1888/dense linop

benchmark/sparse_blas/operations.cpp

@@ -9,6 +9,8 @@
 
 #include <gflags/gflags.h>
 
+#include <ginkgo/core/matrix/dense.hpp>
+
 #include "core/base/array_access.hpp"
 #include "core/factorization/elimination_forest.hpp"
 #include "core/factorization/factorization_kernels.hpp"

cmake/create_test.cmake

@@ -296,11 +296,15 @@ function(ginkgo_create_cuda_test_internal test_name filename test_target_name)
                 --expt-relaxed-constexpr>
         )
     elseif(CMAKE_CUDA_COMPILER_ID STREQUAL "NVIDIA")
+        # 997: remove false positive warning about overloading virtual functions
         target_compile_options(
             ${test_target_name}
             PRIVATE
                 $<$<COMPILE_LANGUAGE:CUDA>:--expt-extended-lambda
-                --expt-relaxed-constexpr>
+                --expt-relaxed-constexpr
+                --diag-suppress
+                997
+                >
         )
     endif()
     ginkgo_set_test_target_properties(${test_target_name} "_cuda" ${ARGN})

common/cuda_hip/CMakeLists.txt

@@ -30,6 +30,7 @@ set(CUDA_HIP_SOURCES
     matrix/batch_dense_kernels.cpp
     matrix/batch_ell_kernels.cpp
     matrix/coo_kernels.cpp
+    matrix/dense_kernels.cpp
     matrix/diagonal_kernels.cpp
     matrix/ell_kernels.cpp
     matrix/multivector_kernels.cpp

common/cuda_hip/factorization/lu_kernels.cpp

@@ -13,6 +13,7 @@
 #include <thrust/tuple.h>
 
 #include <ginkgo/core/matrix/csr.hpp>
+#include <ginkgo/core/matrix/dense.hpp>
 
 #include "common/cuda_hip/base/thrust.hpp"
 #include "common/cuda_hip/base/types.hpp"