Merge remote-tracking branch 'origin/develop' into batched_gemm_c_permute

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f64_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f64_f64_f64.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i32_i8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i8_i8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_b16_f32_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f16_f32_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f32_f32_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f32_f64_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f64_f64_f64.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_b16_f32_b16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f16_f16_f16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f16_f32_f16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f32_f32_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f32_f64_f32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f64_f64_f64.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_i8_i32_i8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_i8_i8_i8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp"
 
 namespace ck {

library/src/utility/conv_util.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "ck/library/utility/conv_util.hpp"
 

profiler/CMakeLists.txt

@@ -6,6 +6,7 @@ include_directories(BEFORE
 set(PROFILER_SOURCE
     src/profiler.cpp
     src/profile_gemm.cpp
+    src/profile_gemm_splitk.cpp
     src/profile_gemm_bias_2d.cpp
     src/profile_gemm_bias_relu.cpp
     src/profile_gemm_bias_relu_add.cpp
@@ -21,27 +22,30 @@ set(PROFILER_SOURCE
     src/profile_conv_bwd_weight.cpp
     src/profile_batched_gemm_reduce.cpp
     src/profile_gemm_add_add_fastgelu.cpp
+    src/profile_normalization.cpp
 )
 
 add_executable(ckProfiler ${PROFILER_SOURCE})
 
 target_link_libraries(ckProfiler PRIVATE host_tensor)
 target_link_libraries(ckProfiler PRIVATE conv_util)
-target_link_libraries(ckProfiler PRIVATE device_gemm_reduce_instance)
-target_link_libraries(ckProfiler PRIVATE device_gemm_bias_add_reduce_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_instance)
+target_link_libraries(ckProfiler PRIVATE device_gemm_splitk_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_bias2d_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_bias_relu_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_bias_relu_add_instance)
+target_link_libraries(ckProfiler PRIVATE device_gemm_reduce_instance)
+target_link_libraries(ckProfiler PRIVATE device_gemm_bias_add_reduce_instance)
+target_link_libraries(ckProfiler PRIVATE device_gemm_add_add_fastgelu_instance)
 target_link_libraries(ckProfiler PRIVATE device_batched_gemm_instance)
+target_link_libraries(ckProfiler PRIVATE device_batched_gemm_reduce_instance)
+target_link_libraries(ckProfiler PRIVATE device_grouped_gemm_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv1d_fwd_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv3d_fwd_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_bias_relu_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_bias_relu_add_instance)
 target_link_libraries(ckProfiler PRIVATE device_convnd_bwd_data_instance)
-target_link_libraries(ckProfiler PRIVATE device_reduce_instance)
-target_link_libraries(ckProfiler PRIVATE device_grouped_gemm_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_bwd_weight_instance)
-target_link_libraries(ckProfiler PRIVATE device_batched_gemm_reduce_instance)
-target_link_libraries(ckProfiler PRIVATE device_gemm_add_add_fastgelu_instance)
+target_link_libraries(ckProfiler PRIVATE device_normalization_instance)
+target_link_libraries(ckProfiler PRIVATE device_reduce_instance)

profiler/include/data_type_enum.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 namespace ck {