kokkos · lucbv · Apr 8, 2024 · Oct 17, 2023 · Oct 17, 2023 · Oct 19, 2023
diff --git a/.github/workflows/docs.yml b/.github/workflows/docs.yml
@@ -23,12 +23,12 @@ jobs:
           doxygen --version
 
       - name: checkout_kokkos_kernels
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: kokkos-kernels
 
       - name: checkout_kokkos
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           repository: kokkos/kokkos
           ref: develop

diff --git a/.github/workflows/format.yml b/.github/workflows/format.yml
@@ -13,7 +13,7 @@ jobs:
   clang-format-check:
     runs-on: ubuntu-20.04
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Install Dependencies
         run: sudo apt install clang-format-8

diff --git a/.github/workflows/osx.yml b/.github/workflows/osx.yml
@@ -50,12 +50,12 @@ jobs:
 
     steps:
       - name: checkout_kokkos_kernels
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: kokkos-kernels
 
       - name: checkout_kokkos
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           repository: kokkos/kokkos
           ref: ${{ github.base_ref }}

diff --git a/.gitignore b/.gitignore
@@ -12,4 +12,7 @@ TAGS
 #Clangd indexing
 compile_commands.json
 .cache/
-.vscode/
+.vscode/
+
+#MacOS hidden files
+.DS_Store
diff --git a/.readthedocs.yaml b/.readthedocs.yaml
@@ -0,0 +1,35 @@
+# Read the Docs configuration file for Sphinx projects
+# See https://docs.readthedocs.io/en/stable/config-file/v2.html for details
+
+# Required
+version: 2
+
+# Set the OS, Python version and other tools you might need
+build:
+  os: ubuntu-22.04
+  tools:
+    python: "3.12"
+    # You can also specify other tool versions:
+    # nodejs: "20"
+    # rust: "1.70"
+    # golang: "1.20"
+
+# Build documentation in the "docs/" directory with Sphinx
+sphinx:
+  configuration: docs/conf.py
+  # You can configure Sphinx to use a different builder, for instance use the dirhtml builder for simpler URLs
+  # builder: "dirhtml"
+  # Fail on all warnings to avoid broken references
+  # fail_on_warning: true
+
+# Optionally build your docs in additional formats such as PDF and ePub
+# formats:
+#   - pdf
+#   - epub
+
+# Optional but recommended, declare the Python requirements required
+# to build your documentation
+# See https://docs.readthedocs.io/en/stable/guides/reproducible-builds.html
+python:
+ install:
+   - requirements: docs/requirements.txt
diff --git a/BUILD.md b/BUILD.md
@@ -227,7 +227,7 @@ endif()
 * KokkosKernels_LAPACK_ROOT: PATH
   * Location of LAPACK install root.
   * Default: None or the value of the environment variable LAPACK_ROOT if set
-* KokkosKernels_LINALG_OPT_LEVEL: BOOL
+* KokkosKernels_LINALG_OPT_LEVEL: BOOL **DEPRECATED**
   * Optimization level for KokkosKernels computational kernels: a nonnegative integer.  Higher levels result in better performance that is more uniform for corner cases, but increase build time and library size.  The default value is 1, which should give performance within ten percent of optimal on most platforms, for most problems.
   * Default: 1
 * KokkosKernels_MAGMA_ROOT: PATH

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,99 @@
 # Change Log
 
+## [4.3.00](https://github.com/kokkos/kokkos-kernels/tree/4.3.00) (2024-03-19)
+[Full Changelog](https://github.com/kokkos/kokkos-kernels/compare/4.2.01...4.3.00)
+
+### New Features
+
+#### BLAS updates
+- Syr2 [\#1942](https://github.com/kokkos/kokkos-kernels/pull/1942)
+
+#### LAPACK updates
+- Adding cuSOLVER [\#2038](https://github.com/kokkos/kokkos-kernels/pull/2038)
+  - Fix for MAGMA with CUDA [\#2044](https://github.com/kokkos/kokkos-kernels/pull/2044)
+- Adding rocSOLVER [\#2034](https://github.com/kokkos/kokkos-kernels/pull/2034)
+  - Fix rocSOLVER issue with Trilinos dependency [\#2037](https://github.com/kokkos/kokkos-kernels/pull/2037)
+- Lapack - SVD [\#2092](https://github.com/kokkos/kokkos-kernels/pull/2092)
+  - Adding benchmark for SVD [\#2103](https://github.com/kokkos/kokkos-kernels/pull/2103)
+  - Quick return to fix cuSOLVER and improve performance [\#2107](https://github.com/kokkos/kokkos-kernels/pull/2107)
+  - Fix Intel MKL tolerance for SVD tests [\#2110](https://github.com/kokkos/kokkos-kernels/pull/2110)
+
+#### Sparse updates
+- Add block support to all SPILUK algorithms [\#2064](https://github.com/kokkos/kokkos-kernels/pull/2064)
+  - Block spiluk follow up [\#2085](https://github.com/kokkos/kokkos-kernels/pull/2085)
+  - Make spiluk_handle::reset backwards compatible [\#2087](https://github.com/kokkos/kokkos-kernels/pull/2087)
+- Sptrsv improvements
+  - Add sptrsv execution space overloads [\#1982](https://github.com/kokkos/kokkos-kernels/pull/1982)
+  - Refactor Test_Sparse_sptrsv [\#2102](https://github.com/kokkos/kokkos-kernels/pull/2102)
+  - Add support for BSR matrices to some trsv routines [\#2104](https://github.com/kokkos/kokkos-kernels/pull/2104)
+- GMRES: Add support for BSR matrices [\#2097](https://github.com/kokkos/kokkos-kernels/pull/2097)
+- Spmv handle [\#2126](https://github.com/kokkos/kokkos-kernels/pull/2126)
+- Option to apply RCM reordering to extracted CRS diagonal blocks [\#2125](https://github.com/kokkos/kokkos-kernels/pull/2125)
+
+#### ODE updates
+- Adding adaptive BDF methods [\#1930](https://github.com/kokkos/kokkos-kernels/pull/1930)
+
+#### Misc updates
+- Add HIPManagedSpace support [\#2079](https://github.com/kokkos/kokkos-kernels/pull/2079)
+
+### Enhancements:
+
+#### BLAS
+- Axpby: improvement on unification attempt logic and on the execution of a diversity of situations [\#1895](https://github.com/kokkos/kokkos-kernels/pull/1895)
+
+#### Misc updates
+- Use execution space operator== [\#2136](https://github.com/kokkos/kokkos-kernels/pull/2136)
+
+#### TPL support
+- Add TPL support for KokkosBlas::dot [\#1949](https://github.com/kokkos/kokkos-kernels/pull/1949)
+- Add CUDA/HIP TPL support for KokkosSparse::spadd [\#1962](https://github.com/kokkos/kokkos-kernels/pull/1962)
+- Don't call optimize_gemv for one-shot MKL spmv [\#2073](https://github.com/kokkos/kokkos-kernels/pull/2073)
+- Async matrix release for MKL >= 2023.2 in SpMV [\#2074](https://github.com/kokkos/kokkos-kernels/pull/2074)
+- BLAS - MKL: fixing HostBlas calls to handle MKL_INT type [\#2112](https://github.com/kokkos/kokkos-kernels/pull/2112)
+
+### Build System:
+- Support CUBLAS_{LIBRARIES,LIBRARY_DIRS,INCLUDE_DIRS,ROOT} and KokkosKernels_CUBLAS_ROOT CMake options [\#2075](https://github.com/kokkos/kokkos-kernels/pull/2075)
+- Link std::filesystem for IntelLLVM in perf_test/sparse [\#2055](https://github.com/kokkos/kokkos-kernels/pull/2055)
+- Fix Cuda TPL finding [\#2098](https://github.com/kokkos/kokkos-kernels/pull/2098)
+- CMake: error out in certain case [\#2115](https://github.com/kokkos/kokkos-kernels/pull/2115)
+
+### Documentation and Testing:
+- par_ilut: Update documentation for fill_in_limit [\#2001](https://github.com/kokkos/kokkos-kernels/pull/2001)
+- Wiki examples for BLAS2 functions are added [\#2122](https://github.com/kokkos/kokkos-kernels/pull/2122)
+- github workflows: update to v4 (use Node 20) [\#2119](https://github.com/kokkos/kokkos-kernels/pull/2119)
+
+### Benchmarks:
+- gemm3 perf test: user CUDA, SYCL, or HIP device for kokkos:initialize [\#2058](https://github.com/kokkos/kokkos-kernels/pull/2058)
+- Lapack: adding svd benchmark [\#2103](https://github.com/kokkos/kokkos-kernels/pull/2103)
+- Benchmark: modifying spmv benchmark to fix interface and run range of spmv tests [\#2135](https://github.com/kokkos/kokkos-kernels/pull/2135)
+
+### Cleanup:
+- Experimental hip cleanup [\#1999](https://github.com/kokkos/kokkos-kernels/pull/1999)
+- iostream clean-up in benchmarks [\#2004](https://github.com/kokkos/kokkos-kernels/pull/2004)
+- Update: implicit capture of 'this' via '[=]' is deprecated in C++20 warnings [\#2076](https://github.com/kokkos/kokkos-kernels/pull/2076)
+- Deprecate KOKKOSLINALG_OPT_LEVEL [\#2072](https://github.com/kokkos/kokkos-kernels/pull/2072)
+- Remove all mentions of HBWSpace [\#2101](https://github.com/kokkos/kokkos-kernels/pull/2101)
+- Change name of yaml-cpp to yamlcpp (trilinos/Trilinos#12710) [\#2099](https://github.com/kokkos/kokkos-kernels/pull/2099)
+- Hands off namespace Kokkos::Impl - cleanup couple violations that snuck in [\#2094](https://github.com/kokkos/kokkos-kernels/pull/2094)
+- Kokkos Kernels: update version guards to drop old version of Kokkos [\#2133](https://github.com/kokkos/kokkos-kernels/pull/2133)
+- Sparse MKL: changing the location of the MKL_SAFE_CALL macro [\#2134](https://github.com/kokkos/kokkos-kernels/pull/2134)
+
+### Bug Fixes:
+- Bspgemm cusparse hang [\#2008](https://github.com/kokkos/kokkos-kernels/pull/2008)
+- bhalf_t fix for isnan function [\#2007](https://github.com/kokkos/kokkos-kernels/pull/2007)
+- Fence Kokkos before timed iterations [\#2066](https://github.com/kokkos/kokkos-kernels/pull/2066)
+- CUDA 11.2.1 / cuSPARSE 11.4.0 changed SpMV enums [\#2011](https://github.com/kokkos/kokkos-kernels/pull/2011)
+- Fix the spadd API [\#2090](https://github.com/kokkos/kokkos-kernels/pull/2090)
+- Axpby reduce deep copy calls [\#2081](https://github.com/kokkos/kokkos-kernels/pull/2081)
+- Correcting BLAS test failures with cuda when ETI_ONLY = OFF (issue #2061) [\#2077](https://github.com/kokkos/kokkos-kernels/pull/2077)
+- Fix weird Trilinos compiler error [\#2117](https://github.com/kokkos/kokkos-kernels/pull/2117)
+- Fix for missing STL inclusion [\#2113](https://github.com/kokkos/kokkos-kernels/pull/2113)
+- Fix build error in trsv on gcc8 [\#2111](https://github.com/kokkos/kokkos-kernels/pull/2111)
+- Add a workaround for compilation errors with cuda-12.2.0 + gcc-12.3 [\#2108](https://github.com/kokkos/kokkos-kernels/pull/2108)
+- Increase tolerance on gesv test (Fix #2123) [\#2124](https://github.com/kokkos/kokkos-kernels/pull/2124)
+- Fix usage of RAII to set cusparse/rocsparse stream [\#2141](https://github.com/kokkos/kokkos-kernels/pull/2141)
+- Spmv bsr matrix fix missing matrix descriptor (rocsparse) [\#2138](https://github.com/kokkos/kokkos-kernels/pull/2138)
+
 ## [4.2.01](https://github.com/kokkos/kokkos-kernels/tree/4.2.01) (2024-01-17)
 [Full Changelog](https://github.com/kokkos/kokkos-kernels/compare/4.2.00...4.2.01)
 

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -10,8 +10,8 @@ SET(KOKKOSKERNELS_TOP_BUILD_DIR ${CMAKE_CURRENT_BINARY_DIR})
 SET(KOKKOSKERNELS_TOP_SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR})
 
 SET(KokkosKernels_VERSION_MAJOR 4)
-SET(KokkosKernels_VERSION_MINOR 2)
-SET(KokkosKernels_VERSION_PATCH 1)
+SET(KokkosKernels_VERSION_MINOR 3)
+SET(KokkosKernels_VERSION_PATCH 0)
 SET(KokkosKernels_VERSION "${KokkosKernels_VERSION_MAJOR}.${KokkosKernels_VERSION_MINOR}.${KokkosKernels_VERSION_PATCH}")
 
 #Set variables for config file
@@ -127,13 +127,13 @@ ELSE()
   IF (NOT KOKKOSKERNELS_HAS_TRILINOS AND NOT KOKKOSKERNELS_HAS_PARENT)
     # This is a standalone build
     FIND_PACKAGE(Kokkos REQUIRED)
-    IF((${Kokkos_VERSION} VERSION_EQUAL "4.1.00") OR (${Kokkos_VERSION} VERSION_GREATER_EQUAL "4.2.00"))
+    IF((${Kokkos_VERSION} VERSION_GREATER_EQUAL "4.1.0") AND (${Kokkos_VERSION} VERSION_LESS_EQUAL "4.3.0"))
       MESSAGE(STATUS "Found Kokkos version ${Kokkos_VERSION} at ${Kokkos_DIR}")
-      IF((${Kokkos_VERSION} VERSION_GREATER "4.2.99"))
+      IF((${Kokkos_VERSION} VERSION_GREATER "4.3.99"))
         MESSAGE(WARNING "Configuring with Kokkos ${Kokkos_VERSION} which is newer than the expected develop branch - version check may need update")
       ENDIF()
     ELSE()
-      MESSAGE(FATAL_ERROR "Kokkos Kernels ${KokkosKernels_VERSION} requires 4.1.00, 4.2.00, 4.2.01 or develop")
+      MESSAGE(FATAL_ERROR "Kokkos Kernels ${KokkosKernels_VERSION} requires Kokkos_VERSION 4.1.0, 4.2.0, 4.2.1 or 4.3.0")
     ENDIF()
   ENDIF()
 
@@ -156,9 +156,16 @@ ELSE()
   KOKKOSKERNELS_ADD_OPTION_AND_DEFINE(
     LINALG_OPT_LEVEL
     KOKKOSLINALG_OPT_LEVEL
-    "Optimization level for KokkosKernels computational kernels: a nonnegative integer.  Higher levels result in better performance that is more uniform for corner cases, but increase build time and library size.  The default value is 1, which should give performance within ten percent of optimal on most platforms, for most problems. Default: 1"
+    "DEPRECATED. Optimization level for KokkosKernels computational kernels: a nonnegative integer.  Higher levels result in better performance that is more uniform for corner cases, but increase build time and library size.  The default value is 1, which should give performance within ten percent of optimal on most platforms, for most problems. Default: 1"
     "1")
 
+  if (KokkosKernels_LINALG_OPT_LEVEL AND NOT KokkosKernels_LINALG_OPT_LEVEL STREQUAL "1")
+   message(WARNING "KokkosKernels_LINALG_OPT_LEVEL is deprecated!")
+ endif()
+  if(KokkosKernels_KOKKOSLINALG_OPT_LEVEL AND NOT KokkosKernels_KOKKOSLINALG_OPT_LEVEL STREQUAL "1")
+    message(WARNING "KokkosKernels_KOKKOSLINALG_OPT_LEVEL is deprecated!")
+  endif()
+
   # Enable experimental features of KokkosKernels if set at configure
   # time. Default is no.
   KOKKOSKERNELS_ADD_OPTION_AND_DEFINE(
@@ -375,8 +382,10 @@ ELSE()
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC MKL)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC CUBLAS)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC CUSPARSE)
+  KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC CUSOLVER)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC ROCBLAS)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC ROCSPARSE)
+  KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC ROCSOLVER)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC METIS)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC ARMPL)
   KOKKOSKERNELS_LINK_TPL(kokkoskernels PUBLIC MAGMA)
@@ -425,7 +434,7 @@ ELSE()
   IF (KOKKOSKERNELS_ALL_COMPONENTS_ENABLED)
     IF (KokkosKernels_ENABLE_PERFTESTS)
       MESSAGE(STATUS "Enabling perf tests.")
-      KOKKOSKERNELS_ADD_TEST_DIRECTORIES(perf_test)
+      add_subdirectory(perf_test) # doesn't require KokkosKernels_ENABLE_TESTS=ON
     ENDIF ()
     IF (KokkosKernels_ENABLE_EXAMPLES)
       MESSAGE(STATUS "Enabling examples.")

diff --git a/CheckHostBlasReturnComplex.cmake b/CheckHostBlasReturnComplex.cmake
@@ -21,8 +21,8 @@ FUNCTION(CHECK_HOST_BLAS_RETURN_COMPLEX VARNAME)
 
 extern \"C\" {
   void F77_BLAS_MANGLE(zdotc,ZDOTC)(
-    std::complex<double>* result, const int* n, 
-    const std::complex<double> x[], const int* incx, 
+    std::complex<double>* result, const int* n,
+    const std::complex<double> x[], const int* incx,
     const std::complex<double> y[], const int* incy);
 }
 
@@ -49,9 +49,9 @@ int main() {
 CHECK_CXX_SOURCE_RUNS("${SOURCE}" KK_BLAS_RESULT_AS_POINTER_ARG)
 
 IF(${KK_BLAS_RESULT_AS_POINTER_ARG})
-  SET(VARNAME OFF)
+  SET(${VARNAME} OFF PARENT_SCOPE)
 ELSE()
-  SET(VARNAME ON)
+  SET(${VARNAME} ON PARENT_SCOPE)
 ENDIF()
 
 ENDFUNCTION()
diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-[![Generic badge](https://readthedocs.org/projects/pip/badge/?version=latest&style=flat)](https://kokkos-kernels.readthedocs.io/en/latest/)
+[![Generic badge](https://readthedocs.org/projects/kokkos-kernels/badge/?version=latest)](https://kokkos-kernels.readthedocs.io/en/latest/)
 
 ![KokkosKernels](https://avatars2.githubusercontent.com/u/10199860?s=200&v=4)
 

diff --git a/batched/KokkosBatched_Util.hpp b/batched/KokkosBatched_Util.hpp
@@ -626,18 +626,6 @@ KOKKOS_INLINE_FUNCTION auto subview_wrapper(ViewType v, IdxType1 i1,
                                             const Trans::NoTranspose) {
   return subview_wrapper(v, i1, i2, i3, layout_tag);
 }
-#if KOKKOS_VERSION < 40099
-template <class ViewType, class IdxType1>
-KOKKOS_INLINE_FUNCTION auto subview_wrapper(ViewType v, IdxType1 i1,
-                                            Kokkos::Impl::ALL_t i2,
-                                            Kokkos::Impl::ALL_t i3,
-                                            const BatchLayout::Left &layout_tag,
-                                            const Trans::Transpose) {
-  auto sv_nt = subview_wrapper(v, i1, i3, i2, layout_tag);
-
-  return transpose_2d_view(sv_nt, layout_tag);
-}
-#else
 template <class ViewType, class IdxType1>
 KOKKOS_INLINE_FUNCTION auto subview_wrapper(ViewType v, IdxType1 i1,
                                             Kokkos::ALL_t i2, Kokkos::ALL_t i3,
@@ -647,7 +635,6 @@ KOKKOS_INLINE_FUNCTION auto subview_wrapper(ViewType v, IdxType1 i1,
 
   return transpose_2d_view(sv_nt, layout_tag);
 }
-#endif
 template <class ViewType, class IdxType1, class IdxType2, class IdxType3>
 KOKKOS_INLINE_FUNCTION auto subview_wrapper(ViewType v, IdxType1 i1,
                                             IdxType2 i2, IdxType3 i3,
@@ -671,16 +658,6 @@ KOKKOS_INLINE_FUNCTION auto subview_wrapper(
     const BatchLayout::Right &layout_tag, const Trans::NoTranspose &) {
   return subview_wrapper(v, i1, i2, i3, layout_tag);
 }
-#if KOKKOS_VERSION < 40099
-template <class ViewType, class IdxType1>
-KOKKOS_INLINE_FUNCTION auto subview_wrapper(
-    ViewType v, IdxType1 i1, Kokkos::Impl::ALL_t i2, Kokkos::Impl::ALL_t i3,
-    const BatchLayout::Right &layout_tag, const Trans::Transpose &) {
-  auto sv_nt = subview_wrapper(v, i1, i3, i2, layout_tag);
-
-  return transpose_2d_view(sv_nt, layout_tag);
-}
-#else
 template <class ViewType, class IdxType1>
 KOKKOS_INLINE_FUNCTION auto subview_wrapper(
     ViewType v, IdxType1 i1, Kokkos::ALL_t i2, Kokkos::ALL_t i3,
@@ -689,7 +666,6 @@ KOKKOS_INLINE_FUNCTION auto subview_wrapper(
 
   return transpose_2d_view(sv_nt, layout_tag);
 }
-#endif
 template <class ViewType, class IdxType1, class IdxType2, class IdxType3>
 KOKKOS_INLINE_FUNCTION auto subview_wrapper(
     ViewType v, IdxType1 i1, IdxType2 i2, IdxType3 i3,