Merge branch 'develop' into aosewski/ggemm_multi_d2

test/grouped_convnd_fwd/CMakeLists.txt

 add_gtest_executable(test_grouped_convnd_fwd test_grouped_convnd_fwd.cpp)
 target_link_libraries(test_grouped_convnd_fwd PRIVATE utility device_grouped_conv1d_fwd_instance device_grouped_conv2d_fwd_instance device_grouped_conv3d_fwd_instance)
 
+add_gtest_executable(test_grouped_convnd_fwd_multi_ab_interface test_grouped_convnd_fwd_multi_ab_interface.cpp)
+target_link_libraries(test_grouped_convnd_fwd_multi_ab_interface PRIVATE utility)
+
+add_gtest_executable(test_grouped_convnd_fwd_multi_d_interface_compatibility test_grouped_convnd_fwd_multi_d_interface_compatibility.cpp)
+target_link_libraries(test_grouped_convnd_fwd_multi_d_interface_compatibility PRIVATE utility device_grouped_conv3d_fwd_instance)

test/grouped_convnd_fwd/test_grouped_convnd_fwd_multi_ab_interface.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <tuple>
+#include <vector>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp"
+
+#include "ck/host_utility/device_prop.hpp"
+
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/algorithm.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+
+#include <gtest/gtest.h>
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using ScaleAdd    = ck::tensor_operation::element_wise::ScaleAdd;
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+template <typename DataType,
+          typename InDataTypes,
+          typename WeiDataTypes,
+          typename InElementOp,
+          typename WeiElementOp>
+class TestGroupedConvndFwdMultiABInterfaceBase : public ::testing::Test
+{
+    protected:
+    static constexpr ck::index_t NDimSpatial = 3;
+    static constexpr ck::index_t NumAs       = 2;
+    static constexpr ck::index_t NumBs       = 2;
+    static constexpr auto ConvSpec =
+        ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+    static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+    using InLayout                 = ck::tensor_layout::convolution::GNDHWC;
+    using WeiLayout                = ck::tensor_layout::convolution::GKZYXC;
+    using OutLayout                = ck::tensor_layout::convolution::GNDHWK;
+    using OutElementOp             = PassThrough;
+
+    using DeviceGroupedConvNDMultiABFwdInstance =
+        ck::tensor_operation::device::DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle<
+            NDimSpatial,
+            InLayout,
+            WeiLayout,
+            ck::Tuple<>,
+            OutLayout,
+            InDataTypes,
+            WeiDataTypes,
+            DataType,
+            DataType,
+            ck::Tuple<>,
+            DataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            ConvSpec,    // ConvForwardSpecialization
+            GemmSpec,    // GemmSpecialization
+            1,           //
+            256,         // BlockSize
+            128,         // MPerBlock
+            256,         // NPerBlock
+            32,          // KPerBlock
+            8,           // AK1
+            8,           // BK1
+            32,          // MPerXdl
+            32,          // NPerXdl
+            2,           // MXdlPerWave
+            4,           // NXdlPerWave
+            S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+            S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+            S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+            2,           // ABlockTransferSrcVectorDim
+            8,           // ABlockTransferSrcScalarPerVector
+            8,           // ABlockTransferDstScalarPerVector_AK1
+            1,           // ABlockLdsExtraM
+            S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+            S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+            S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+            2,           // BBlockTransferSrcVectorDim
+            8,           // BBlockTransferSrcScalarPerVector
+            8,           // BBlockTransferDstScalarPerVector_BK1
+            1,           // BBlockLdsExtraN
+            1,
+            1,
+            S<1, 32, 1, 8>,
+            8>;
+
+    const ck::utils::conv::ConvParam conv_param{
+        3, 1, 16, 16, 8, {3, 3, 3}, {17, 17, 17}, {2, 2, 2}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}};
+
+    void SetUp() override
+    {
+        if(!ck::is_xdl_supported())
+        {
+            GTEST_SKIP();
+        }
+    }
+
+    template <typename ADataType, typename BDataType>
+    bool Run(ADataType as, BDataType bs)
+    {
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_lengths{};
+        std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_strides{};
+        std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_lengths{};
+        std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_strides{};
+        std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_lengths{};
+        std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_strides{};
+        std::array<ck::index_t, NDimSpatial> conv_filter_strides{};
+        std::array<ck::index_t, NDimSpatial> conv_filter_dilations{};
+        std::array<ck::index_t, NDimSpatial> input_left_pads{};
+        std::array<ck::index_t, NDimSpatial> input_right_pads{};
+
+        auto copy = [](const auto& x, auto& y) { ck::ranges::copy(x, y.begin()); };
+
+        copy(in_g_n_c_wis_desc.GetLengths(), a_g_n_c_wis_lengths);
+        copy(in_g_n_c_wis_desc.GetStrides(), a_g_n_c_wis_strides);
+        copy(wei_g_k_c_xs_desc.GetLengths(), b_g_k_c_xs_lengths);
+        copy(wei_g_k_c_xs_desc.GetStrides(), b_g_k_c_xs_strides);
+        copy(out_g_n_k_wos_desc.GetLengths(), e_g_n_k_wos_lengths);
+        copy(out_g_n_k_wos_desc.GetStrides(), e_g_n_k_wos_strides);
+        copy(conv_param.conv_filter_strides_, conv_filter_strides);
+        copy(conv_param.conv_filter_dilations_, conv_filter_dilations);
+        copy(conv_param.input_left_pads_, input_left_pads);
+        copy(conv_param.input_right_pads_, input_right_pads);
+
+        std::array<const void*, 0> ds{};
+
+        // do Conv
+        auto conv     = DeviceGroupedConvNDMultiABFwdInstance{};
+        auto invoker  = conv.MakeInvoker();
+        auto argument = conv.MakeArgument(as,
+                                          bs,
+                                          ds,
+                                          nullptr,
+                                          a_g_n_c_wis_lengths,
+                                          a_g_n_c_wis_strides,
+                                          b_g_k_c_xs_lengths,
+                                          b_g_k_c_xs_strides,
+                                          {},
+                                          {},
+                                          e_g_n_k_wos_lengths,
+                                          e_g_n_k_wos_strides,
+                                          conv_filter_strides,
+                                          conv_filter_dilations,
+                                          input_left_pads,
+                                          input_right_pads,
+                                          InElementOp{},
+                                          WeiElementOp{},
+                                          OutElementOp{});
+
+        return conv.IsSupportedArgument(argument);
+    }
+};
+
+class TestGroupedConvndFwdMultiAInterface
+    : public TestGroupedConvndFwdMultiABInterfaceBase<float,
+                                                      ck::Tuple<float, float>,
+                                                      float,
+                                                      ScaleAdd,
+                                                      PassThrough>
+{
+};
+
+class TestGroupedConvndFwdMultiBInterface
+    : public TestGroupedConvndFwdMultiABInterfaceBase<float,
+                                                      float,
+                                                      ck::Tuple<float, float>,
+                                                      PassThrough,
+                                                      ScaleAdd>
+{
+};
+
+class TestGroupedConvndFwdMultiABInterface
+    : public TestGroupedConvndFwdMultiABInterfaceBase<float,
+                                                      ck::Tuple<float, float>,
+                                                      ck::Tuple<float, float>,
+                                                      ScaleAdd,
+                                                      ScaleAdd>
+{
+};
+
+class TestGroupedConvndFwdInterface
+    : public TestGroupedConvndFwdMultiABInterfaceBase<float, float, float, PassThrough, PassThrough>
+{
+};
+
+TEST_F(TestGroupedConvndFwdMultiAInterface, MultiA)
+{
+    std::array<const void*, NumAs> as{nullptr, nullptr};
+    const void* b = nullptr;
+
+    EXPECT_TRUE(this->template Run(as, b));
+}
+
+TEST_F(TestGroupedConvndFwdMultiBInterface, MultiB)
+{
+    const void* a = nullptr;
+    std::array<const void*, NumBs> bs{nullptr, nullptr};
+
+    EXPECT_TRUE(this->template Run(a, bs));
+}
+
+TEST_F(TestGroupedConvndFwdMultiABInterface, MultiAB)
+{
+    std::array<const void*, NumAs> as{nullptr, nullptr};
+    std::array<const void*, NumBs> bs{nullptr, nullptr};
+
+    EXPECT_TRUE(this->template Run(as, bs));
+}
+
+TEST_F(TestGroupedConvndFwdInterface, SingleAB)
+{
+    const void* a = nullptr;
+    const void* b = nullptr;
+
+    EXPECT_TRUE(this->template Run(a, b));
+}

test/grouped_convnd_fwd/test_grouped_convnd_fwd_multi_d_interface_compatibility.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <tuple>
+#include <vector>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp"
+#include "ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp"
+
+#include <gtest/gtest.h>
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+class TestGroupedConvndFwdMultiDInterfaceCompatibility : public ::testing::Test
+{
+    protected:
+    static constexpr ck::index_t NDimSpatial = 3;
+
+    using InDataType  = float;
+    using WeiDataType = float;
+    using OutDataType = float;
+    using InLayout    = ck::tensor_layout::convolution::GNDHWC;
+    using WeiLayout   = ck::tensor_layout::convolution::GKZYXC;
+    using OutLayout   = ck::tensor_layout::convolution::GNDHWK;
+
+    using DeviceOp = ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD<NDimSpatial,
+                                                                                 InLayout,
+                                                                                 WeiLayout,
+                                                                                 ck::Tuple<>,
+                                                                                 OutLayout,
+                                                                                 InDataType,
+                                                                                 WeiDataType,
+                                                                                 ck::Tuple<>,
+                                                                                 OutDataType,
+                                                                                 PassThrough,
+                                                                                 PassThrough,
+                                                                                 PassThrough>;
+
+    bool Run()
+    {
+        const auto op_ptrs = ck::tensor_operation::device::instance::DeviceOperationInstanceFactory<
+            DeviceOp>::GetInstances();
+        return op_ptrs.size() != 0;
+    }
+};
+
+TEST_F(TestGroupedConvndFwdMultiDInterfaceCompatibility, CompatibilityTest)
+{
+    EXPECT_TRUE(this->Run());
+}

test/grouped_gemm/test_grouped_gemm_interface.cpp

@@ -108,6 +108,10 @@ TEST_F(TestGGemmSplitKInterface_MKNKMN, KLoops)
 
     // kloops % 2
     Ks = std::vector<int>{256, 512, 320, 768};
+    EXPECT_FALSE(
+        DefaultGGemmInstance{}.IsSupported(Ms, Ns, Ks, StrideAs, StrideBs, StrideCs, kbatch));
+
+    Ks = std::vector<int>{256, 512, 384, 768};
     EXPECT_TRUE(
         DefaultGGemmInstance{}.IsSupported(Ms, Ns, Ks, StrideAs, StrideBs, StrideCs, kbatch));
 

test/normalization/CMakeLists.txt

-add_custom_target(test_normalization)
-add_gtest_executable(test_layernorm2d_fp32 test_layernorm2d_fp32.cpp)
-if(result EQUAL 0)
-  target_link_libraries(test_layernorm2d_fp32 PRIVATE utility device_normalization_instance)
-  add_dependencies(test_normalization test_layernorm2d_fp32)
-endif()
-add_gtest_executable(test_groupnorm_fp32 test_groupnorm_fp32.cpp)
-if(result EQUAL 0)
-  target_link_libraries(test_groupnorm_fp32 PRIVATE utility device_normalization_instance)
-  add_dependencies(test_normalization test_groupnorm_fp32)
-endif()
-add_gtest_executable(test_layernorm2d_fp16 test_layernorm2d_fp16.cpp)
-if(result EQUAL 0)
-  target_link_libraries(test_layernorm2d_fp16 PRIVATE utility device_normalization_instance)
-  add_dependencies(test_normalization test_layernorm2d_fp16)
-endif()
-add_gtest_executable(test_groupnorm_fp16 test_groupnorm_fp16.cpp)
-if(result EQUAL 0)
-  target_link_libraries(test_groupnorm_fp16 PRIVATE utility device_normalization_instance)
-  add_dependencies(test_normalization test_groupnorm_fp16)
-endif()

test/normalization_fwd/CMakeLists.txt

+add_custom_target(test_normalization_fwd)
+add_gtest_executable(test_layernorm2d_fwd_fp32 test_layernorm2d_fwd_fp32.cpp)
+if(result EQUAL 0)
+  target_link_libraries(test_layernorm2d_fwd_fp32 PRIVATE utility device_normalization_fwd_instance)
+  add_dependencies(test_normalization_fwd test_layernorm2d_fwd_fp32)
+endif()
+
+add_gtest_executable(test_groupnorm_fwd_fp32 test_groupnorm_fwd_fp32.cpp)
+if(result EQUAL 0)
+  target_link_libraries(test_groupnorm_fwd_fp32 PRIVATE utility device_normalization_fwd_instance)
+  add_dependencies(test_normalization_fwd test_groupnorm_fwd_fp32)
+endif()
+
+add_gtest_executable(test_layernorm2d_fwd_fp16 test_layernorm2d_fwd_fp16.cpp)
+if(result EQUAL 0)
+  target_link_libraries(test_layernorm2d_fwd_fp16 PRIVATE utility device_normalization_fwd_instance)
+  add_dependencies(test_normalization_fwd test_layernorm2d_fwd_fp16)
+endif()
+
+add_gtest_executable(test_layernorm4d_fwd_fp16 test_layernorm4d_fwd_fp16.cpp)
+if(result EQUAL 0)
+  target_link_libraries(test_layernorm4d_fwd_fp16 PRIVATE utility device_normalization_fwd_instance)
+  add_dependencies(test_normalization_fwd test_layernorm4d_fwd_fp16)
+endif()
+
+add_gtest_executable(test_groupnorm_fwd_fp16 test_groupnorm_fwd_fp16.cpp)
+if(result EQUAL 0)
+  target_link_libraries(test_groupnorm_fwd_fp16 PRIVATE utility device_normalization_fwd_instance)
+  add_dependencies(test_normalization_fwd test_groupnorm_fwd_fp16)
+endif()

test/normalization/test_groupnorm_fp16.cpp → test/normalization_fwd/test_groupnorm_fwd_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/profile_groupnorm_impl.hpp"
+#include "profiler/profile_groupnorm_fwd_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_groupnorm_fp32.cpp → test/normalization_fwd/test_groupnorm_fwd_fp32.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/profile_groupnorm_impl.hpp"
+#include "profiler/profile_groupnorm_fwd_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_layernorm2d_fp16.cpp → test/normalization_fwd/test_layernorm2d_fwd_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/profile_layernorm_impl.hpp"
+#include "profiler/profile_layernorm_fwd_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization/test_layernorm2d_fp32.cpp → test/normalization_fwd/test_layernorm2d_fwd_fp32.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include "gtest/gtest.h"
-#include "profiler/profile_layernorm_impl.hpp"
+#include "profiler/profile_layernorm_fwd_impl.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;

test/normalization_fwd/test_layernorm4d_fwd_fp16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "gtest/gtest.h"
+#include "profiler/profile_layernorm_fwd_impl.hpp"
+
+using F16 = ck::half_t;
+using F32 = float;
+using ck::index_t;
+
+template <typename Tuple>
+class TestLayernorm4d : public ::testing::Test
+{
+    protected:
+    using XDataType              = std::tuple_element_t<0, Tuple>;
+    using GammaDataType          = std::tuple_element_t<1, Tuple>;
+    using BetaDataType           = std::tuple_element_t<2, Tuple>;
+    using ComputeDataType        = std::tuple_element_t<3, Tuple>;
+    using YDataType              = std::tuple_element_t<4, Tuple>;
+    using SaveMeanInvStdDataType = std::tuple_element_t<5, Tuple>;
+
+    void Run()
+    {
+        // [N, D], reduce D
+        std::vector<std::vector<ck::index_t>> lengths = {
+            {1, 1, 1, 1}, {7, 7, 7, 7}, {256, 16, 16, 8}};
+
+        for(auto length : lengths)
+        {
+            bool success = ck::profiler::profile_layernorm_impl<XDataType,
+                                                                GammaDataType,
+                                                                BetaDataType,
+                                                                ComputeDataType,
+                                                                YDataType,
+                                                                SaveMeanInvStdDataType,
+                                                                true,
+                                                                4>(true, 2, false, false, length);
+            EXPECT_TRUE(success);
+        }
+    }
+};
+
+using KernelTypes = ::testing::Types<
+    // XDataType, GammaDataType, BetaDataType, ComputeDataType, YDataType>
+    std::tuple<F16, F16, F16, F32, F16, F32>>;
+
+TYPED_TEST_SUITE(TestLayernorm4d, KernelTypes);
+TYPED_TEST(TestLayernorm4d, Test_FP16) { this->Run(); }

test/transpose/CMakeLists.txt

+list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
+set(target 0)
+foreach(gpu IN LISTS GPU_TARGETS)
+ if(gpu IN_LIST gpu_list AND target EQUAL 0)
+   add_gtest_executable(test_transpose test_transpose.cpp)
+   target_link_libraries(test_transpose PRIVATE utility device_transpose_instance)
+   set(target 1)
+ endif()
+endforeach()

test/transpose/test_transpose.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <tuple>
+
+#include "gtest/gtest.h"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "test_transpose_util.hpp"
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <typename Tuple>
+class TestTranspose : public ::testing::Test
+{
+};
+
+// clang-format off
+using KernelTypes = ::testing::Types<
+    std::tuple<      F16,       F16>,
+    std::tuple<      F32,       F32>
+    >;
+// clang-format on
+
+TYPED_TEST_SUITE(TestTranspose, KernelTypes);
+
+//#include "test_transpose_ut_cases.inc"

test/transpose/test_transpose_ut_cases.inc

+#pragma once
+
+TYPED_TEST(TestTranspose, Test1)
+{
+    // for 16, 8, 16, 32, 8
+    std::vector<int> Ms{1, 2, 3, 4, 5, 6};
+    std::vector<index_t> lengths{16, 8, 16, 32, 8};
+    /**constexpr int N = 16;
+    constexpr int C = 8;
+    constexpr int D = 16;
+    constexpr int H = 32;
+    constexpr int W = 8;**/
+
+    this->Run();
+}
+
+
+TYPED_TEST(TestTranpose, Test2)
+{
+    std::vector<int> Ms{127, 255, 312, 799, 1573};
+    std::vector<index_t> lengths{16, 8, 16, 32, 16};
+    /**constexpr int N = 16;
+    constexpr int C = 8;
+    constexpr int D = 16;
+    constexpr int H = 32;
+    constexpr int W = 8;**/
+
+    this->Run();
+}
+

test/transpose/test_transpose_util.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <string>
+#include <sstream>
+#include <tuple>
+#include <vector>
+#include <gtest/gtest.h>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "include/ck/utility/data_type.hpp"
+#include "profiler/profile_transpose_impl.hpp"
+
+namespace ck {
+namespace test {
+
+template <typename Tuple>
+class TestTranspose : public testing::Test
+{
+    using F32 = float;
+
+    protected:
+    using ADataType = std::tuple_element_t<0, Tuple>;
+    using BDataType = std::tuple_element_t<1, Tuple>;
+
+    public:
+    static constexpr bool verify_              = true;
+    static constexpr int init_method_          = 1; // decimal value initialization
+    static constexpr bool log_                 = false;
+    static constexpr bool bench_               = false; // measure kernel performance
+    std::vector<std::vector<index_t>> lengths_ = {{16, 32, 16, 32, 16}, {16, 8, 16, 32, 8}};
+
+    void Run()
+    {
+        for(auto length : this->lengths_)
+        {
+            this->RunSingle(length);
+        }
+    }
+
+    void RunSingle()
+    {
+        bool pass = ck::profiler::profile_transpose_impl<ADataType, BDataType, 5>(
+            verify_, init_method_, log_, bench_, lengths_);
+        EXPECT_TRUE(pass);
+    }
+};
+
+} // namespace test
+} // namespace ck