Master release 4.4.01 (#2337)

* Set version number to 4.4.01 * [4.4.01] Patches to 4.4.01 (#2327) * Restore size_t as default offset, in Tribits builds (#2313) If building KokkosKernels standalone, leave int as the default offset (this was the case since #2140). But if building KokkosKernels as a Trilinos/Tribits package, then make size_t the default offset because this is what the Tpetra stack currently uses. Signed-off-by: Brian Kelley <bmkelle@sandia.gov> * Improve crs/bsr sorting performance (#2293) * CRS sorting improvements - Wrote bulk sort/permutation based sorting for CRS graph, matrix, and BSR matrix (bulk = one large sort of all the entries, using row-major dense index as keys) - This is more performant for imbalanced entries per row - If matrix dimensions are too large to do bulk sort, fall back to sorting within each row with a thread. * Add perf test for sort_crs_matrix * sort_crs: improve parallel labels * Work around kokkos issue 7036 * sort_crs: replace radix sort lambda with functor (Lambda segfaults with nvcc+openmp) --------- Signed-off-by: Brian Kelley <bmkelle@sandia.gov> * SpAdd handle: delete sort_option getter/setter (#2296) SpAdd handle was originally a copy-paste of the spgemm handle way back in #122, and included get_sort_option() and set_sort_option() from spgemm. But these try to use the member bool sort_option, which doesn't exist. Somehow these functions never produced compile errors until someone tried to call them. * Improve GH action to produce release artifacts (#2312) * coo2csr: add parens to function calls (#2318) * Update changelog * Update master_history.txt * .github/workflows: Group jobs under common github-AT2 name (#2320) * Update master_history.txt --------- Signed-off-by: Brian Kelley <bmkelle@sandia.gov> Co-authored-by: brian-kelley <bmkelle@sandia.gov> Co-authored-by: Damien L-G <dalg24@gmail.com> Co-authored-by: Carl Pearson <cwpearson@users.noreply.github.com> Co-authored-by: Evan Harvey <57234914+e10harvey@users.noreply.github.com> * Update master_history.txt for 4.4.01 --------- Signed-off-by: Brian Kelley <bmkelle@sandia.gov> Co-authored-by: brian-kelley <bmkelle@sandia.gov> Co-authored-by: Damien L-G <dalg24@gmail.com> Co-authored-by: Carl Pearson <cwpearson@users.noreply.github.com> Co-authored-by: Evan Harvey <57234914+e10harvey@users.noreply.github.com>
kokkos · Sep 13, 2024 · 0b43169 · 0b43169
1 parent 0608a33
commit 0b43169
Show file tree

Hide file tree

Showing 19 changed files with 849 additions and 595 deletions.
diff --git a/.github/workflows/at2.yml b/.github/workflows/at2.yml
@@ -0,0 +1,29 @@
+name: github-AT2
+
+on: 
+  pull_request:
+    paths-ignore:
+    - '**/*.rst'
+    - '**/*.md'
+    - '**/requirements.txt'
+    - '**/*.py'
+    - 'docs/**'
+    types: [ opened, reopened, synchronize ]
+
+permissions:
+  contents: none
+
+# Cancels any in progress 'workflow' associated with this PR
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  mi210:
+    uses: ./.github/workflows/mi210.yml
+  h100:
+    uses: ./.github/workflows/h100.yml
+  bdw:
+    uses: ./.github/workflows/bdw.yml
+  #spr:
+    #uses: ./.github/workflows/spr.yml
diff --git a/.github/workflows/bdw.yml b/.github/workflows/bdw.yml
@@ -1,25 +1,7 @@
-name: github-BDW
+name: Reusable BDW workflow
 
 on:
-  pull_request:
-    paths-ignore:
-    - '**/*.rst'
-    - '**/*.md'
-    - '**/requirements.txt'
-    - '**/*.py'
-    - 'docs/**'
-    types: [ opened, reopened, synchronize ]
-  pull_request_review:
-    types:
-      - submitted
-
-permissions:
-  contents: none
-
-# Cancels any in progress 'workflow' associated with this PR
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
+  workflow_call
 
 jobs:
 #  PR_BDW_GNU1020_OPENMP_LEFT_REL_NOETI:

diff --git a/.github/workflows/h100.yml b/.github/workflows/h100.yml
@@ -1,26 +1,7 @@
-name: github-H100
+name: Reusable H100 workflow
 
-# Only allow manual runs until at2 runners are available.
 on:
-  pull_request:
-    paths-ignore:
-    - '**/*.rst'
-    - '**/*.md'
-    - '**/requirements.txt'
-    - '**/*.py'
-    - 'docs/**'
-    types: [ opened, reopened, synchronize ]
-  pull_request_review:
-    types:
-      - submitted
-
-permissions:
-  contents: none
-
-# Cancels any in progress 'workflow' associated with this PR
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
+  workflow_call
 
 jobs:
   PR_HOPPER90_CUDA1180_CUDA_LEFT_RIGHT_REL:

diff --git a/.github/workflows/mi210.yml b/.github/workflows/mi210.yml
@@ -1,25 +1,7 @@
-name: github-MI210
+name: Reusable MI210 workflow
 
-on: 
-  pull_request:
-    paths-ignore:
-    - '**/*.rst'
-    - '**/*.md'
-    - '**/requirements.txt'
-    - '**/*.py'
-    - 'docs/**'
-    types: [ opened, reopened, synchronize ]
-  pull_request_review:
-    types:
-      - submitted
-
-permissions:
-  contents: none
-
-# Cancels any in progress 'workflow' associated with this PR
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
+on:
+  workflow_call
 
 jobs:
 #  PR_VEGA90A_ROCM561_HIP_SERIAL_LEFT_REL:

diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
@@ -13,33 +13,26 @@ jobs:
       hashes: ${{ steps.hash.outputs.hashes }}
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
       - name: Build artifacts
         run: |
-          git archive -o kokkos-kernels-${{ github.ref_name }}.zip HEAD
-          git archive -o kokkos-kernels-${{ github.ref_name }}.tar.gz HEAD
+          git archive --prefix=kokkos-kernels-${{ github.ref_name }}/ -o kokkos-kernels-${{ github.ref_name }}.zip HEAD
+          git archive --prefix=kokkos-kernels-${{ github.ref_name }}/ -o kokkos-kernels-${{ github.ref_name }}.tar.gz HEAD
 
       - name: Generate hashes
         shell: bash
         id: hash
         run: |
           # sha256sum generates sha256 hash for all artifacts.
           # base64 -w0 encodes to base64 and outputs on a single line.
-          echo "hashes=$(sha256sum kokkos-kernels-${{ github.ref_name }}.zip kokkos-kernels-${{ github.ref_name }}.tar.gz | base64 -w0)" >> "$GITHUB_OUTPUT"
+          sha256sum kokkos-kernels-${{ github.ref_name }}.zip kokkos-kernels-${{ github.ref_name }}.tar.gz > kokkos-kernels-${{ github.ref_name }}-SHA-256.txt
+          echo "hashes=$(base64 -w0 kokkos-kernels-${{ github.ref_name }}-SHA-256.txt)" >> "$GITHUB_OUTPUT"
 
-      - name: Upload source code (zip)
-        uses: actions/upload-artifact@89ef406dd8d7e03cfd12d9e0a4a378f454709029 # v4.3.5
+      - name: Upload artifacts
+        uses: actions/upload-artifact@50769540e7f4bd5e21e526ee35c689e35e0d6874 # v4.4.0
         with:
-          name: kokkos-kernels-${{ github.ref_name }}.zip
-          path: kokkos-kernels-${{ github.ref_name }}.zip
-          if-no-files-found: error
-          retention-days: 5
-
-      - name: Upload source code (tar.gz)
-        uses: actions/upload-artifact@89ef406dd8d7e03cfd12d9e0a4a378f454709029 # v4.3.5
-        with:
-          name: kokkos-kernels-${{ github.ref_name }}.tar.gz
-          path: kokkos-kernels-${{ github.ref_name }}.tar.gz
+          name: release-artifacts
+          path: kokkos-kernels-${{ github.ref_name }}*
           if-no-files-found: error
           retention-days: 5
 
@@ -65,19 +58,14 @@ jobs:
     runs-on: ubuntu-latest
     if: startsWith(github.ref, 'refs/tags/')
     steps:
-      - name: Download kokkos-kernels-${{ github.ref_name }}.zip
+      - name: Download artifacts
         uses: actions/download-artifact@fa0a91b85d4f404e444e00e005971372dc801d16 # v4.1.8
         with:
-          name: kokkos-kernels-${{ github.ref_name }}.zip
-
-      - name: Download kokkos-kernels-${{ github.ref_name }}.tar.gz
-        uses: actions/download-artifact@fa0a91b85d4f404e444e00e005971372dc801d16 # v4.1.8
-        with:
-          name: kokkos-kernels-${{ github.ref_name }}.tar.gz
-
+          name: release-artifacts
       - name: Upload assets
         uses: softprops/action-gh-release@c062e08bd532815e2082a85e87e3ef29c3e6d191 # v2.0.8
         with:
           files: |
             kokkos-kernels-${{ github.ref_name }}.zip
             kokkos-kernels-${{ github.ref_name }}.tar.gz
+            kokkos-kernels-${{ github.ref_name }}-SHA-256.txt
diff --git a/.github/workflows/spr.yml b/.github/workflows/spr.yml
@@ -1,26 +1,7 @@
-name: github-SPR
+name: Reusable SPR workflow
 
-# Only allow manual runs until at2 runners are available.
-on: workflow_dispatch
-  #pull_request:
-  #  paths-ignore:
-  #  - '**/*.rst'
-  #  - '**/*.md'
-  #  - '**/requirements.txt'
-  #  - '**/*.py'
-  #  - 'docs/**'
-  #  types: [ opened, reopened, synchronize ]
-  #pull_request_review:
-  #  types:
-  #    - submitted
-
-permissions:
-  contents: none
-
-# Cancels any in progress 'workflow' associated with this PR
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
+on:
+  workflow_call
 
 jobs:
   PR_SPR_ONEAPI202310_OPENMP_LEFT_MKLBLAS_MKLLAPACK_REL:

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,19 @@
 # Change Log
 
+## [4.4.01](https://github.com/kokkos/kokkos-kernels/tree/4.4.01)
+[Full Changelog](https://github.com/kokkos/kokkos-kernels/compare/4.4.00...4.4.01)
+
+### Build System:
+- Restore size_t as default offset, in Tribits builds [\#2313](https://github.com/kokkos/kokkos-kernels/pull/2313)
+
+### Enhancements:
+- Improve crs/bsr sorting performance [\#2293](https://github.com/kokkos/kokkos-kernels/pull/2293)
+
+### Bug Fixes:
+- SpAdd handle: delete sort_option getter/setter [\#2296](https://github.com/kokkos/kokkos-kernels/pull/2296)
+- Improve GH action to produce release artifacts [\#2312](https://github.com/kokkos/kokkos-kernels/pull/2312)
+- coo2csr: add parens to function calls [\#2318](https://github.com/kokkos/kokkos-kernels/pull/2318)
+
 ## [4.4.00](https://github.com/kokkos/kokkos-kernels/tree/4.4.00)
 [Full Changelog](https://github.com/kokkos/kokkos-kernels/compare/4.3.01...4.4.00)
 

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -11,7 +11,7 @@ SET(KOKKOSKERNELS_TOP_SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR})
 
 SET(KokkosKernels_VERSION_MAJOR 4)
 SET(KokkosKernels_VERSION_MINOR 4)
-SET(KokkosKernels_VERSION_PATCH 0)
+SET(KokkosKernels_VERSION_PATCH 1)
 SET(KokkosKernels_VERSION "${KokkosKernels_VERSION_MAJOR}.${KokkosKernels_VERSION_MINOR}.${KokkosKernels_VERSION_PATCH}")
 
 #Set variables for config file

diff --git a/cmake/kokkoskernels_eti_offsets.cmake b/cmake/kokkoskernels_eti_offsets.cmake
@@ -1,5 +1,15 @@
-SET(KOKKOSKERNELS_INST_OFFSET_SIZE_T_DEFAULT OFF)
-SET(KOKKOSKERNELS_INST_OFFSET_INT_DEFAULT ${KOKKOSKERNELS_ADD_DEFAULT_ETI})
+IF(KOKKOSKERNELS_HAS_TRILINOS)
+  # In a Trilinos build, size_t is the default offset because this is what Tpetra uses
+  # TODO: update this when Tpetra can use different offsets
+  SET(KOKKOSKERNELS_INST_OFFSET_SIZE_T_DEFAULT ${KOKKOSKERNELS_ADD_DEFAULT_ETI})
+  SET(KOKKOSKERNELS_INST_OFFSET_INT_DEFAULT OFF)
+ELSE()
+  # But in a standalone KokkosKernels build, int is the default offset type
+  # This provides the maximum TPL compatibility
+  SET(KOKKOSKERNELS_INST_OFFSET_SIZE_T_DEFAULT OFF)
+  SET(KOKKOSKERNELS_INST_OFFSET_INT_DEFAULT ${KOKKOSKERNELS_ADD_DEFAULT_ETI})
+ENDIF()
+
 SET(OFFSETS
   OFFSET_INT
   OFFSET_SIZE_T
@@ -12,14 +22,14 @@ KOKKOSKERNELS_ADD_OPTION(
   INST_OFFSET_INT
   ${KOKKOSKERNELS_INST_OFFSET_INT_DEFAULT}
   BOOL
-  "Whether to pre instantiate kernels for the offset type int.  This option is KokkosKernels_INST_OFFSET_INT=OFF by default. Default: OFF"
+  "Whether to pre instantiate kernels for the offset type int.  This option is KokkosKernels_INST_OFFSET_INT=OFF by default. Default: ${KOKKOSKERNELS_INST_OFFSET_INT_DEFAULT}"
   )
 
 KOKKOSKERNELS_ADD_OPTION(
   INST_OFFSET_SIZE_T
   ${KOKKOSKERNELS_INST_OFFSET_SIZE_T_DEFAULT}
   BOOL
-  "Whether to pre instantiate kernels for the offset type size_t.  This option is KokkosKernels_INST_OFFSET_SIZE_T=ON by default. Default: ON"
+  "Whether to pre instantiate kernels for the offset type size_t.  This option is KokkosKernels_INST_OFFSET_SIZE_T=ON by default. Default: ${KOKKOSKERNELS_INST_OFFSET_SIZE_T_DEFAULT}"
   )
 
 IF (KOKKOSKERNELS_INST_OFFSET_INT)

diff --git a/common/src/KokkosKernels_SimpleUtils.hpp b/common/src/KokkosKernels_SimpleUtils.hpp
@@ -358,13 +358,19 @@ struct ReduceMaxFunctor {
 };
 
 template <typename view_type, typename MyExecSpace>
-void kk_view_reduce_max(size_t num_elements, view_type view_to_reduce,
+void kk_view_reduce_max(const MyExecSpace &exec, size_t num_elements, view_type view_to_reduce,
                         typename view_type::non_const_value_type &max_reduction) {
-  typedef Kokkos::RangePolicy<MyExecSpace> my_exec_space;
-  Kokkos::parallel_reduce("KokkosKernels::Common::ReduceMax", my_exec_space(0, num_elements),
+  typedef Kokkos::RangePolicy<MyExecSpace> policy_t;
+  Kokkos::parallel_reduce("KokkosKernels::Common::ReduceMax", policy_t(exec, 0, num_elements),
                           ReduceMaxFunctor<view_type>(view_to_reduce), max_reduction);
 }
 
+template <typename view_type, typename MyExecSpace>
+void kk_view_reduce_max(size_t num_elements, view_type view_to_reduce,
+                        typename view_type::non_const_value_type &max_reduction) {
+  kk_view_reduce_max(MyExecSpace(), num_elements, view_to_reduce, max_reduction);
+}
+
 // xorshift hash/pseudorandom function (supported for 32- and 64-bit integer
 // types only)
 template <typename Value>
@@ -429,10 +435,14 @@ struct SequentialFillFunctor {
   val_type start;
 };
 
+template <typename ExecSpace, typename V>
+void sequential_fill(const ExecSpace &exec, const V &v, typename V::non_const_value_type start = 0) {
+  Kokkos::parallel_for(Kokkos::RangePolicy<ExecSpace>(exec, 0, v.extent(0)), SequentialFillFunctor<V>(v, start));
+}
+
 template <typename V>
 void sequential_fill(const V &v, typename V::non_const_value_type start = 0) {
-  Kokkos::parallel_for(Kokkos::RangePolicy<typename V::execution_space>(0, v.extent(0)),
-                       SequentialFillFunctor<V>(v, start));
+  sequential_fill(typename V::execution_space(), v, start);
 }
 
 }  // namespace Impl

diff --git a/common/src/KokkosKernels_Utils.hpp b/common/src/KokkosKernels_Utils.hpp
@@ -1076,6 +1076,12 @@ void view_reduce_max(size_t num_elements, view_type view_to_reduce,
   kk_view_reduce_max<view_type, MyExecSpace>(num_elements, view_to_reduce, max_reduction);
 }
 
+template <typename view_type, typename MyExecSpace>
+void view_reduce_max(const MyExecSpace &exec, size_t num_elements, view_type view_to_reduce,
+                     typename view_type::non_const_value_type &max_reduction) {
+  kk_view_reduce_max<view_type, MyExecSpace>(exec, num_elements, view_to_reduce, max_reduction);
+}
+
 template <typename size_type>
 struct ReduceRowSizeFunctor {
   const size_type *rowmap_view_begins;

diff --git a/master_history.txt b/master_history.txt
@@ -27,3 +27,4 @@ tag: 4.2.01     date: 01/30/2024  master: f429f6ec    release: bcf9854b
 tag: 4.3.00     date: 04/03/2024  master: afd65f03    release: ebbf4b78
 tag: 4.3.01     date: 05/07/2024  master: 1b0a15f5    release: 58785c1b
 tag: 4.4.00     date: 08/08/2024  master: d1a91b8a    release: 1145f529
+tag: 4.4.01     date: 09/12/2024  master: 0608a337    release: 6b340287
diff --git a/perf_test/sparse/CMakeLists.txt b/perf_test/sparse/CMakeLists.txt
@@ -116,6 +116,15 @@ KOKKOSKERNELS_ADD_EXECUTABLE(
         SOURCES KokkosSparse_mdf.cpp
 )
 
+# For the sake of build times, don't build this CRS sorting perf test by default.
+# It can be enabled if needed by setting -DKokkosKernels_ENABLE_SORT_CRS_PERFTEST=ON.
+if (KokkosKernels_ENABLE_SORT_CRS_PERFTEST)
+  KOKKOSKERNELS_ADD_EXECUTABLE(
+    sparse_sort_crs
+    SOURCES KokkosSparse_sort_crs.cpp
+)
+endif ()
+
 if (KokkosKernels_ENABLE_BENCHMARK)
   KOKKOSKERNELS_ADD_BENCHMARK(
     sparse_par_ilut