code formatting

library/src/tensor_operation_instance/gpu/CMakeLists.txt

@@ -92,13 +92,12 @@ function(add_instance_library INSTANCE_NAME)
          message("removing gemm_universal_streamk_f8 instance ${source} ")
          list(REMOVE_ITEM ARGN "${source}")
     endif()
-    # Do not build gemm_universal_streamk_bf16 for any targets except gfx94 #Emin (atomicAdd Issue)
     endforeach()
-    foreach(source IN LISTS ARGN)
-    if(NOT INST_TARGETS MATCHES "gfx94" AND source MATCHES "gemm_xdl_universal_streamk" AND source MATCHES "_bf16_")
-         message("removing gemm_universal_streamk_bf16 instance ${source} ")
-         list(REMOVE_ITEM ARGN "${source}")
-    endif()
+    # foreach(source IN LISTS ARGN)
+    # if(NOT INST_TARGETS MATCHES "gfx94" AND source MATCHES "gemm_xdl_universal_streamk" AND source MATCHES "_bf16_")
+    #      message("removing gemm_universal_streamk_bf16 instance ${source} ")
+    #      list(REMOVE_ITEM ARGN "${source}")
+    # endif()
     endforeach()
     endif()
     #only continue if there are some source files left on the list
@@ -190,6 +189,10 @@ FOREACH(subdir_path ${dir_list})
             message("bf8 instance found!")
             set(add_inst 1)
         endif()
+        if(("${cmake_instance}" MATCHES "_bf16" OR "${cmake_instance}" MATCHES "_b16") AND DTYPES MATCHES "bf16")
+        message("bf16 instance found!")
+        set(add_inst 1)
+        endif()
         if(("${cmake_instance}" MATCHES "_fp16" OR "${cmake_instance}" MATCHES "_f16") AND DTYPES MATCHES "fp16")
             message("fp16 instance found!")
             set(add_inst 1)

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

@@ -101,9 +101,6 @@ list(APPEND GEMM_UNIVERSAL_STREAMK_INSTANCES
         device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_default_instance.cpp
         device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v1_kpadding_instance.cpp
         device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_default_instance.cpp
-        device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_kpadding_instance.cpp
-        
-        
-)
+        device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_mk_nk_mn_mem_v2_kpadding_instance.cpp)
 
 add_instance_library(device_gemm_universal_streamk_instance ${GEMM_UNIVERSAL_STREAMK_INSTANCES})

library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instance.cpp

@@ -9,15 +9,9 @@ namespace device {
 namespace instance {
 
 void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_default_instances(
-    std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
-                                                      Row,
-                                                      Row,
-                                                      BF16,
-                                                      BF16,
-                                                      BF16,
-                                                      PassThrough,
-                                                      PassThrough,
-                                                      PassThrough>>>& instances)
+    std::vector<std::unique_ptr<
+        DeviceGemm_Streamk_V2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& 
+        instances)
 {
     add_device_operation_instances(
         instances,

library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instance.cpp

@@ -9,19 +9,12 @@ namespace device {
 namespace instance {
 
 void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_kpadding_instances(
-    std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
-                                                      Row,
-                                                      Row,
-                                                      BF16,
-                                                      BF16,
-                                                      BF16,
-                                                      PassThrough,
-                                                      PassThrough,
-                                                      PassThrough>>>& instances)
+    std::vector<std::unique_ptr<
+        DeviceGemm_Streamk_V2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough,PassThrough>>>& 
+        instances)
 {
     add_device_operation_instances(
-        instances,
-        device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmKPadding>{});
+        instances, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmKPadding>{});
 }
 
 } // namespace instance

library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_bf16_bf16_bf16/device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instance.cpp

@@ -9,19 +9,11 @@ namespace device {
 namespace instance {
 
 void add_device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_mnkpadding_instances(
-    std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col,
-                                                      Row,
-                                                      Row,
-                                                      BF16,
-                                                      BF16,
-                                                      BF16,
-                                                      PassThrough,
-                                                      PassThrough,
-                                                      PassThrough>>>& instances)
+    std::vector<std::unique_ptr<DeviceGemm_Streamk_V2<Col, Row, Row, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>& 
+    instances)
 {
     add_device_operation_instances(
-        instances,
-        device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmMNKPadding>{});
+        instances, device_gemm_xdl_universal_streamk_bf16_bf16_bf16_km_kn_mn_comp_instances<GemmMNKPadding>{});
 }
 
 } // namespace instance