[petsc-users] Strange crash on VecScatter

Matthew Knepley knepley at gmail.com
Tue Nov 29 08:56:44 CST 2011


2011/11/29 Gong Ding <gdiso at ustc.edu>

> Hi,
> I meet some very strange problem on AIX6.1 with IBM POE.
>
> When I start the simulation with 3 or more processors, the following error
> will occure.
> Fatal Error:  at line 71 in
> /gpfs1/cogenda/cogenda/packages/petsc/include/../src/vec/vec/utils/vpscat.h
> One or two processors are ok.
>
> However, when I start my code with -vecscatter_window to avoid the
> MPI_Startall_irecv call of line 71, everything is ok.
>
> Some information:
> The petsc is configured with
> --known-level1-dcache-size=32768 --known-level1-dcache-linesize=32
> --known-level1-dcache-assoc=2 --known-memcmp-ok=1 --known-endian=big
> --known-sizeof-char=1
> --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-double=8 --known-sizeof-size_t=8 --known-bits-per-byte=8
> --known-sizeof-MPI_Comm=4 --known-sizeof-MPI_Fint=4
> --known-mpi-long-double=1 --download-f-blas-lapack=1
> --download-mumps=1 --download-blacs=1 --download-parmetis=1
> --download-scalapack=1 --download-superlu=1 --with-debugging=0
> --with-clanguage=cxx
> --with-cc=\"mpcc_r -q64\" --with-fc=\"mpxlf_r -q64\" --with-cxx=\"mpCC_r
> -q64\" --with-batch=1 --with-shared-libraries=1
> --known-mpi-shared-libraries=1
> --with-x=0 --with-pic=1
>
> And I add
> #ifndef PETSC_HAVE_MPI_MISSING_TYPESIZE
> #define PETSC_HAVE_MPI_MISSING_TYPESIZE
> #endif
> to petscconf.h for avoiding some mpi compile problem.
>
> Does anyone meet this problem?
>
> If I have to use this workaround, how to I add -vecscatter_window in the
> code instead in command line?


PetscOptionSetValue('-vecscatter_window', '1')

   Matt


>
> Gong Ding
>
>


-- 
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/20111129/fb7791c7/attachment.htm>


More information about the petsc-users mailing list