Commits

Show all
Author Commit Message Labels Comments Date
Mark Adams
PCGAMG: change default nsmooths, reuse_interpolation, threshold -pc_gamg_agg_nsmooths 1 (was 0) Although unsmoothed aggregation is more robust, smoothed is faster for the problems that AMG is most frequently used for. -pc_gamg_reuse_interpolation FALSE (was TRUE) Lagging interpolation by default causes GAMG to spuriously fail for many nonlinear problems and transient problems with moving coefficient structure, and with variable time steps. -pc_…
Matt Knepley
DMPlex: Add missing declaration - This should merge alright with the other branch which fixes it
Matt Knepley
DMPlex: Submesh fixes - Coordinates must use a field - Allow same subpointMap to be specified again
Matt Knepley
DMPlex: Fix submesh marking for uninterpolated meshes - This now works for embedded submeshes
Matt Knepley
DMPlex: Interpolation corner cases - Do not interpolate a 1D mesh - No need to copy coordinates of the same DM
Matt Knepley
DMPlex: Interpolation of a submesh now starts at the faces - Do not know a better way to do this right now
Matt Knepley
DMPlex: Fix error in work space allocation
Matt Knepley
ISieve: Coordinate conversion should add a field
Jed Brown
Merge branch 'knepley/pylith' * knepley/pylith: DMPlex: Fixed DMPlexCreateSohesiveSubmesh() - We should take the fault face directly from the cohesive cell rather than using the face insertion routine DMPlex: Add argument to DMPlexCreateSubmesh() that allows user to choose stratum - Added Fortran binding - Updated test DMPlex: Added DMPlexCreateCohesiveSubmesh() - Needed to have submesh faces in the same order as cohesiv…
Jed Brown
Merge branch 'balay/mpich-3.0.4' Upgrade to v3.0.4-106-g3adb59c. This nighly is used because it fixes the FreeBSD bug identified in this branch. * balay/mpich-3.0.4: mpich: revert workarround on freebsd as its fixed in the current tarball mpich: update to pre 3.0.5 snapshot. This fixes a malloc bug in mpiexec [this bug is visible on freebsd - but exists on other arches aswell] http://trac.mpich.org/projects/mpich/ticket/182…
Jed Brown
Merge branch 'hzhang/update-nightlytestoutput' Preallocation was simplified in 'karlrupp/fix-mpiaij-prealloc', but that branch forgot to update test output. * hzhang/update-nightlytestoutput: Mat: fix test output after MPIAIJ preallocation change update output file of nightly test
Matt Knepley
SNES ex12: Fix leak
Jed Brown
Mat: fix test output after MPIAIJ preallocation change The behavior change was introduced in 'karlrupp/fix-mpiaij-prealloc', which was merged to 'master' in 714d7381235fa6b645d.
Jed Brown
Merge branch 'toby/fix-sor-baij' Consolidates code and adds support for BAIJ SOR with nonzero initial guess (required for use as a MG smoother). * toby/fix-sor-baij: Mat: add MatSOR/BAIJ test to nightlies Mat: move new MatSOR macros to a common header blockmatmult.h MatSOR_SeqBAIJ: refactor, enable nonzero initial guess. MatSOR, test ex43: removed BlockMat from test MatSOR: added test ex43, currently fails MatSOR_Seq…
Jed Brown
Merge branch 'toby/pcml-nnz' * toby/pcml-nnz: PCML: set N_nonzeros so thatfine-grid operator complexity is reported
Jed Brown
Merge branch 'dalcinl/ts-monitor' * dalcinl/ts-monitor: TS: remove last-step '-1 TS' monitor lines from output TS: Remove last call to TSMonitor() with step=-1
Jed Brown
Merge branch 'jed/configure-check-once' * jed/configure-check-once: config: test functions and headers only once
Jed Brown
Merge branch 'jed/exodusii-5.24' * jed/exodusii-5.24: ExodusII: link Fortran if and only if using Fortran ExodusII: require HDF5 due to transitive dependency via --enable-netcdf-4 ExodusII: update to 5.24
Jed Brown
Merge branch 'jed/petscdlopen-extern' * jed/petscdlopen-extern: PetscDL: use PETSC_EXTERN and remove obsolete header
Jed Brown
Merge branch 'jed/fix-optionsinsertfile-large' * jed/fix-optionsinsertfile-large: PetscOptionsInsertFile: fix overflow for large file sizes [fix #39]
Jed Brown
Merge branch 'jed/matmatmult-sort' New version provides in the neighborhood of 20% speedup for MatMatMult as used in PCGAMG. * jed/matmatmult-sort: MatMatMultSymbolic_SeqAIJ_SeqAIJ: switch to "sort" version as default segbuffer: g++ does not like casting int*restrict* to void*, void** okay MatMatMult_SeqAIJ_SeqAIJ_Sort: new implementation using char flags and sorting segbuffer: add PetscSegBufferGetInts() to encourage use …
Satish Balay
fun3d: fix c89 warning
Jed Brown
PetscStackCallBLAS: new macro instead of PetscStackCall (skip CHKMEMQ) When CHKMEMQ occurs on every call to BLAS or Lapack, we sometimes see unacceptable performance degradation (orders of magnitude). PetscStackCallBLAS is currently identical to PetscStackCall except for skipping CHKMEMQ, but such checking could be made optional in the future.
Jed Brown
Stack: consolidate handling of PetscStack This is in preparation for addition of PetscStackCallBLAS(), for which I would like to avoid further duplication (it has already led to petscstackp->petscroutine becoming stale in case of PetscStackPush). This change has only one functional change: PetscStackPush will now report the file and line number of the call site instead of "unknown" and 0. This is more useful because it distinguishes mul…
Jed Brown
PetscDL: use PETSC_EXTERN and remove obsolete header
Jed Brown
TS: remove last-step '-1 TS' monitor lines from output
Jed Brown
config: test functions and headers only once Many functions and headers are mentioned in both BuildSystem and PETSc, thus being tested more than once. Removing this redundancy speeds up my configure from 125 seconds to 110 seconds.
Jed Brown
BuildSystem: fix comment s/generateLibraryList/genarateLibList/
Jed Brown
ExodusII: link Fortran if and only if using Fortran
Jed Brown
ExodusII: require HDF5 due to transitive dependency via --enable-netcdf-4
  1. Prev
  2. Next