agg_nsmooths 1 -lspaint_pc_gamg_sym_graph true -lspaint_ksp_monitor -pres_ksp_type preonly -pres_pc_type redistribute -pres_redistribute_ksp_type bcgsl -pres_redistribute_pc_type gamg -pres_redistribute_pc_gamg_threshold 0.05 -pres_redistribute_mg_levels_ksp_type richardson -pres_redistribute_mg_levels_pc_type sor -pres_redistribute_mg_coarse_ksp_type richardson -pres_redistribute_mg_coarse_pc_type sor -pres_redistribute_mg_coarse_pc_sor_its 5 -pres_redistribute_pc_gamg_type agg -pres_redistribute_pc_gamg_agg_nsmooths 2 -pres_redistribute_pc_gamg_sym_graph true -pres_redistribute_ksp_initial_guess_nonzero 0 -vel_ksp_monitor -pres_redistribute_ksp_monitor
> ==18312== Parent PID: 18308
> ==18312==
> ==18311== Invalid read of size 8
> ==18311== at 0x4DF2BDC: PetscCheckPointer (checkptr.c:52)
> ==18311== by 0x54370DE: MatSetValues_MPIAIJ (mpiaij.c:506)
> ==18311== by 0x5462AA8: MatGetSubMatrix_MPIAIJ_Private (mpiaij.c:3860)
> ==18311== by 0x5461799: MatGetSubMatrix_MPIAIJ (mpiaij.c:3733)
> ==18311== by 0x551B5E8: MatGetSubMatrix (matrix.c:7322)
> ==18311== by 0x5817535: createLevel (gamg.c:404)
> ==18311== by 0x5819420: PCSetUp_GAMG (gamg.c:630)
> ==18311== by 0x5764D93: PCSetUp (precon.c:890)
> ==18311== by 0x589B211: KSPSetUp (itfunc.c:278)
> ==18311== by 0x589C39A: KSPSolve (itfunc.c:399)
> ==18311== by 0x57555E1: kspsolve_ (itfuncf.c:219)
> ==18311== by 0x61B7A6: axbsolve_ (Axb.F90:139)
> ==18311== Address 0x83d9c20 is 9,376 bytes inside a block of size 9,380 alloc'd
> ==18311== at 0x4A06548: memalign (vg_replace_malloc.c:727)
> ==18311== by 0x4DF2C7D: PetscMallocAlign (mal.c:==18309==
> ==18309== HEAP SUMMARY:
> ==18309== in use at exit: 3,262,243 bytes in 17,854 blocks
> ==18309== total heap usage: 329,856 allocs, 312,002 frees, 136,790,072 bytes allocated
> ==18309==
> ==18309== LEAK SUMMARY:
> ==18309== definitely lost: 0 bytes in 0 blocks
> ==18309== indirectly lost: 0 bytes in 0 blocks
> ==18309== possibly lost: 0 bytes in 0 blocks
> ==18309== still reachable: 3,262,243 bytes in 17,854 blocks
> ==18309== suppressed: 0 bytes in 0 blocks
> ==18309== Rerun with --leak-check=full to see details of leaked memory
> ==18309==
> ==18309== For counts of detected and suppressed errors, rerun with: -v
> ==18309== Use --track-origins=yes to see where uninitialised values come from
> ==18309== ERROR SUMMARY: 24 errors from 1 contexts (suppressed: 6 from 6)
> ,141 frees, 130,288,754 bytes allocated
> ==18310======18310== LEAK SUMMARY:
> ==18310== definitely lost: 0 bytes in 0 blocks
> ==18310== indirectly lost: 0 bytes in 0 blocks
> ==18310== possibly lost: 0 bytes in 0 blocks
> ==18310== still reachable: 3,376,159 bytes in 18,772 blocks
> ==18310== suppressed: 0 bytes in 0 blocks
> ==18310== Rerun with --leak-check=full to see details of leaked memory
> ==18310==
> ==18310== For counts of detected and suppressed errors, rerun with: -v
> ==18310== ERROR SUMMARY: 2 errors from 1 contexts (suppressed: 6 from 6)
>
>
>
> KSP Object:(pres_) 4 MPI processes
> type: preonly
> maximum iterations=20000, initial guess is zero
> tolerances: relative=5e-08, absolute=1e-50, divergence=10000
> left preconditioning
> using NONE norm type for convergence test
> PC Object:(pres_) 4 MPI processes
> type: redistribute
> Number rows eliminated 820 Percentage rows eliminated 11.6213
> Redistribute preconditioner:
> KSP Object: (pres_redistribute_) 4 MPI processes
> type: bcgsl
> BCGSL: Ell = 2
> BCGSL: Delta = 0
> maximum iterations=20000, initial guess is zero
> tolerances: relative=5e-08, absolute=1e-50, divergence=10000
> left preconditioning
> has attached null space
> using PRECONDITIONED norm type for convergence test
> PC Object: (pres_redistribute_) 4 MPI processes
> type: gamg
> MG: type is MULTIPLICATIVE, levels=3 cycles=v
> Cycles per PCApply=1
> Using Galerkin computed coarse grid matrices
> Coarse grid solver -- level -------------------------------
> KSP Object: (pres_redistribute_mg_coarse_) 4 MPI processes
> type: preonly
> maximum iterations=1, initial guess is zero
> tolerances: relative=1e-05, absolute=1e-50, divergence=10000
> left preconditioning
> using NONE norm type for convergence test
> PC Object: (pres_redistribute_mg_coarse_) 4 MPI processes
> type: sor
> SOR: type = local_symmetric, iterations = 5, local iterations = 1, omega = 1
> linear system matrix = precond matrix:
> Mat Object: 4 MPI processes
> type: mpiaij
> rows=61, cols=61
> total: nonzeros=1846, allocated nonzeros=1846
> total number of mallocs used during MatSetValues calls =0
> not using I-node (on process 0) routines
> Down solver (pre-smoother) on level 1 -------------------------------
> KSP Object: (pres_redistribute_mg_levels_1_) 4 MPI processes
> type: richardson
> Richardson: damping factor=1
> maximum iterations=2
> tolerances: relative=1e-05, absolute=1e-50, divergence=10000
> left preconditioning
> using nonzero initial guess
> using NONE norm type for convergence test
> PC Object: (pres_redistribute_mg_levels_1_) 4 MPI processes
> type: sor
> SOR: type = local_symmetric, iterations = 1, local iterations = 1, omega = 1
> linear system matrix = precond matrix:
> Mat Object: 4 MPI processes
> type: mpiaij
> rows=870, cols=870
> total: nonzeros=16020, allocated nonzeros=16020
> total number of mallocs used during MatSetValues calls =0
> not using I-node (on process 0) routines
> Up solver (post-smoother) same as down solver (pre-smoother)
> Down solver (pre-smoother) on level 2 -------------------------------
> KSP Object: (pres_redistribute_mg_levels_2_) 4 MPI processes
> type: richardson
> Richardson: damping factor=1
> maximum iterations=2
> tolerances: relative=1e-05, absolute=1e-50, divergence=10000
> left preconditioning
> using nonzero initial guess
> using NONE norm type for convergence test
> PC Object: (pres_redistribute_mg_levels_2_) 4 MPI processes
> type: sor
> SOR: type = local_symmetric, iterations = 1, local iterations = 1, omega = 1
> linear system matrix = precond matrix:
> Mat Object: 4 MPI processes
> type: mpiaij
> rows=6236, cols=6236
> total: nonzeros=31220, allocated nonzeros=31220
> total number of mallocs used during MatSetValues calls =0
> not using I-node (on process 0) routines
> Up solver (post-smoother) same as down solver (pre-smoother)
> linear system matrix = precond matrix:
> Mat Object: 4 MPI processes
> type: mpiaij
> rows=6236, cols=6236
> total: nonzeros=31220, allocated nonzeros=31220
> total number of mallocs used during MatSetValues calls =0
> not using I-node (on process 0) routines
> linear system matrix = precond matrix:
> Mat Object: 4 MPI processes
> type: mpiaij
> rows=7056, cols=7056
> total: nonzeros=32416, allocated nonzeros=33180
> total number of mallocs used during MatSetValues calls =0
> not using I-node (on process 0) routines
>
>
>
>
>
> --
> 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-dev/attachments/20130809/a1070f6e/attachment.html>
More information about the petsc-dev
mailing list