added int4/int8 dequantization support

cmake/ClangTidy.cmake

@@ -146,16 +146,18 @@ function(clang_tidy_check TARGET)
         if((NOT "${SOURCE}" MATCHES "(h|hpp|hxx)$") AND (NOT "${SOURCE}" MATCHES "TARGET_OBJECTS"))
             string(MAKE_C_IDENTIFIER "${SOURCE}" tidy_file)
             set(tidy_target tidy-target-${TARGET}-${tidy_file})
-            add_custom_target(${tidy_target}
-                # for some targets clang-tidy not able to get information from .clang-tidy
-                DEPENDS ${SOURCE}
-                COMMAND ${CLANG_TIDY_COMMAND} "-config=\{CheckOptions: \[\{key: bugprone-reserved-identifier.AllowedIdentifiers,value: __HIP_PLATFORM_HCC__\; __HIP_PLATFORM_AMD__\; __HIP_ROCclr__\}\]\}" ${SOURCE} "-export-fixes=${CLANG_TIDY_FIXIT_DIR}/${TARGET}-${tidy_file}.yaml"
-                WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
-                COMMENT "clang-tidy: Running clang-tidy on target ${SOURCE}..."
-            )
-            add_dependencies(${tidy_target} ${TARGET})
-            add_dependencies(${tidy_target} tidy-base)
-            add_dependencies(tidy ${tidy_target})
+            if(NOT TARGET ${tidy_target})
+                add_custom_target(${tidy_target}
+                    # for some targets clang-tidy not able to get information from .clang-tidy
+                    DEPENDS ${SOURCE}
+                    COMMAND ${CLANG_TIDY_COMMAND} "-config=\{CheckOptions: \[\{key: bugprone-reserved-identifier.AllowedIdentifiers,value: __HIP_PLATFORM_HCC__\; __HIP_PLATFORM_AMD__\; __HIP_ROCclr__\}\]\}" ${SOURCE} "-export-fixes=${CLANG_TIDY_FIXIT_DIR}/${TARGET}-${tidy_file}.yaml"
+                    WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+                    COMMENT "clang-tidy: Running clang-tidy on target ${SOURCE}..."
+                )
+                add_dependencies(${tidy_target} ${TARGET})
+                add_dependencies(${tidy_target} tidy-base)
+                add_dependencies(tidy ${tidy_target})
+            endif()
         endif()
     endforeach()
 endfunction()

example/65_gemm_multiply_multiply/CMakeLists.txt

 add_example_executable(example_gemm_multiply_multiply_xdl_fp8 gemm_multiply_multiply_xdl_fp8.cpp)
 add_example_executable(example_gemm_multiply_multiply_xdl_fp8_ab_scale gemm_multiply_multiply_xdl_fp8_ab_scale.cpp)
+add_example_executable(example_gemm_multiply_multiply_xdl_fp8_b_scale gemm_multiply_multiply_xdl_fp8_b_scale.cpp)
 add_example_executable(example_gemm_add_add_xdl_fp16 gemm_add_add_xdl_fp16.cpp)

script/cmake-ck-dev.sh

@@ -14,8 +14,8 @@ fi
 cmake                                                                                             \
 -D CMAKE_PREFIX_PATH=/opt/rocm                                                                    \
 -D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc                                                         \
--D CMAKE_CXX_FLAGS="-Xclang -mllvm -Xclang -enable-post-misched=0 -std=c++17 -O3 -ftemplate-backtrace-limit=0  -fPIE  -Wno-gnu-line-marker"     \
--D CMAKE_BUILD_TYPE=Release                                                                       \
+-D CMAKE_CXX_FLAGS="-Xclang -mllvm -Xclang -enable-post-misched=0 -std=c++17 -O0 -ftemplate-backtrace-limit=0  -fPIE  -Wno-gnu-line-marker"     \
+-D CMAKE_BUILD_TYPE=Debug                                                                       \
 -D BUILD_DEV=ON                                                                                   \
 -D GPU_TARGETS=$GPU_TARGETS                                                                       \
 -D CMAKE_VERBOSE_MAKEFILE:BOOL=ON                                                                 \