Fix cmake.

CMakeLists.txt

@@ -26,7 +26,23 @@ set(version 1.1.0)
 project(composable_kernel VERSION ${version} LANGUAGES CXX HIP)
 include(CTest)
 
-find_package(Python3 3.6 COMPONENTS Interpreter REQUIRED)
+# Usage: for customized Python location cmake -DCK_USE_ALTERNATIVE_PYTHON="/opt/Python-3.8.13/bin/python3.8"
+# CK Codegen requires dataclass which is added in Python 3.7
+# Python version 3.8 is required for general good practice as it is default for Ubuntu 20.04
+if(NOT CK_USE_ALTERNATIVE_PYTHON)
+   find_package(Python3 3.8 COMPONENTS Interpreter REQUIRED)
+else()
+   message("Using alternative python version")
+   set(EXTRA_PYTHON_PATH)
+   # this is overly restrictive, we may need to be more flexible on the following
+   string(REPLACE "/bin/python3.8" "" EXTRA_PYTHON_PATH "${CK_USE_ALTERNATIVE_PYTHON}")
+   message("alternative python path is: ${EXTRA_PYTHON_PATH}")
+   find_package(Python3 3.6 COMPONENTS Interpreter REQUIRED)
+   add_definitions(-DPython3_EXECUTABLE="${CK_USE_ALTERNATIVE_PYTHON}")
+   set(Python3_EXECUTABLE "${CK_USE_ALTERNATIVE_PYTHON}")
+   set(PYTHON_EXECUTABLE "${CK_USE_ALTERNATIVE_PYTHON}")
+   set(ENV{LD_LIBRARY_PATH} "${EXTRA_PYTHON_PATH}/lib:$ENV{LD_LIBRARY_PATH}")
+endif()
 
 list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
 
@@ -62,17 +78,14 @@ if (DTYPES)
     endif()
     message("DTYPES macro set to ${DTYPES}")
 else()
-    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
+    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16 -DCK_ENABLE_FP8 -DCK_ENABLE_BF8)
     set(CK_ENABLE_INT8 "ON")
     set(CK_ENABLE_FP16 "ON")
     set(CK_ENABLE_FP32 "ON")
     set(CK_ENABLE_FP64 "ON")
     set(CK_ENABLE_BF16 "ON")
-    if (GPU_TARGETS MATCHES "gfx94")
-        add_definitions(-DCK_ENABLE_FP8 -DCK_ENABLE_BF8)
-        set(CK_ENABLE_FP8 "ON")
-        set(CK_ENABLE_BF8 "ON")
-    endif()
+    set(CK_ENABLE_FP8 "ON")
+    set(CK_ENABLE_BF8 "ON")
 endif()
 
 #for f8/bf8_t type
@@ -191,12 +204,18 @@ endif()
 configure_file(include/ck/config.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/ck/config.h)
 
 if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 500723302)
-   message("Adding the fno-offload-uniform-block compiler flag")
-   add_compile_options(-fno-offload-uniform-block)
+  check_cxx_compiler_flag("-fno-offload-uniform-block" HAS_NO_OFFLOAD_UNIFORM_BLOCK)
+  if(HAS_NO_OFFLOAD_UNIFORM_BLOCK)
+    message("Adding the fno-offload-uniform-block compiler flag")
+    add_compile_options(-fno-offload-uniform-block)
+  endif()
 endif()
 if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600140090)
-   message("Adding the enable-post-misched=0 compiler flag")
-   add_compile_options("SHELL: -mllvm -enable-post-misched=0")
+  check_cxx_compiler_flag("-mllvm -enable-post-misched=0" HAS_ENABLE_POST_MISCHED)
+  if(HAS_ENABLE_POST_MISCHED)
+    message("Adding the enable-post-misched=0 compiler flag")
+    add_compile_options("SHELL: -mllvm -enable-post-misched=0")
+  endif()
 endif()
 set(check-coerce)
 check_cxx_compiler_flag(" -mllvm -amdgpu-coerce-illegal-types=1" check-coerce)
@@ -550,12 +569,7 @@ if(NOT DEFINED INSTANCES_ONLY)
         PACKAGE_NAME examples
    )
    add_subdirectory(example)
-   if(GPU_TARGETS MATCHES "gfx9" AND NOT INSTANCES_ONLY)
-      add_subdirectory(codegen)
-   endif()
-   if(BUILD_TESTING)
-      add_subdirectory(test)
-   endif()
+   add_subdirectory(test)
 
    rocm_package_setup_component(profiler
         LIBRARY_NAME composablekernel
@@ -572,6 +586,10 @@ if(NOT DEFINED INSTANCES_ONLY)
   endif()
 endif()
 
+if(NOT DEFINED PROFILER_ONLY AND (GPU_TARGETS MATCHES "gfx9" OR DEFINED INSTANCES_ONLY))
+  add_subdirectory(codegen)
+endif()
+
 #Create an interface target for the include only files and call it "composablekernels"
 include(CMakePackageConfigHelpers)
 

codegen/CMakeLists.txt

-cmake_minimum_required(VERSION 3.16)
-project(composable_kernel_host LANGUAGES CXX HIP)
-find_package(ROCM)
-include(ROCMInstallTargets)
-include(ROCMTest)
-  
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
-  
+
 set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
 set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)
 set(CK_ROOT ${CMAKE_CURRENT_SOURCE_DIR}/..)
-  
+
 add_compile_options(-std=c++17)
 find_package(hip)
 add_custom_target(codegen)
-  
+
 # add include directories
 include_directories(BEFORE
             ${PROJECT_BINARY_DIR}/include
             ${PROJECT_SOURCE_DIR}/include
             ${PROJECT_SOURCE_DIR}/library/include
             ${HIP_INCLUDE_DIRS}
-            ${CK_ROOT}/include/
             )
-  
+
 list(APPEND CMAKE_MODULE_PATH ${CK_ROOT}/cmake)
 include(Embed)
 file(GLOB_RECURSE KERNEL_FILES CONFIGURE_DEPENDS
@@ -32,31 +25,29 @@ file(GLOB_RECURSE KERNEL_FILES CONFIGURE_DEPENDS
 #message(STATUS "KERNEL_FILES: ${KERNEL_FILES}")
 #message(STATUS "RELATIVE: ${CK_ROOT}/include")
 add_embed_library(ck_headers ${KERNEL_FILES} RELATIVE ${CK_ROOT}/include)
-  
+
 file(GLOB SOURCES CONFIGURE_DEPENDS src/*.cpp)
-  
+
 ##message(STATUS "SOURCE_FILES: ${SOURCES}")
 # TODO: Use object library
 add_library(ck_host STATIC ${SOURCES})
 target_link_libraries(ck_host PRIVATE ck_headers)
-  
+
 set_target_properties(ck_host PROPERTIES
 	LINKER_LANGUAGE CXX
         POSITION_INDEPENDENT_CODE ON)
-  
+
 target_include_directories(ck_host PUBLIC
     $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
-    $<BUILD_INTERFACE:${CMAKE_CURRENT_BINARY_DIR}/solution_instances>
-    $<BUILD_INTERFACE:${CMAKE_CURRENT_BINARY_DIR}/embed/ck_headers/include>
 )
-  
+
 add_executable(ck-template-driver driver/main.cpp)
 target_link_libraries(ck-template-driver ck_host)
-  
+
 rocm_install(
-    TARGETS ck_host
+    TARGETS ck_host ck_headers
     EXPORT ck_hostTargets
 )
 rocm_install(DIRECTORY include/ck DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
-  
+
 add_subdirectory(test)