Convert device-aware MPI implementation for LinearAlgebra::distributed::Vector #14571
Conversation
|
/rebuild |
masterleinad
force-pushed
the
kokkos_la_d_vector_device_aware_mpi
branch
from
49c65ac
to
340ec56
Compare
masterleinad
force-pushed
the
kokkos_la_d_vector_device_aware_mpi
branch
from
1a53a05
to
10e4a84
Compare
masterleinad
changed the title
|
Can you rebase? |
What would rebasing change? |
|
I should have been more specific - my intention is that we should squash some commits. |
masterleinad
force-pushed
the
kokkos_la_d_vector_device_aware_mpi
branch
from
d74fa6e
to
ac60c6e
Compare
Ah, sure, here you go! |
| @@ -65,8 +65,12 @@ macro(feature_mpi_configure_external) | |||
| # (in Modules/FindMPI.cmake) at some point. For the time being this is an | |||
| # advanced configuration option. | |||
| # | |||
| option(DEAL_II_MPI_WITH_CUDA_SUPPORT "Enable MPI Cuda support" OFF) | |||
| mark_as_advanced(DEAL_II_MPI_WITH_CUDA_SUPPORT) | |||
| if(DEAL_II_MPI_WITH_CUDA_SUPPORT) | |||
There was a problem hiding this comment.
Where does DEAL_II_MPI_WITH_CUDA_SUPPORT get set? AFAICT, based on the changes to cuda.html, we would not expect this to be set by a user any more.
There was a problem hiding this comment.
Only for backward-compatibility.
| @@ -448,7 +448,10 @@ | |||
| # define DEAL_II_MPI_VERSION_GTE(major,minor) false | |||
| #endif | |||
|
|
|||
| #cmakedefine DEAL_II_MPI_WITH_DEVICE_SUPPORT | |||
| #ifdef DEAL_II_MPI_WITH_DEVICE_SUPPORT | |||
| #cmakedefine DEAL_II_MPI_WITH_CUDA_SUPPORT | |||
There was a problem hiding this comment.
Is this now just for backwards compatibility? I don't see anywhere in the library where it is used any more.
There was a problem hiding this comment.
Only for backward-compatibility.
|
Anyone who wanted to look at this PR has had enough time to do it. So I am merging it. |
Based on top of #14537.
- mpi/parallel_partitioner_device_06.mpirun=4.debug (Failed)- mpi/parallel_partitioner_device_07.mpirun=4.debug (Failed)are known to fail.Also, this needs work for running on GPUs.