Modularize ckProfiler operations (#514)

* Re-structure ckProfiler source files

* Rename profiler.cpp to main.cpp

* Modularize ckProfiler operations

* Add description for profiler operations

* Use longer name to avoid name collision

* Use macro to delay expansion

* Use std::move() to avoid object copying

* Prohibit users from calling dtor

* Use macro to eliminate redundant code

* Make friend function hidden

* Add missing include directive <iostream>

* Fix wrong include directives

* Remove int8 from batchnorm-forward instances since it is not needed for forward training and could fail test

Co-authored-by: Qianfeng Zhang <Qianfeng.Zhang@amd.com>

[ROCm/composable_kernel commit: 8784a72e23]

test/CMakeLists.txt

@@ -1,5 +1,6 @@
 include_directories(BEFORE
     ${PROJECT_SOURCE_DIR}/
+    ${PROJECT_SOURCE_DIR}/profiler/include
 )
 
 include(googletest)

test/batched_gemm/batched_gemm_bf16.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_batched_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_impl.hpp"
 
 namespace {
 using ADataType = ck::bhalf_t;

test/batched_gemm/batched_gemm_fp16.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_batched_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_impl.hpp"
 
 namespace {
 using ADataType = ck::half_t;

test/batched_gemm/batched_gemm_fp32.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_batched_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_impl.hpp"
 
 namespace {
 using ADataType = float;

test/batched_gemm/batched_gemm_int8.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_batched_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_impl.hpp"
 
 namespace {
 using ADataType = int8_t;

test/batched_gemm_gemm/test_batched_gemm_gemm_util.hpp

@@ -6,7 +6,7 @@
 #include <vector>
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_batched_gemm_gemm_xdl_cshuffle.hpp"
-#include "profiler/include/profile_batched_gemm_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_gemm_impl.hpp"
 
 using ck::tensor_operation::device::GemmSpecialization;
 

test/batched_gemm_reduce/batched_gemm_reduce_fp16.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_batched_gemm_reduce_impl.hpp"
+#include "profiler/profile_batched_gemm_reduce_impl.hpp"
 
 int main()
 {

test/batched_gemm_softmax_gemm/test_batched_gemm_softmax_gemm_util.hpp

@@ -6,7 +6,7 @@
 #include <vector>
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp"
-#include "profiler/include/profile_batched_gemm_softmax_gemm_impl.hpp"
+#include "profiler/profile_batched_gemm_softmax_gemm_impl.hpp"
 using ck::tensor_operation::device::GemmSpecialization;
 
 template <ck::index_t N>

test/batched_gemm_softmax_gemm_permute/test_batched_gemm_softmax_gemm_permute_util.hpp

@@ -7,7 +7,7 @@
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp"
-#include "profiler/include/profile_batched_gemm_softmax_gemm_permute_impl.hpp"
+#include "profiler/profile_batched_gemm_softmax_gemm_permute_impl.hpp"
 
 using ck::tensor_operation::device::GemmSpecialization;
 using ck::tensor_operation::device::MaskingSpecialization;

test/batchnorm/batchnorm_bwd_rank_4.cpp

@@ -8,7 +8,7 @@
 #include <tuple>
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_batchnorm_backward_impl.hpp"
+#include "profiler/profile_batchnorm_backward_impl.hpp"
 
 using F16  = ck::half_t;
 using F32  = float;

test/batchnorm/batchnorm_fwd_rank_4.cpp

@@ -8,7 +8,7 @@
 #include <tuple>
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_batchnorm_forward_impl.hpp"
+#include "profiler/profile_batchnorm_forward_impl.hpp"
 
 using F16  = ck::half_t;
 using F32  = float;

test/convnd_bwd_data/convnd_bwd_data.cpp

@@ -8,7 +8,7 @@
 #include <tuple>
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_conv_bwd_data_impl.hpp"
+#include "profiler/profile_conv_bwd_data_impl.hpp"
 
 template <typename Tuple>
 class TestConvndBwdData : public ::testing::Test

test/convnd_fwd/convnd_fwd.cpp

@@ -8,7 +8,7 @@
 #include <tuple>
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_conv_fwd_impl.hpp"
+#include "profiler/profile_conv_fwd_impl.hpp"
 
 template <typename Tuple>
 class TestConvndFwd : public ::testing::Test

test/elementwise_normalization/test_elementwise_layernorm_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/include/profile_elementwise_layernorm_impl.hpp"
+#include "profiler/profile_elementwise_layernorm_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/gemm_reduce/gemm_reduce_fp16.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_gemm_reduce_impl.hpp"
+#include "profiler/profile_gemm_reduce_impl.hpp"
 
 int main()
 {

test/grouped_convnd_bwd_weight/grouped_convnd_bwd_weight.cpp

@@ -9,7 +9,7 @@
 
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_grouped_conv_bwd_weight_impl.hpp"
+#include "profiler/profile_grouped_conv_bwd_weight_impl.hpp"
 
 template <typename Tuple>
 class TestGroupedConvndBwdWeight : public ::testing::Test

test/grouped_convnd_fwd/grouped_convnd_fwd.cpp

@@ -7,7 +7,7 @@
 #include <vector>
 #include <gtest/gtest.h>
 
-#include "profiler/include/profile_grouped_conv_fwd_impl.hpp"
+#include "profiler/profile_grouped_conv_fwd_impl.hpp"
 
 class TestGroupedConvNdFwd : public ::testing::Test
 {

test/grouped_gemm/grouped_gemm_fp16.cpp

@@ -3,7 +3,7 @@
 
 #include <iostream>
 
-#include "profiler/include/profile_grouped_gemm_impl.hpp"
+#include "profiler/profile_grouped_gemm_impl.hpp"
 
 namespace {
 

test/normalization/test_groupnorm_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/include/profile_groupnorm_impl.hpp"
+#include "profiler/profile_groupnorm_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_groupnorm_fp32.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/include/profile_groupnorm_impl.hpp"
+#include "profiler/profile_groupnorm_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_layernorm2d_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/include/profile_layernorm_impl.hpp"
+#include "profiler/profile_layernorm_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_layernorm2d_fp32.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/include/profile_layernorm_impl.hpp"
+#include "profiler/profile_layernorm_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/reduce/reduce_no_index.cpp

@@ -4,7 +4,7 @@
 #include <getopt.h>
 
 #include "ck/library/utility/host_common_util.hpp"
-#include "profiler/include/profile_reduce_impl.hpp"
+#include "profiler/profile_reduce_impl.hpp"
 
 using namespace ck;
 

test/reduce/reduce_with_index.cpp

@@ -4,7 +4,7 @@
 #include <getopt.h>
 
 #include "ck/library/utility/host_common_util.hpp"
-#include "profiler/include/profile_reduce_impl.hpp"
+#include "profiler/profile_reduce_impl.hpp"
 
 using namespace ck;
 

test/softmax/test_softmax_util.hpp

@@ -13,7 +13,7 @@
 #include "ck/tensor_operation/gpu/device/impl/device_softmax_impl.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 #include "include/ck/utility/data_type.hpp"
-#include "profiler/include/profile_softmax_impl.hpp"
+#include "profiler/profile_softmax_impl.hpp"
 
 namespace ck {