On Mar 2, 2011, at 9:27 AM, Roy Stogner wrote:
>
> On Wed, 2 Mar 2011, Cody Permann wrote:
>
>>> point_locator(), and topological_neighbor(), and perhaps an expanding
>>> range of future functions, should not be used inside element loops
>>> unless the loop is preceded by an explicit PointLocat
On Wed, Mar 2, 2011 at 10:24 AM, Roy Stogner wrote:
>
> On Tue, 1 Mar 2011, David Andrs wrote:
>
>> Attached patch is improving 2 things in PETSc vector:
>> 1) improving the detection of PETSc vector type in a copy c-tor (current
>> check was ok for parallel runs, but not for serial runs)
>> 2) en
On Wed, 2 Mar 2011, Cody Permann wrote:
>> point_locator(), and topological_neighbor(), and perhaps an expanding
>> range of future functions, should not be used inside element loops
>> unless the loop is preceded by an explicit PointLocator construction.
>
> Yes - I did in fact run into this in
On Tue, Mar 1, 2011 at 4:14 PM, David Andrs wrote:
>
> Hi!
>
> Attached patch is improving 2 things in PETSc vector:
> 1) improving the detection of PETSc vector type in a copy c-tor (current
> check was ok for parallel runs, but not for serial runs)
> 2) enabling the assignment of parallel vector
On Mar 1, 2011, at 4:27 PM, Roy Stogner wrote:
>
> Take a look at revision 4214 to src/mesh/mesh_refinement.C
>
> It's the "implicit PointLocator construction fails" bug again, but
> this time with no threads involved. If a problem with periodic
> boundaries was run under certain conditions an
>> Can you try it again with Petsc 3 just to be sure? Sorry for the
>> inconvenience.
>
> Still works fine here, thanks.
Works for me with PETSC 3.0 and 3.1.
Sure looks weird having to put a const in there so you can pass its address
:-P
---
On Wed, 2 Mar 2011, John Peterson wrote:
> Can you try it again with Petsc 3 just to be sure? Sorry for the
> inconvenience.
Still works fine here, thanks.
---
Roy
--
Free Software Download: Index, Search & Analyze Lo
On Tue, 1 Mar 2011, David Andrs wrote:
> Attached patch is improving 2 things in PETSc vector:
> 1) improving the detection of PETSc vector type in a copy c-tor (current
> check was ok for parallel runs, but not for serial runs)
> 2) enabling the assignment of parallel vector into ghosted vector
On Wed, 2 Mar 2011, Boyce Griffith wrote:
> On 3/1/11 6:41 PM, Roy Stogner wrote:
>>
>> On Thu, 2 Dec 2010, Boyce Griffith wrote:
>>
>>> The patched code does the "right thing" only when
>>> asymmetric_constraint_rows=true when calling
>>> DofMap::constrain_element_{matrix,vector,matrix_and_vec
Hi!
Attached patch is improving 2 things in PETSc vector:
1) improving the detection of PETSc vector type in a copy c-tor (current
check was ok for parallel runs, but not for serial runs)
2) enabling the assignment of parallel vector into ghosted vector
Please review and check in, if ok.
Thanks
Take a look at revision 4214 to src/mesh/mesh_refinement.C
It's the "implicit PointLocator construction fails" bug again, but
this time with no threads involved. If a problem with periodic
boundaries was run under certain conditions and on the right
partitioning, then some of the MPI ranks (thos
On Tue, 1 Mar 2011, Johannes Huber wrote:
> Hi Roy,
>> Anyway, Tet10::has_affine_map() should be fixed in svn now. I'm
>> tempted not to bother releasing a 0.7.0.5 version with the bugfix,
>> since false negatives in has_affine_map() should only mess up
>> optimization and not cause any incorrec
12 matches
Mail list logo
