[petsc-users] MatView with relatively large matrix in Parallel

[Barry Smith]

   There is currently no fast binary view for parallel BAIJ matrices (when we wrote this stuff originally we didn't think people would ever be saving big sparse matrices). You can do MatConvert() to MPIAIJ and then MatView().  Don't worry the conversion won't be a very noticable time sink.


   Barry

On Sep 21, 2010, at 9:30 AM, Gaetan Kenway wrote:

> Hello
> 
> I am a PETSc user and have run into a problem using MatView. I am trying to output a matrix to file so I can load it instead of computing it for faster debugging. The matrix I'm trying to output is drdwt. It is a parallel block aij matrix with block size of 5. The matrix is assembled and the following code works when I run it in serial:
> 
>  call PetscViewerBinaryOpen(sumb_petsc_comm_world,drdw_name,FILE_MODE_WRITE,bin_viewer,ierr)
>  call MatView(drdwt,bin_viewer,ierr)
>  call PetscViewerDestroy(bin_viewer,ierr)
> 
> The matrix size is approximately 300k by 300k and I get an output file that is approximately 245MB in size which is expected.  However, when I run the same code in parallel on 3 processors it hangs at the MatView call until I am forced to kill the processes. I've let it go for 20 minutes with no sign of stopping.
> 
> I am not sure what is causing this.  I'm using openmpi-1.4.1  and petsc3.1 on 32 bit Ubuntu 10.10.
> 
> Thank you,
> 
> Gaetan Kenway
> 
>