[petsc-dev] error reading binary file

Mark F. Adams mark.adams at columbia.edu
Tue Jun 12 17:05:00 CDT 2012


On Jun 12, 2012, at 5:52 PM, Jed Brown wrote:

> Okay, every matrix type except MPIAIJ respects -matload_block_size. Can you fix that?
> 
> 
> Can you fix that?

I'll take a shot at it Friday.  And I assume I need to add '-matload_block_size 3' to the info files.

Mark

> 
> On Tue, Jun 12, 2012 at 4:52 PM, Mark F. Adams <mark.adams at columbia.edu> wrote:
> Mine works in serial but not in parallel.
> 
> It looks like vecs are created in MatLoad via MatMPIAIJSetPreallocation_MPIAIJ and they are not split wrt blocks.  here is the stack trace again:
> 
> [0]PETSC ERROR: PetscSplitOwnership() line 93 in /Users/markadams/Codes/petsc-dev/src/sys/utils/psplit.c
> [0]PETSC ERROR: PetscLayoutSetUp() line 147 in /Users/markadams/Codes/petsc-dev/src/vec/vec/impls/mpi/pmap.c
> [0]PETSC ERROR: MatMPIAIJSetPreallocation_MPIAIJ() line 3194 in /Users/markadams/Codes/petsc-dev/src/mat/impls/aij/mpi/mpiaij.c
> [0]PETSC ERROR: MatMPIAIJSetPreallocation() line 3928 in /Users/markadams/Codes/petsc-dev/src/mat/impls/aij/mpi/mpiaij.c
> [0]PETSC ERROR: MatLoad_MPIAIJ() line 3442 in /Users/markadams/Codes/petsc-dev/src/mat/impls/aij/mpi/mpiaij.c
> [0]PETSC ERROR: MatLoad() line 881 in /Users/markadams/Codes/petsc-dev/src/mat/interface/matrix.c
> [0]PETSC ERROR: main() line 109 in src/ksp/ksp/examples/tutorials/ex10.c
> 
> 
> 
> On Jun 12, 2012, at 5:43 PM, Jed Brown wrote:
> 
>> On Tue, Jun 12, 2012 at 4:45 PM, Mark F. Adams <mark.adams at columbia.edu> wrote:
>> Satish beat me to it.  I still get the same error ... did you use -mat_block_size 3?
>> 
>> I will look in to further.
>> 
>> ~/petsc/src/ksp/ksp/examples/tutorials$ ./ex10 -f0 binaryoutput-sym -ksp_monitor -pc_type gamg -mat_block_size 3
>> [0]PCSetData_AGG bs=3 MM=89373
>>   0 KSP Residual norm 4.854898290598e-03 
>>   1 KSP Residual norm 1.748948970672e-03 
>>   2 KSP Residual norm 1.098105829371e-03 
>>   3 KSP Residual norm 7.267836867590e-04 
>>   4 KSP Residual norm 4.752091560022e-04 
>>   5 KSP Residual norm 3.152074302240e-04 
>>   6 KSP Residual norm 1.998139269819e-04 
>>   7 KSP Residual norm 1.142295077940e-04 
>>   8 KSP Residual norm 6.107885553654e-05 
>>   9 KSP Residual norm 2.903395117932e-05 
>>  10 KSP Residual norm 1.748667302607e-05 
>>  11 KSP Residual norm 9.864255584167e-06 
>>  12 KSP Residual norm 5.077178570024e-06 
>>  13 KSP Residual norm 2.996102603244e-06 
>>  14 KSP Residual norm 2.293954251541e-06 
>>  15 KSP Residual norm 1.977076000505e-06 
>>  16 KSP Residual norm 1.729762967197e-06 
>>  17 KSP Residual norm 1.422352992308e-06 
>>  18 KSP Residual norm 1.063366577331e-06 
>>  19 KSP Residual norm 7.479518697501e-07 
>>  20 KSP Residual norm 4.861597752713e-07 
>>  21 KSP Residual norm 2.996075502312e-07 
>>  22 KSP Residual norm 1.962056634115e-07 
>>  23 KSP Residual norm 1.216036063867e-07 
>>  24 KSP Residual norm 8.556036439634e-08 
>>  25 KSP Residual norm 5.911194128322e-08 
>>  26 KSP Residual norm 3.784597093891e-08 
>> Number of iterations =  26
>> Residual norm 7.51457e-10
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120612/061d3494/attachment.html>


More information about the petsc-dev mailing list