diff --git a/CMakeLists.txt b/CMakeLists.txt
index 4e12462a41..e6e296af2d 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -604,22 +604,22 @@ ENDFOREACH()
 add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
 add_subdirectory(library)
 
-if(NOT GPU_ARCHS AND USER_GPU_TARGETS)
-   rocm_package_setup_component(tests
-        LIBRARY_NAME composablekernel
-        PACKAGE_NAME tests # Prevent -static suffix on package name
-   )
+# if(NOT GPU_ARCHS AND USER_GPU_TARGETS)
+#    rocm_package_setup_component(tests
+#         LIBRARY_NAME composablekernel
+#         PACKAGE_NAME tests # Prevent -static suffix on package name
+#    )
 
-   rocm_package_setup_component(examples
-        LIBRARY_NAME composablekernel
-        PACKAGE_NAME examples
-   )
-   add_subdirectory(example)
-   add_subdirectory(tile_engine)
-   if(BUILD_TESTING)
-       add_subdirectory(test)
-   endif()
-endif()
+#    rocm_package_setup_component(examples
+#         LIBRARY_NAME composablekernel
+#         PACKAGE_NAME examples
+#    )
+#    add_subdirectory(example)
+#    add_subdirectory(tile_engine)
+#    if(BUILD_TESTING)
+#        add_subdirectory(test)
+#    endif()
+# endif()
 
 rocm_package_setup_component(profiler
     LIBRARY_NAME composablekernel
diff --git a/library/src/tensor_operation_instance/gpu/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
index a79aa99bfc..63b36f630b 100755
--- a/library/src/tensor_operation_instance/gpu/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
@@ -1,5 +1,17 @@
 function(add_instance_library INSTANCE_NAME)
     message("adding instance ${INSTANCE_NAME}")
+    if(NOT "${INSTANCE_NAME}" MATCHES "device_gemm_add_multiply")
+        foreach(source IN LISTS ARGN)
+            list(REMOVE_ITEM ARGN "${source}")
+        endforeach()
+    else()
+        foreach(source IN LISTS ARGN)
+            if(NOT "${source}" MATCHES "device_gemm_add_multiply")
+                message("removing instance ${source} ")
+                list(REMOVE_ITEM ARGN "${source}")
+            endif()
+        endforeach()
+    endif()
     set(result 1)
     if(DEFINED DTYPES)
         foreach(source IN LISTS ARGN)
@@ -183,8 +195,7 @@ function(add_instance_library INSTANCE_NAME)
     set(result ${result} PARENT_SCOPE)
 endfunction(add_instance_library INSTANCE_NAME)
 
-
-file(GLOB dir_list LIST_DIRECTORIES true *)
+file(GLOB dir_list LIST_DIRECTORIES true gemm_add_multiply)
 set(CK_DEVICE_OTHER_INSTANCES)
 set(CK_DEVICE_GEMM_INSTANCES)
 set(CK_DEVICE_CONV_INSTANCES)
@@ -352,6 +363,7 @@ if(CK_DEVICE_OTHER_INSTANCES)
             DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
         )
 endif()
+message("CK_DEVICE_GEMM_INSTANCES: ${CK_DEVICE_GEMM_INSTANCES}")
 if(CK_DEVICE_GEMM_INSTANCES)
         add_library(device_gemm_operations ${CK_DEVICE_GEMM_INSTANCES})
         add_library(composablekernels::device_gemm_operations ALIAS device_gemm_operations)
diff --git a/profiler/src/CMakeLists.txt b/profiler/src/CMakeLists.txt
index 39f47dd84f..e0c144c2a3 100644
--- a/profiler/src/CMakeLists.txt
+++ b/profiler/src/CMakeLists.txt
@@ -1,101 +1,101 @@
 # ckProfiler
 set(PROFILER_SOURCES
     profiler.cpp
-    profile_gemm.cpp
-    profile_reduce.cpp
-    profile_groupnorm_bwd_data.cpp
-    profile_groupnorm_fwd.cpp
-    profile_layernorm_bwd_data.cpp
-    profile_layernorm_bwd_gamma_beta.cpp
-    profile_groupnorm_bwd_gamma_beta.cpp
-    profile_layernorm_fwd.cpp
-    profile_max_pool2d_fwd.cpp
-    profile_pool3d_fwd.cpp
-    profile_avg_pool3d_bwd.cpp
-    profile_max_pool3d_bwd.cpp
-    profile_avg_pool2d_bwd.cpp
-    profile_max_pool2d_bwd.cpp
-    profile_softmax.cpp
-    profile_batchnorm_fwd.cpp
-    profile_batchnorm_bwd.cpp
-    profile_batchnorm_infer.cpp
-    profile_conv_tensor_rearrange.cpp
-    profile_transpose.cpp
-    profile_permute_scale.cpp
+    # profile_gemm.cpp
+    # profile_reduce.cpp
+    # profile_groupnorm_bwd_data.cpp
+    # profile_groupnorm_fwd.cpp
+    # profile_layernorm_bwd_data.cpp
+    # profile_layernorm_bwd_gamma_beta.cpp
+    # profile_groupnorm_bwd_gamma_beta.cpp
+    # profile_layernorm_fwd.cpp
+    # profile_max_pool2d_fwd.cpp
+    # profile_pool3d_fwd.cpp
+    # profile_avg_pool3d_bwd.cpp
+    # profile_max_pool3d_bwd.cpp
+    # profile_avg_pool2d_bwd.cpp
+    # profile_max_pool2d_bwd.cpp
+    # profile_softmax.cpp
+    # profile_batchnorm_fwd.cpp
+    # profile_batchnorm_bwd.cpp
+    # profile_batchnorm_infer.cpp
+    # profile_conv_tensor_rearrange.cpp
+    # profile_transpose.cpp
+    # profile_permute_scale.cpp
 )
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
-  if(DTYPES MATCHES "fp32" OR DTYPES MATCHES "fp64" OR NOT DEFINED DTYPES)
-    list(APPEND PROFILER_SOURCES profile_contraction_bilinear.cpp)
-    list(APPEND PROFILER_SOURCES profile_contraction_scale.cpp)
-  endif()
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    list(APPEND PROFILER_SOURCES profile_gemm_reduce.cpp)
-    list(APPEND PROFILER_SOURCES profile_batched_gemm_gemm.cpp)
-    list(APPEND PROFILER_SOURCES profile_batched_gemm_add_relu_gemm_add.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_add.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_add_add_fastgelu.cpp)
-    list(APPEND PROFILER_SOURCES profile_grouped_gemm.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_streamk.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_fastgelu.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_add_relu.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_add_silu.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_add_relu_add_layernorm.cpp)
-    list(APPEND PROFILER_SOURCES profile_grouped_gemm_fixed_nk.cpp)
-    list(APPEND PROFILER_SOURCES profile_grouped_gemm_fastgelu.cpp)
-    list(APPEND PROFILER_SOURCES profile_grouped_gemm_tile_loop.cpp)
-    list(APPEND PROFILER_SOURCES profile_grouped_gemm_multiply_tile_loop.cpp)
-  endif()
-  list(APPEND PROFILER_SOURCES profile_gemm_multiply_add.cpp)
-  if(SUPPORTED_GPU_TARGETS MATCHES "gfx94")
-    list(APPEND PROFILER_SOURCES profile_gemm_multiply_multiply.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_multiply_multiply_wp.cpp)
-    list(APPEND PROFILER_SOURCES profile_gemm_ab_scale.cpp)
-  endif()
-  list(APPEND PROFILER_SOURCES profile_batched_gemm.cpp)
-  list(APPEND PROFILER_SOURCES profile_batched_gemm_reduce.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_add_multiply.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_bias_add_reduce.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_splitk.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_b_scale.cpp)
-  list(APPEND PROFILER_SOURCES profile_batched_gemm_b_scale.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_universal_batched.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_universal_reduce.cpp)
-  list(APPEND PROFILER_SOURCES profile_gemm_universal_streamk.cpp)
-  list(APPEND PROFILER_SOURCES profile_conv_fwd_bias_relu.cpp)
-  list(APPEND PROFILER_SOURCES profile_conv_fwd_bias_relu_add.cpp)
-  list(APPEND PROFILER_SOURCES profile_conv_bwd_data.cpp)
-  list(APPEND PROFILER_SOURCES profile_conv_fwd.cpp)
-  list(APPEND PROFILER_SOURCES profile_grouped_conv_fwd_outelementop.cpp)
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
+#   if(DTYPES MATCHES "fp32" OR DTYPES MATCHES "fp64" OR NOT DEFINED DTYPES)
+#     list(APPEND PROFILER_SOURCES profile_contraction_bilinear.cpp)
+#     list(APPEND PROFILER_SOURCES profile_contraction_scale.cpp)
+#   endif()
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+#     list(APPEND PROFILER_SOURCES profile_gemm_reduce.cpp)
+#     list(APPEND PROFILER_SOURCES profile_batched_gemm_gemm.cpp)
+#     list(APPEND PROFILER_SOURCES profile_batched_gemm_add_relu_gemm_add.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add_add_fastgelu.cpp)
+#     list(APPEND PROFILER_SOURCES profile_grouped_gemm.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_streamk.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_fastgelu.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add_relu.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add_silu.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add_relu_add_layernorm.cpp)
+#     list(APPEND PROFILER_SOURCES profile_grouped_gemm_fixed_nk.cpp)
+#     list(APPEND PROFILER_SOURCES profile_grouped_gemm_fastgelu.cpp)
+#     list(APPEND PROFILER_SOURCES profile_grouped_gemm_tile_loop.cpp)
+#     list(APPEND PROFILER_SOURCES profile_grouped_gemm_multiply_tile_loop.cpp)
+#   endif()
+#   list(APPEND PROFILER_SOURCES profile_gemm_multiply_add.cpp)
+#   if(SUPPORTED_GPU_TARGETS MATCHES "gfx94")
+#     list(APPEND PROFILER_SOURCES profile_gemm_multiply_multiply.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_multiply_multiply_wp.cpp)
+#     list(APPEND PROFILER_SOURCES profile_gemm_ab_scale.cpp)
+#   endif()
+#   list(APPEND PROFILER_SOURCES profile_batched_gemm.cpp)
+#   list(APPEND PROFILER_SOURCES profile_batched_gemm_reduce.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_add_multiply.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_bias_add_reduce.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_splitk.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_b_scale.cpp)
+#   list(APPEND PROFILER_SOURCES profile_batched_gemm_b_scale.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_universal_batched.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_universal_reduce.cpp)
+#   list(APPEND PROFILER_SOURCES profile_gemm_universal_streamk.cpp)
+#   list(APPEND PROFILER_SOURCES profile_conv_fwd_bias_relu.cpp)
+#   list(APPEND PROFILER_SOURCES profile_conv_fwd_bias_relu_add.cpp)
+#   list(APPEND PROFILER_SOURCES profile_conv_bwd_data.cpp)
+#   list(APPEND PROFILER_SOURCES profile_conv_fwd.cpp)
+#   list(APPEND PROFILER_SOURCES profile_grouped_conv_fwd_outelementop.cpp)
 
-endif()
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+# #     list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
+#   endif()
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
-  if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
-    list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
+#   if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
+#     list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
+#   endif()
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12" OR SUPPORTED_GPU_TARGETS MATCHES "gfx9")
-  list(APPEND PROFILER_SOURCES profile_gemm_universal.cpp)
-  list(APPEND PROFILER_SOURCES profile_grouped_conv_fwd.cpp)
-  list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_data.cpp)
-  list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_weight.cpp)
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    list(APPEND PROFILER_SOURCES profile_gemm_add_fastgelu.cpp)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12" OR SUPPORTED_GPU_TARGETS MATCHES "gfx9")
+#   list(APPEND PROFILER_SOURCES profile_gemm_universal.cpp)
+#   list(APPEND PROFILER_SOURCES profile_grouped_conv_fwd.cpp)
+#   list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_data.cpp)
+#   list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_weight.cpp)
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+#     list(APPEND PROFILER_SOURCES profile_gemm_add_fastgelu.cpp)
+#   endif()
+# endif()
 
-if(DL_KERNELS)
-  list(APPEND PROFILER_SOURCES profile_batched_gemm_multi_d.cpp)
-  list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_weight.cpp)
-endif()
+# if(DL_KERNELS)
+#   list(APPEND PROFILER_SOURCES profile_batched_gemm_multi_d.cpp)
+#   list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_weight.cpp)
+# endif()
 
 set(PROFILER_EXECUTABLE ckProfiler)
 
@@ -108,102 +108,102 @@ if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600241132)
 endif()
 
 target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE utility getopt::getopt)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_fwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_bwd_data_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_bwd_gamma_beta_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_softmax_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_reduce_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batchnorm_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_pool2d_fwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_pool3d_fwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_avg_pool2d_bwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_avg_pool3d_bwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_max_pool_bwd_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_image_to_column_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_column_to_image_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_transpose_instance)
-target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_permute_scale_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_instance)
+target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_multiply_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_fwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_bwd_data_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_normalization_bwd_gamma_beta_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_softmax_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_reduce_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batchnorm_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_pool2d_fwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_pool3d_fwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_avg_pool2d_bwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_avg_pool3d_bwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_max_pool_bwd_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_image_to_column_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_column_to_image_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_transpose_instance)
+# target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_permute_scale_instance)
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
-  if(DTYPES MATCHES "fp32" OR DTYPES MATCHES "fp64" OR NOT DEFINED DTYPES)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_contraction_bilinear_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_contraction_scale_instance)
-  endif()
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_add_fastgelu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_fastgelu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_gemm_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_add_relu_gemm_add_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_streamk_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_silu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_add_layernorm_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fixed_nk_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fastgelu_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_tile_loop_instance)
-  endif()
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_reduce_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_add_instance)
-  if(SUPPORTED_GPU_TARGETS MATCHES "gfx94")
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_multiply_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_multiply_wp_instance)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_ab_scale_instance)
-  endif()
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_splitk_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_b_scale_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_b_scale_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_batched_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_reduce_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_streamk_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_multiply_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_reduce_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bias_add_reduce_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_bias_relu_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_bias_relu_add_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_fwd_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv1d_bwd_data_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv3d_bwd_data_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_bwd_data_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_bwd_weight_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_weight_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_convscale_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_convinvscale_instance)
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
+#   if(DTYPES MATCHES "fp32" OR DTYPES MATCHES "fp64" OR NOT DEFINED DTYPES)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_contraction_bilinear_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_contraction_scale_instance)
+#   endif()
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_add_fastgelu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_fastgelu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_gemm_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_add_relu_gemm_add_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_streamk_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_silu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_add_layernorm_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fixed_nk_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fastgelu_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_tile_loop_instance)
+#   endif()
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_reduce_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_add_instance)
+#   if(SUPPORTED_GPU_TARGETS MATCHES "gfx94")
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_multiply_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_multiply_multiply_wp_instance)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_ab_scale_instance)
+#   endif()
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_splitk_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_b_scale_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_b_scale_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_batched_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_reduce_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_streamk_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_reduce_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bias_add_reduce_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_bias_relu_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_fwd_bias_relu_add_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_fwd_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv1d_bwd_data_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv3d_bwd_data_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_conv2d_bwd_data_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_bwd_weight_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_weight_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_convscale_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_convinvscale_instance)
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
+#   endif()
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
-  if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
+#   if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
+#   endif()
+# endif()
 
-if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_data_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_data_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_fwd_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_weight_instance)
-  if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
-    target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance)
-  endif()
-endif()
+# if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_data_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_data_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_fwd_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_weight_instance)
+#   if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+#     target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance)
+#   endif()
+# endif()
 
-if(DL_KERNELS)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_multi_d_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_bwd_weight_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_weight_instance)
-  target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_weight_instance)
-endif()
+# if(DL_KERNELS)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_multi_d_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv1d_bwd_weight_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_weight_instance)
+#   target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_bwd_weight_instance)
+# endif()
 rocm_install(TARGETS ${PROFILER_EXECUTABLE} COMPONENT profiler)