add installable cmake package

CMakeLists.txt

@@ -72,8 +72,9 @@ message(STATUS "Build with HIP ${HIP_VERSION}")
 
 
 rocm_create_package(
-    NAME CK-${CK_BACKEND}
+    NAME composablekernel
     DESCRIPTION "High Performance Composable Kernel for AMD GPUs"
+    MAINTAINER "MIOpen Kernels Dev Team <dl.MIOpen@amd.com>"
     LDCONFIG
 )
 
@@ -245,3 +246,25 @@ add_subdirectory(library)
 add_subdirectory(example)
 add_subdirectory(test)
 add_subdirectory(profiler)
+
+#Create an interface target for the include only files and call it "coposablekernels"
+include(CMakePackageConfigHelpers)
+
+set(version 1.0.0)
+write_basic_package_version_file(
+    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
+    VERSION "${version}"
+    COMPATIBILITY AnyNewerVersion
+)
+
+configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in 
+        "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
+        INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel 
+        NO_CHECK_REQUIRED_COMPONENTS_MACRO
+)
+
+install(FILES 
+    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
+    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
+    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel 
+)

Config.cmake.in

+@PACKAGE_INIT@
+
+set(_composable_kernel_supported_components device_operations host_tensor)
+
+foreach(_comp ${composable_kernel_FIND_COMPONENTS})
+	if(NOT _comp IN_LIST _composable_kernel_supported_components)
+		set(composable_kernel_FOUND False)
+		set(composable_kernel_NOT_FOUND_MESSAGE "Unsupported component: ${_comp}")
+	endif()
+	include("${CMAKE_CURRENT_LIST_DIR}/composable_kernel${_comp}Targets.cmake")
+endforeach()
\ No newline at end of file

Dockerfile

@@ -11,13 +11,7 @@ ARG DEB_ROCM_REPO=http://repo.radeon.com/rocm/apt/.apt_$ROCMVERSION/
 RUN apt-get update
 RUN apt-get install -y wget gnupg
 RUN wget -qO - http://repo.radeon.com/rocm/rocm.gpg.key | apt-key add -
-RUN if ! [ -z $OSDB_BKC_VERSION ]; then \
-       echo "Using BKC VERISION: $OSDB_BKC_VERSION";\
-       sh -c "echo deb [arch=amd64 trusted=yes] http://compute-artifactory.amd.com/artifactory/list/rocm-osdb-deb/ compute-rocm-dkms-no-npi-hipclang ${OSDB_BKC_VERSION} > /etc/apt/sources.list.d/rocm.list" ;\
-       cat  /etc/apt/sources.list.d/rocm.list;\
-    else \
-       sh -c "echo deb [arch=amd64] $DEB_ROCM_REPO ubuntu main > /etc/apt/sources.list.d/rocm.list" ;\
-    fi
+RUN sh -c "echo deb [arch=amd64] $DEB_ROCM_REPO ubuntu main > /etc/apt/sources.list.d/rocm.list"
 RUN wget --no-check-certificate -qO - https://apt.kitware.com/keys/kitware-archive-latest.asc 2>/dev/null | apt-key add -
 RUN sh -c "echo deb https://apt.kitware.com/ubuntu/ bionic main | tee -a /etc/apt/sources.list"
 

example/01_gemm/gemm_xdl_fp16.cpp

@@ -41,7 +41,6 @@ using AElementOp = ck::tensor_operation::element_wise::PassThrough;
 using BElementOp = ck::tensor_operation::element_wise::PassThrough;
 using CElementOp = ck::tensor_operation::element_wise::PassThrough;
 
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization_t::Default;
 
 // clang-format off
 #if 0

library/CMakeLists.txt

 add_subdirectory(src/host_tensor)
 add_subdirectory(src/tensor_operation_instance/gpu)
+
+# rocm_install_targets(
+#   TARGETS device_operations host_tensor
+#   INCLUDE
+#     ${PROJECT_SOURCE_DIR}/include
+#     ${PROJECT_SOURCE_DIR}/library/include
+# )
\ No newline at end of file

library/src/host_tensor/CMakeLists.txt

@@ -10,10 +10,38 @@ set(HOST_TENSOR_SOURCE
     host_tensor.cpp
 )
 
-add_library(host_tensor SHARED ${HOST_TENSOR_SOURCE})
+add_library(host_tensor STATIC ${HOST_TENSOR_SOURCE})
+add_library(composable_kernel::host_tensor ALIAS host_tensor)
+
 target_compile_features(host_tensor PUBLIC)
 set_target_properties(host_tensor PROPERTIES POSITION_INDEPENDENT_CODE ON)
 target_include_directories(host_tensor SYSTEM PUBLIC $<BUILD_INTERFACE:${HALF_INCLUDE_DIR}>)
-install(TARGETS host_tensor LIBRARY DESTINATION lib)
+
+target_include_directories(host_tensor PUBLIC 
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck>"
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/utility>"
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/host_tensor>"
+)
+
+install(TARGETS host_tensor 
+        EXPORT host_tensorTargets
+        LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+        INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+)
+
+install(DIRECTORY
+    ${PROJECT_SOURCE_DIR}/include/ck
+    ${PROJECT_SOURCE_DIR}/include/ck/utility
+    ${PROJECT_SOURCE_DIR}/library/include/ck/library/host_tensor
+    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+)
+
+install(EXPORT host_tensorTargets
+    FILE composable_kernelhost_tensorTargets.cmake 
+    NAMESPACE composable_kernel::
+    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
+)
 
 clang_tidy_check(host_tensor)

library/src/tensor_operation_instance/gpu/CMakeLists.txt

@@ -28,3 +28,82 @@ add_subdirectory(conv2d_fwd_bias_relu_add)
 add_subdirectory(conv2d_fwd_bias_relu_atomic_add)
 add_subdirectory(conv2d_bwd_data)
 add_subdirectory(reduce)
+
+add_library(device_operations STATIC 
+    $<TARGET_OBJECTS:device_conv1d_fwd_instance> 
+    $<TARGET_OBJECTS:device_batched_gemm_instance> 
+    $<TARGET_OBJECTS:device_conv2d_bwd_data_instance> 
+    $<TARGET_OBJECTS:device_conv2d_fwd_instance> 
+    $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_instance> 
+    $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_add_instance>
+    $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_atomic_add_instance>
+    $<TARGET_OBJECTS:device_gemm_instance>
+    $<TARGET_OBJECTS:device_gemm_bias_relu_instance>
+    $<TARGET_OBJECTS:device_gemm_bias_relu_add_instance>
+    $<TARGET_OBJECTS:device_gemm_bias2d_instance>
+    $<TARGET_OBJECTS:device_reduce_instance>
+)
+add_library(composablekernels::device_operations ALIAS device_operations)
+set(DEV_OPS_INC_DIRS 
+    ${PROJECT_SOURCE_DIR}/include/ck
+    ${PROJECT_SOURCE_DIR}/include/ck/utility
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_description
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor
+    ${PROJECT_SOURCE_DIR}/include/ck/problem_transform
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/device
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/grid
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/block
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/warp
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/thread
+    ${PROJECT_SOURCE_DIR}/include/ck/tensor_operation/gpu/element
+    ${PROJECT_SOURCE_DIR}/library/include/ck/library/host_tensor
+    ${PROJECT_SOURCE_DIR}/library/include/ck/library/tensor_operation_instance
+    ${PROJECT_SOURCE_DIR}/library/include/ck/library/tensor_operation_instance/gpu/reduce
+    ${PROJECT_SOURCE_DIR}/external/include/half
+)
+# target_include_directories(device_operations PUBLIC 
+#             $<INSTALL_INTERFACE:${DEV_OPS_INC_DIRS}>
+# )
+
+target_compile_features(device_operations PUBLIC)
+set_target_properties(device_operations PROPERTIES POSITION_INDEPENDENT_CODE ON)
+target_include_directories(device_operations PUBLIC 
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/utility>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_description>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/problem_transform>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/device>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/grid>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/block>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/warp>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/thread>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/element>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/host_tensor>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/tensor_operation_instance>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/tensor_operation_instance/gpu/reduce>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/half>
+)
+
+#once new arches are enabled make this an option on the main cmake file
+# and pass down here to be exported
+
+target_compile_definitions(device_operations 
+PUBLIC -DCK_AMD_GPU_GFX908
+PRIVATE -amdgpu-target=gfx908
+PRIVATE -O3
+)
+# install(TARGETS device_operations LIBRARY DESTINATION lib)
+install(TARGETS device_operations
+        EXPORT device_operationsTargets
+        LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+        INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+)
+install(DIRECTORY ${DEV_OPS_INC_DIRS} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+install(EXPORT device_operationsTargets 
+        FILE composable_kerneldevice_operationsTargets.cmake 
+        NAMESPACE composable_kernel::
+        DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
+)

library/src/tensor_operation_instance/gpu/batched_gemm/CMakeLists.txt

@@ -6,9 +6,9 @@ set(DEVICE_BATCHED_GEMM_INSTANCE_SOURCE
    device_batched_gemm_xdl_f16_f16_f16_gkm_gnk_gmn_instance.cpp;
 )
 
-add_library(device_batched_gemm_instance SHARED ${DEVICE_BATCHED_GEMM_INSTANCE_SOURCE})
-target_compile_features(device_batched_gemm_instance PUBLIC)
+add_library(device_batched_gemm_instance OBJECT ${DEVICE_BATCHED_GEMM_INSTANCE_SOURCE})
+# target_compile_features(device_batched_gemm_instance PUBLIC)
 set_target_properties(device_batched_gemm_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_batched_gemm_instance LIBRARY DESTINATION lib)
+# install(TARGETS device_batched_gemm_instance LIBRARY DESTINATION lib)
 
 clang_tidy_check(device_batched_gemm_instance)

library/src/tensor_operation_instance/gpu/conv1d_fwd/CMakeLists.txt

@@ -3,9 +3,9 @@ set(DEVICE_CONV1D_FWD_INSTANCE_SOURCE
    device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp;
 )
 
-add_library(device_conv1d_fwd_instance SHARED ${DEVICE_CONV1D_FWD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv1d_fwd_instance PUBLIC)
+add_library(device_conv1d_fwd_instance OBJECT ${DEVICE_CONV1D_FWD_INSTANCE_SOURCE}) 
+# target_compile_features(device_conv1d_fwd_instance PUBLIC)
 set_target_properties(device_conv1d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv1d_fwd_instance LIBRARY DESTINATION lib) 
+# install(TARGETS device_conv1d_fwd_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv1d_fwd_instance)

library/src/tensor_operation_instance/gpu/conv2d_bwd_data/CMakeLists.txt

@@ -6,9 +6,7 @@ set(DEVICE_CONV2D_BWD_DATA_INSTANCE_SOURCE
    device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp;
 ) 
 
-add_library(device_conv2d_bwd_data_instance SHARED ${DEVICE_CONV2D_BWD_DATA_INSTANCE_SOURCE})
-target_compile_features(device_conv2d_bwd_data_instance PUBLIC)
+add_library(device_conv2d_bwd_data_instance OBJECT ${DEVICE_CONV2D_BWD_DATA_INSTANCE_SOURCE})
 set_target_properties(device_conv2d_bwd_data_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_bwd_data_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_bwd_data_instance)

library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt

@@ -6,9 +6,7 @@ set(DEVICE_CONV2D_FWD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp;
    device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
-add_library(device_conv2d_fwd_instance SHARED ${DEVICE_CONV2D_FWD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_instance PUBLIC)
+add_library(device_conv2d_fwd_instance OBJECT ${DEVICE_CONV2D_FWD_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_instance)

library/src/tensor_operation_instance/gpu/conv2d_fwd_bias_relu/CMakeLists.txt

@@ -2,9 +2,7 @@
 set(DEVICE_CONV2D_FWD_BIAS_RELU_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_c_shuffle_bias_relu_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
-add_library(device_conv2d_fwd_bias_relu_instance SHARED ${DEVICE_CONV2D_FWD_BIAS_RELU_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_bias_relu_instance PUBLIC)
+add_library(device_conv2d_fwd_bias_relu_instance OBJECT ${DEVICE_CONV2D_FWD_BIAS_RELU_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_bias_relu_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_bias_relu_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_bias_relu_instance)

library/src/tensor_operation_instance/gpu/conv2d_fwd_bias_relu_add/CMakeLists.txt

@@ -2,9 +2,7 @@
 set(DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_c_shuffle_bias_relu_add_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
-add_library(device_conv2d_fwd_bias_relu_add_instance SHARED ${DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_bias_relu_add_instance PUBLIC)
+add_library(device_conv2d_fwd_bias_relu_add_instance OBJECT ${DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_bias_relu_add_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_bias_relu_add_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_bias_relu_add_instance)

library/src/tensor_operation_instance/gpu/conv2d_fwd_bias_relu_atomic_add/CMakeLists.txt

@@ -3,9 +3,7 @@ set(DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_c_shuffle_bias_relu_atomic_add_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
 
-add_library(device_conv2d_fwd_bias_relu_atomic_add_instance SHARED ${DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_bias_relu_atomic_add_instance PUBLIC)
+add_library(device_conv2d_fwd_bias_relu_atomic_add_instance OBJECT ${DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_bias_relu_atomic_add_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_bias_relu_atomic_add_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_bias_relu_atomic_add_instance)

library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt

@@ -25,10 +25,7 @@ set(DEVICE_GEMM_INSTANCE_SOURCE
    device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp;
 )
 
-add_library(device_gemm_instance SHARED ${DEVICE_GEMM_INSTANCE_SOURCE}) 
-
-target_compile_features(device_gemm_instance PUBLIC)
+add_library(device_gemm_instance OBJECT ${DEVICE_GEMM_INSTANCE_SOURCE}) 
 set_target_properties(device_gemm_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_gemm_instance LIBRARY DESTINATION lib)
 
 clang_tidy_check(device_gemm_instance)

library/src/tensor_operation_instance/gpu/gemm_bias2d/CMakeLists.txt

@@ -10,9 +10,7 @@ set(DEVICE_GEMM_BIAS2D_INSTANCE_SOURCE
    device_gemm_xdl_c_shuffle_bias_2d_f16_f16_f16_mk_nk_mn_instance.cpp;
 )
 
-add_library(device_gemm_bias2d_instance SHARED ${DEVICE_GEMM_BIAS2D_INSTANCE_SOURCE})
-target_compile_features(device_gemm_bias2d_instance PUBLIC)
+add_library(device_gemm_bias2d_instance OBJECT ${DEVICE_GEMM_BIAS2D_INSTANCE_SOURCE})
 set_target_properties(device_gemm_bias2d_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_gemm_bias2d_instance LIBRARY DESTINATION lib)
 
 clang_tidy_check(device_gemm_bias2d_instance)

library/src/tensor_operation_instance/gpu/gemm_bias_relu/CMakeLists.txt

@@ -6,9 +6,7 @@ set(DEVICE_GEMM_BIAS_RELU_INSTANCE_SOURCE
    device_gemm_xdl_c_shuffle_bias_relu_f16_f16_f16_km_nk_mn_instance.cpp;
 )
 
-add_library(device_gemm_bias_relu_instance SHARED ${DEVICE_GEMM_BIAS_RELU_INSTANCE_SOURCE}) 
-target_compile_features(device_gemm_bias_relu_instance PUBLIC)
+add_library(device_gemm_bias_relu_instance OBJECT ${DEVICE_GEMM_BIAS_RELU_INSTANCE_SOURCE}) 
 set_target_properties(device_gemm_bias_relu_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_gemm_bias_relu_instance LIBRARY DESTINATION lib)
 
 clang_tidy_check(device_gemm_bias_relu_instance)

library/src/tensor_operation_instance/gpu/gemm_bias_relu_add/CMakeLists.txt

@@ -6,9 +6,7 @@ set(DEVICE_GEMM_BIAS_RELU_ADD_INSTANCE_SOURCE
    device_gemm_xdl_c_shuffle_bias_relu_add_f16_f16_f16_km_nk_mn_instance.cpp;
 )
 
-add_library(device_gemm_bias_relu_add_instance SHARED ${DEVICE_GEMM_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
-target_compile_features(device_gemm_bias_relu_add_instance PUBLIC)
+add_library(device_gemm_bias_relu_add_instance OBJECT ${DEVICE_GEMM_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
 set_target_properties(device_gemm_bias_relu_add_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_gemm_bias_relu_add_instance LIBRARY DESTINATION lib)
 
 clang_tidy_check(device_gemm_bias_relu_add_instance)

library/src/tensor_operation_instance/gpu/reduce/CMakeLists.txt

@@ -25,9 +25,7 @@ set(DEVICE_REDUCE_INSTANCE_SOURCE
    device_reduce_instance_multiblock_partial_reduce_f64_f64_f64.cpp;
 )
 
-add_library(device_reduce_instance SHARED ${DEVICE_REDUCE_INSTANCE_SOURCE}) 
-target_compile_features(device_reduce_instance PUBLIC)
+add_library(device_reduce_instance OBJECT ${DEVICE_REDUCE_INSTANCE_SOURCE}) 
 set_target_properties(device_reduce_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_reduce_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_reduce_instance)