MPI has some asinine semantics with MPI_Ops being used differently with collectives and with one-sided. I pushed a fix yesterday. On Mon, Apr 23, 2012 at 08:46, Alexander Grayver <[email protected]>wrote:

On 23.04.2012 15:12, Jed Brown wrote:

...

My fault, now fixed:

http://petsc.cs.iit.edu/petsc/**petsc-dev/rev/8a997a5a65d3<http://petsc.cs.iit.edu/petsc/petsc-dev/rev/8a997a5a65d3>

Jed,

MatAXPY works fine, although during tests there was crash:

Possible error running C/C++ src/snes/examples/tutorials/**ex19 with 2 MPI processes See http://www.mcs.anl.gov/petsc/**documentation/faq.html<http://www.mcs.anl.gov/petsc/documentation/faq.html> lid velocity = 0.0016, prandtl # = 1, grashof # = 1 [0]PETSC ERROR: PetscSplitReductionApply() line 328 in /home/lib/petsc-dev1/src/vec/**vec/utils/comb.c [0]PETSC ERROR: PetscSplitReductionEnd() line 247 in /home/lib/petsc-dev1/src/vec/**vec/utils/comb.c [0]PETSC ERROR: VecNormEnd() line 672 in /home/lib/petsc-dev1/src/vec/** vec/utils/comb.c [0]PETSC ERROR: SNESSolve_LS() line 172 in /home/lib/petsc-dev1/src/snes/* *impls/ls/ls.c [0]PETSC ERROR: SNESSolve() line 3458 in /home/lib/petsc-dev1/src/snes/** interface/snes.c [0]PETSC ERROR: main() line 151 in src/snes/examples/tutorials/**ex19.c [1]PETSC ERROR: PetscSplitReductionApply() line 328 in /home/lib/petsc-dev1/src/vec/**vec/utils/comb.c [1]PETSC ERROR: PetscSplitReductionEnd() line 247 in /home/lib/petsc-dev1/src/vec/**vec/utils/comb.c [1]PETSC ERROR: VecNormEnd() line 672 in /home/lib/petsc-dev1/src/vec/** vec/utils/comb.c [1]PETSC ERROR: SNESSolve_LS() line 172 in /home/lib/petsc-dev1/src/snes/* *impls/ls/ls.c [1]PETSC ERROR: SNESSolve() line 3458 in /home/lib/petsc-dev1/src/snes/** interface/snes.c [1]PETSC ERROR: main() line 151 in src/snes/examples/tutorials/**ex19.c

-- Regards, Alexander