Merge branch 'jczhang/2025-03-05/revise-PetscGetKokkosExecutionSpace' into 'main'

Return execution space instead of reference to simplify the code

See merge request petsc/petsc!8182

Kokkos: return execution space instead of reference to simplify the code

A Kokkos execution space is a shared pointer. We don't need to reference it

Merge branch 'jczhang/2025-01-30/feature-support-AMD-MI300A' into 'main'

Add support of AMD MI300A

Closes #1703

See merge request petsc/petsc!8110

Kokkos: add support of AMD MI300A

* Use HostMirrorMemorySpace instead of HostSpace to fix compile errors on MI300A

* Replace Kokkos::HostSpace with HostMirrorMemorySpace to fix compile errors on MI

Kokkos: add support of AMD MI300A

* Use HostMirrorMemorySpace instead of HostSpace to fix compile errors on MI300A

* Replace Kokkos::HostSpace with HostMirrorMemorySpace to fix compile errors on MI300A, since the latter is what Kokkos::DualView use for its host view

* Fix a subtle bug in KokkosDualViewSync() w.r.t MI300A. Suppose we want to sync a petsc VecKokkos v on host. On MI300A, the host copy v_h and the device copy v_d share the memory. So in the old code, we used if (v_dual.need_sync_host()) to skip the device to host memory copy. But I should not skip the exec.fence(). As the device might still have kernels writing v_d, we still need to sync the device/stream to make v_d ready for use on CPU (via v_h).

show more ...

Merge branch 'jczhang/2024-11-17/update-kokkos-4.5' into 'main'

Update to Kokkos to 4.5.00

See merge request petsc/petsc!8013

Update to Kokkos to 4.5.00

Merge branch 'jczhang/2024-09-09/fix-sf-mpi-large-count-datatype' into 'main'

PetscSF: Support MPI derived data types created with large count

Closes #1625

See merge request petsc/petsc!7825

PetscSF: support MPI derived data types created with large count

When user created MPI derived data types with large count routines, e.g., MPI_Type_contiguous_c(4, MPI_DOUBLE, &newtype), and passed

PetscSF: support MPI derived data types created with large count

When user created MPI derived data types with large count routines, e.g., MPI_Type_contiguous_c(4, MPI_DOUBLE, &newtype), and passed newtype to PetscSF,
is it legal to decode the datatype with MPI_Type_get_envelope()? The MPI Forum debated but has yet to resolve this issue.

Our approach is to decode the datatype with MPI_Type_get_envelope_c() if available; otherwise with MPI_Type_get_envelope().

But currently we don't really support real big count values, say a type created by MPI_Type_contiguous_c(2^33, MPI_DOUBLE, &newtype).

We could support that, but just see no needs for now. We will error out if this kind of types are given to PetscSF.

show more ...

Merge branch 'tisaac/bugfix-sfwindow' into 'main'

Add PetscSFRegisterPersistent() and fix PETSCSFWINDOW for empty ranks

See merge request petsc/petsc!7786

PetscSF: add PetscSFGetDatatypeSize_Internal() for internal use

Merge branch 'jolivet/clang-format-18' into 'main'

clang-format version 18

See merge request petsc/petsc!6902

Remove some unneeded parentheses

checkbadSource: rules for PetscFunctionBegin and derivatives

Merge branch 'jczhang/2023-08-14/make-device2host-memcopy-async-safe' into 'main'

Make DualView sync_host() asynchronous by adding an execution space argument

See merge request petsc/petsc!6811

Kokkos: make dual view sync_host() asynchronous by adding an execution space argument

sync_host() without the argument is synchronous (like cudaMemcpy D2H). With the argument,
it is asynchronous (l

Kokkos: make dual view sync_host() asynchronous by adding an execution space argument

sync_host() without the argument is synchronous (like cudaMemcpy D2H). With the argument,
it is asynchronous (like cudaMemcpyAsync D2H).

To make sure the host array contains valide data immediately after dualview.sync_host(),
we need to do exec.fence();

show more ...

Merge remote-tracking branch 'origin/release'

Merge branch 'jczhang/2023-04-11/fix-sfkok-stream/release' into 'release'

Let SF Kokkos backend use an execution space made from PetscDefaultCuda/HipStream

See merge request petsc/petsc!6317

Let SF Kokkos backend use an execution space made from PetscDefaultCuda/HipStream

Merge branch 'jacobf/2022-12-10/petscerrorcode-nodiscard' into 'main'

Feature: Non-discardable PetscErrorCode

See merge request petsc/petsc!5923

Make PetscErrorCode a non-discardable enum

Merge branch 'jacobf/2022-09-21/2-bike-2-shed' into 'main'

Feature: Bicycle Storage Facility 2

See merge request petsc/petsc!5661

source code format changes due to .clang-format changes

Merge branch 'barry/2022-08-21/clang-format-source' into 'main'

format repository with clang-format

See merge request petsc/petsc!5541

clang-format: convert PETSc sources to comply with clang-format

Merge branch 'jolivet/feature-less-checkfalse' into 'main'

Dividing by four the number of PetscCheckFalse()

See merge request petsc/petsc!5072