In implementing Vec in the GPU a VecScatter would take the entries
directly from GPU copy of the
vector and put them appropriately into the GPU copy of the other
vector, thus it would be completely
transparent from the MatMult_MPIAIJ() code. In other words, if you
have to modify the MatM
Ahmed,
If you take a look in
src/mat/impls/aij/mpi/
at mpiaij.h and mpiaij.c, you can see that an MPIAIJ matrix is stored locally
as two "sequential" matrics, A and B. A contains the "diagonal" portion of
the matrix, and B contains the "off-diagonal" portion. See page 59 of the
PETSc Use
On Thu, Sep 18, 2008 at 1:34 PM, Ahmed El Zein wrote:
> On Thu, 2008-09-18 at 10:20 -0500, Matthew Knepley wrote:
>> > A question that I had regarding the PETSc code when I was thinking
>> about
>> > this was:
>> > You have the SeqAIJ matrix type and the the MPIAIJ type built around
>> it
>> > (or
On Wed, 2008-09-17 at 21:33 -0500, Barry Smith wrote:
> Ahmed,
>
> This is very cool.
> On Sep 17, 2008, at 8:05 PM, Ahmed El Zein wrote:
>
> > Harald,
> > I am working on implementing SpMV on a NVIDIA GPU with CUDA for SML
> > applications (just about finished actually).
> >
> > As the SML a
Harald,
I am working on implementing SpMV on a NVIDIA GPU with CUDA for SML
applications (just about finished actually).
As the SML application I am planing to base my work on uses PETSc, I
have written some functions that will convert AIJ matrices and Vectors
to SP and copy them to the GPU, multi
On Wed, Sep 17, 2008 at 10:51 PM, Ahmed El Zein wrote:
> On Wed, 2008-09-17 at 21:33 -0500, Barry Smith wrote:
>> Ahmed,
>>
>> This is very cool.
>> On Sep 17, 2008, at 8:05 PM, Ahmed El Zein wrote:
>>
>> > Harald,
>> > I am working on implementing SpMV on a NVIDIA GPU with CUDA for SML
>> > a