Add nompi stub #425
Add nompi stub #425
Conversation
|
@cwsmith, build is failing for this PR. I think we should add |
|
@flagdanger and I are working on a very large update of PCU that should make this all easier by getting rid of the singleton. Pull request for that coming within the next week. |
|
@bobpaw sorting out a few things to make the pull request ready to merge, but #388 the draft pull request for replacing the singleton in PCU. Although we replaced the interface used in the test cases, this pull request is designed to work with no modification to existing codes by using a global PCU state when the old style interface is used with |
|
@jacobmerson, we will plan to use your PR for the future of core, but the sponsor wants this functionality soon, so we will use this branch until yours is finished. |
|
@bobpaw Does |
No, at the moment it only checks if PCU has been initialized using a global state variable. We could check with |
|
@bobpaw OK. I'm wondering if it will make sense to make the MPI check in that call, and if it fails call MPI_Init. But... handling MPI_Finalize could be a bit more tricky; other libraries may still be running and using MPI beyond the lifetime of PCU. I'd prefer symmetry in the interface so your original suggestion may be best. |
Added pcu_pnompi.c. Added SCOREC_NO_MPI CMake option. Removed direct MPI calls in parma, phasta, and PUMI. Added incomplete reimplementations of those MPI calls.
Add stubs for MPI_Init and MPI_Finalize. Add pcu_pnompi_types.h for MPI typedefs. Add status message about SCOREC_NO_MPI flag. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Replace MPI_Comm_free with PCU_Comm_Free_One. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
This will avoid object collisions when linking to libraries that already use MPI. - i.e. the intended use case for SCOREC_NO_MPI. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Changed add_nompi_msg to record receiver. - Since there's no MPI, sender == receiver but I want to quiet the warnings. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Add return type for pcu_pmpi_free, pcu_pmpi_split. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Add linked list traversal to find a matching message. Removed error message that did not actually indicate errors. - If no message is found it's perfectly reasonable and in fact correct to return 0 and that's how pmpi knows to return false. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Run smoke tests without mpirun given SCOREC_NO_MPI. Add SCOREC_NO_MPI to the GitHub Actions test matrix. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
I use clock_gettime, which is one of the many methods mpich uses depending on the system. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Also update TriBITS package file. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
- Add mpi_test_depends function to replace set_test_properties. This function checks if SCOREC_NO_MPI is defined and assumes nonexistent tests were multiproc ones. Signed-off-by: Aiden Woodruff <woodra@rpi.edu>
Add nompi stub
Add a stub implementation of MPI in
pcu/pcu_pnompi.c.Remove direct calls to MPI functions.
This adds a
SCOREC_NO_MPIbuild option for users running on single-process, single-machine setups that do not want to compile with MPI. Message passing is simulated in memory.