CUTLASS 3.0.0 (#786)

* CUTLASS 3.0.0

test/unit/CMakeLists.txt

@@ -28,6 +28,8 @@
 
 include(CTest)
 
+set(CUTLASS_UNIT_TEST_COMMON_DIR ${CMAKE_CURRENT_LIST_DIR}/common)
+
 cutlass_add_library(
   cutlass_test_unit_infra 
   OBJECT
@@ -42,6 +44,7 @@ target_link_libraries(
   $<$<BOOL:${CUTLASS_ENABLE_CUBLAS}>:nvidia::cublas>
   gtest
   cudart
+  cuda_driver
   )
 
 cutlass_add_library(
@@ -69,6 +72,12 @@ function(cutlass_test_unit_add_executable NAME)
 
   target_compile_definitions(${NAME} PUBLIC CUTLASS_TARGET_NAME="${NAME}")
 
+  target_include_directories(
+    ${NAME}
+    PRIVATE
+    ${CUTLASS_UNIT_TEST_COMMON_DIR}
+  )
+
   target_link_libraries(
     ${NAME}
     PRIVATE
@@ -76,6 +85,10 @@ function(cutlass_test_unit_add_executable NAME)
     cutlass_test_unit_infra_lib
   )
 
+  if (CUTLASS_ENABLE_OPENMP_TESTS AND OpenMP_CXX_FOUND)
+    target_link_libraries(${NAME} PRIVATE OpenMP::OpenMP_CXX)
+  endif()
+
   string(REGEX REPLACE cutlass_ "" NAME_STEM ${NAME})
 
   set(RESULT_CACHE_FILE "${CUTLASS_TEST_UNIT_RESULTS_CACHE_DIR}/cached_results_${NAME}.txt")
@@ -99,6 +112,7 @@ add_custom_target(test_unit)
 
 set(SUBDIRS
   core
+  cute
   gemm
   conv
   layout
@@ -106,6 +120,7 @@ set(SUBDIRS
   epilogue
   reduction
   util
+  pipeline
   )
 
 if(TARGET nvidia::nvrtc AND TARGET nvidia::cuda_driver)