Merge branch 'develop' into bmatrix_skip_lds

library/src/tensor_operation_instance/gpu/conv2d_fwd_bias_relu_add/CMakeLists.txt

@@ -2,9 +2,7 @@
 set(DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_c_shuffle_bias_relu_add_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
-add_library(device_conv2d_fwd_bias_relu_add_instance SHARED ${DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_bias_relu_add_instance PUBLIC)
+add_library(device_conv2d_fwd_bias_relu_add_instance OBJECT ${DEVICE_CONV2D_FWD_BIAS_RELU_ADD_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_bias_relu_add_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_bias_relu_add_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_bias_relu_add_instance)

library/src/tensor_operation_instance/gpu/conv2d_fwd_bias_relu_atomic_add/CMakeLists.txt

@@ -3,9 +3,7 @@ set(DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_c_shuffle_bias_relu_atomic_add_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
 
-add_library(device_conv2d_fwd_bias_relu_atomic_add_instance SHARED ${DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv2d_fwd_bias_relu_atomic_add_instance PUBLIC)
+add_library(device_conv2d_fwd_bias_relu_atomic_add_instance OBJECT ${DEVICE_CONV2D_FWD_BIAS_RELU_ATOMIC_ADD_INSTANCE_SOURCE}) 
 set_target_properties(device_conv2d_fwd_bias_relu_atomic_add_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv2d_fwd_bias_relu_atomic_add_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv2d_fwd_bias_relu_atomic_add_instance)

library/src/tensor_operation_instance/gpu/conv3d_fwd/CMakeLists.txt

@@ -5,9 +5,8 @@ set(DEVICE_CONV3D_FWD_INSTANCE_SOURCE
    device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
    device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp;
 )
-add_library(device_conv3d_fwd_instance SHARED ${DEVICE_CONV3D_FWD_INSTANCE_SOURCE}) 
+add_library(device_conv3d_fwd_instance OBJECT ${DEVICE_CONV3D_FWD_INSTANCE_SOURCE}) 
 target_compile_features(device_conv3d_fwd_instance PUBLIC)
 set_target_properties(device_conv3d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_conv3d_fwd_instance LIBRARY DESTINATION lib) 
 
 clang_tidy_check(device_conv3d_fwd_instance)

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

@@ -14,7 +14,7 @@ set(DEVICE_CONVND_BWD_DATA_INSTANCE_SOURCE
    device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp;
 ) 
 
-add_library(device_convnd_bwd_data_instance SHARED ${DEVICE_CONVND_BWD_DATA_INSTANCE_SOURCE})
+add_library(device_convnd_bwd_data_instance OBJECT ${DEVICE_CONVND_BWD_DATA_INSTANCE_SOURCE})
 target_compile_features(device_convnd_bwd_data_instance PUBLIC)
 set_target_properties(device_convnd_bwd_data_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
 install(TARGETS device_convnd_bwd_data_instance LIBRARY DESTINATION lib) 

library/src/tensor_operation_instance/gpu/device_conv2d.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+#include "host_interface.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_conv2d_fwd_instance {
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances);
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(
+    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances);
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(
+    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances);
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances);
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(
+    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances);
+
+} // namespace device_conv2d_fwd_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+struct DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl
+{
+    std::unique_ptr<DeviceConvFwdPtr_t::BaseArgument>
+    MakeArgumentPointer(void* in_ptr,
+                        void* wei_ptr,
+                        void* out_ptr,
+                        size_t N,
+                        size_t K,
+                        size_t C,
+                        std::vector<ck::index_t> input_spatial_lengths,
+                        std::vector<ck::index_t> filter_spatial_lengths,
+                        std::vector<ck::index_t> output_spatial_lengths,
+                        std::vector<ck::index_t> conv_filter_strides,
+                        std::vector<ck::index_t> conv_filter_dilations,
+                        std::vector<ck::index_t> input_left_pads,
+                        std::vector<ck::index_t> input_right_pads) const
+    {
+        return el->MakeArgumentPointer(in_ptr,
+                                       wei_ptr,
+                                       out_ptr,
+                                       N,
+                                       K,
+                                       C,
+                                       input_spatial_lengths,
+                                       filter_spatial_lengths,
+                                       output_spatial_lengths,
+                                       conv_filter_strides,
+                                       conv_filter_dilations,
+                                       input_left_pads,
+                                       input_right_pads,
+                                       PassThrough{},
+                                       PassThrough{},
+                                       PassThrough{});
+    }
+    std::unique_ptr<DeviceConvFwdPtr_t::BaseInvoker> MakeInvokerPointer() const
+    {
+        return el->MakeInvokerPointer();
+    }
+
+    std::string GetTypeString() { return el->GetTypeString(); }
+    bool IsSupportedArgument(const DeviceConvFwdPtr_t::BaseArgument* arg)
+    {
+        return el->IsSupportedArgument(arg);
+    }
+
+    ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough> el;
+};
+
+DeviceConvFwdPtr_t::DeviceConvFwdPtr_t() : pImpl(nullptr) {}
+DeviceConvFwdPtr_t::~DeviceConvFwdPtr_t()                    = default;
+DeviceConvFwdPtr_t::DeviceConvFwdPtr_t(DeviceConvFwdPtr_t&&) = default;
+DeviceConvFwdPtr_t::DeviceConvFwdPtr_t(DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl& other)
+    : pImpl(std::make_unique<DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl>(std::move(other)))
+{
+}
+
+std::unique_ptr<DeviceConvFwdPtr_t::BaseArgument>
+DeviceConvFwdPtr_t::MakeArgumentPointer(void* in_ptr,
+                                        void* wei_ptr,
+                                        void* out_ptr,
+                                        size_t N,
+                                        size_t K,
+                                        size_t C,
+                                        std::vector<ck::index_t> input_spatial_lengths,
+                                        std::vector<ck::index_t> filter_spatial_lengths,
+                                        std::vector<ck::index_t> output_spatial_lengths,
+                                        std::vector<ck::index_t> conv_filter_strides,
+                                        std::vector<ck::index_t> conv_filter_dilations,
+                                        std::vector<ck::index_t> input_left_pads,
+                                        std::vector<ck::index_t> input_right_pads) const
+{
+    return pImpl->MakeArgumentPointer(in_ptr,
+                                      wei_ptr,
+                                      out_ptr,
+                                      N,
+                                      K,
+                                      C,
+                                      input_spatial_lengths,
+                                      filter_spatial_lengths,
+                                      output_spatial_lengths,
+                                      conv_filter_strides,
+                                      conv_filter_dilations,
+                                      input_left_pads,
+                                      input_right_pads);
+}
+
+std::unique_ptr<DeviceConvFwdPtr_t::BaseInvoker> DeviceConvFwdPtr_t::MakeInvokerPointer() const
+{
+    return pImpl->MakeInvokerPointer();
+}
+
+std::string DeviceConvFwdPtr_t::GetTypeString() { return pImpl->GetTypeString(); }
+bool DeviceConvFwdPtr_t::IsSupportedArgument(const DeviceConvFwdPtr_t::BaseArgument* arg_ptr)
+{
+    return pImpl->IsSupportedArgument(arg_ptr);
+}
+
+using namespace ck::tensor_operation::device::device_conv2d_fwd_instance;
+void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances_t(
+    std::vector<DeviceConvFwdPtr_t>& instances)
+{
+    std::vector<
+        ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>
+        local_instances;
+    add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(local_instances);
+    for(auto& kinder : local_instances)
+    {
+        DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl tmp{std::move(kinder)};
+        instances.emplace_back(tmp);
+    }
+    return;
+}
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances_t(
+    std::vector<DeviceConvFwdPtr_t>& instances)
+{
+    std::vector<
+        ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>
+        local_instances;
+    add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(local_instances);
+    for(auto& kinder : local_instances)
+    {
+        DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl tmp{std::move(kinder)};
+        instances.emplace_back(tmp); // Perhaps we can do better
+    }
+    return;
+}
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances_t(
+    std::vector<DeviceConvFwdPtr_t>& instances)
+{
+    std::vector<
+        ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>
+        local_instances;
+    add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(local_instances);
+    for(auto& kinder : local_instances)
+    {
+        DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl tmp{std::move(kinder)};
+        instances.emplace_back(tmp); // Perhaps we can do better
+    }
+    return;
+}
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances_t(
+    std::vector<DeviceConvFwdPtr_t>& instances)
+{
+    std::vector<
+        ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>
+        local_instances;
+    add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(local_instances);
+    for(auto& kinder : local_instances)
+    {
+        DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl tmp{std::move(kinder)};
+        instances.emplace_back(tmp); // Perhaps we can do better
+    }
+    return;
+}
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances_t(
+    std::vector<DeviceConvFwdPtr_t>& instances)
+{
+    std::vector<
+        ck::tensor_operation::device::DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>
+        local_instances;
+    add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(local_instances);
+    for(auto& kinder : local_instances)
+    {
+        DeviceConvFwdPtr_t::DeviceConvFwdPtrImpl tmp{std::move(kinder)};
+        instances.emplace_back(tmp);
+    }
+    return;
+}

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

-# device_gemm_instance
 set(DEVICE_GEMM_INSTANCE_SOURCE
+   device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp;
+   device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp;
+   device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp;
+   device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp;
    device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp;
    device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp;
    device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp;
@@ -8,10 +11,10 @@ set(DEVICE_GEMM_INSTANCE_SOURCE
    device_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp;
    device_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp;
    device_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp;
-   device_gemm_xdl_c_shuffle_int8_int8_int8_mk_kn_mn_instance.cpp;
-   device_gemm_xdl_c_shuffle_int8_int8_int8_mk_nk_mn_instance.cpp;
-   device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instance.cpp;
-   device_gemm_xdl_c_shuffle_int8_int8_int8_km_nk_mn_instance.cpp;
+   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp;
+   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp;
+   device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp;
+   device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp;
    device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp;
    device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp;
    device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp;
@@ -33,12 +36,21 @@ set(DEVICE_GEMM_INSTANCE_SOURCE
    device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instance.cpp;
    device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instance.cpp;
    device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp;
+   device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp;
+   device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp;
+   device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp;
+   device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp;
+   device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp;
+   device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp;
+   device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp;
+   device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp;
+   device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp;
+   device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp;
+   device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp;
+   device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp;
 )
 
-add_library(device_gemm_instance SHARED ${DEVICE_GEMM_INSTANCE_SOURCE})
+add_library(device_gemm_instance OBJECT ${DEVICE_GEMM_INSTANCE_SOURCE})
 
 target_compile_features(device_gemm_instance PUBLIC)
 set_target_properties(device_gemm_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-install(TARGETS device_gemm_instance LIBRARY DESTINATION lib)
-
-clang_tidy_check(device_gemm_instance)

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[k, n] = c[m, n]
+using device_gemm_dl_f16_f16_f16_km_kn_mn_instances = std::tuple<
+    // clang-format off
+        // #########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        // #########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        // #########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|               Order|                 |                   |
+        // #########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F16,   F16,   F16,     F32,     Col,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  2,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 2>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 2>,      S<2, 1, 4, 2>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_f16_f16_f16_km_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f16_f16_f16_km_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[n, k] = c[m, n]
+using device_gemm_dl_f16_f16_f16_km_nk_mn_instances = std::tuple<
+    // clang-format off
+        // #########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        // #########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        // #########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        // #########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F16,   F16,   F16,     F32,     Col,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  2,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 2>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 2>,      S<8, 1, 1, 2>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 2>,      S<1, 2, 0, 3>,        S<1, 1, 1, 2>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_f16_f16_f16_km_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f16_f16_f16_km_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
+using device_gemm_dl_f16_f16_f16_mk_kn_mn_instances = std::tuple<
+    // clang-format off
+        // #########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        // #########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        // #########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        // #########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F16,   F16,   F16,     F32,     Row,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  2,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 2>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 2>,      S<1, 2, 0, 3>,        S<1, 1, 1, 2>,      S<2, 1, 4, 2>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_f16_f16_f16_mk_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f16_f16_f16_mk_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[n, k] = c[m, n]
+using device_gemm_dl_f16_f16_f16_mk_nk_mn_instances =
+    std::tuple<
+        // clang-format off
+        //  ########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        //  ########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        //  ########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        //  ########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F16,   F16,   F16,     F32,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  2,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 2>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 2>,      S<1, 2, 0, 3>,        S<1, 1, 1, 2>,      S<8, 1, 1, 2>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 2>,      S<1, 2, 0, 3>,        S<1, 1, 1, 2>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+        // clang-format on
+        >;
+
+void add_device_gemm_dl_f16_f16_f16_mk_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f16_f16_f16_mk_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[k, n] = c[m, n]
+using device_gemm_dl_f32_f32_f32_km_kn_mn_instances = std::tuple<
+    // clang-format off
+        //  ########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|      BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        //  ########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|     DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        //  ########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder| Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        //  ########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                    |                    |                 |                   |
+        DeviceGemmDl<   F32,   F32,   F32,     F32,     Col,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  1,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 1>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 1>,      S<2, 1, 4, 1>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,       S<1, 1, 4, 1>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_f32_f32_f32_km_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f32_f32_f32_km_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[n, k] = c[m, n]
+using device_gemm_dl_f32_f32_f32_km_nk_mn_instances =
+    std::tuple<
+        // clang-format off
+        //  ########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        //  ########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        //  ########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        //  ########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F32,   F32,   F32,     F32,     Col,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  1,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 1>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 1>,      S<8, 1, 1, 1>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 1>,      S<1, 2, 0, 3>,        S<1, 1, 1, 1>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+        // clang-format on
+        >;
+
+void add_device_gemm_dl_f32_f32_f32_km_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f32_f32_f32_km_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
+using device_gemm_dl_f32_f32_f32_mk_kn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //  ########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        //  ########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        //  ########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        //  ########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F32,   F32,   F32,     F32,     Row,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  1,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 1>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 1>,      S<1, 2, 0, 3>,        S<1, 1, 1, 1>,      S<2, 1, 4, 1>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 1>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+        // clang-format on
+        >;
+
+void add_device_gemm_dl_f32_f32_f32_mk_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f32_f32_f32_mk_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[n, k] = c[m, n]
+using device_gemm_dl_f32_f32_f32_mk_nk_mn_instances =
+    std::tuple<
+        // clang-format off
+        //  ########| AData| BData| CData| AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        //  ########|  Type|  Type|  Type|    Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        //  ########|      |      |      |        |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        //  ########|      |      |      |        |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl<   F32,   F32,   F32,     F32,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  1,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 1>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 1>,      S<1, 2, 0, 3>,        S<1, 1, 1, 1>,      S<8, 1, 1, 1>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 1>,      S<1, 2, 0, 3>,        S<1, 1, 1, 1>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+        // clang-format on
+        >;
+
+void add_device_gemm_dl_f32_f32_f32_mk_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_f32_f32_f32_mk_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[k, n] = c[m, n]
+using device_gemm_dl_i8_i8_i8_km_kn_mn_instances = std::tuple<
+    // clang-format off
+        // #########|  AData|   BData|   CData|    AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer|  CThreadTransfer|    CThreadTransfer|
+        // #########|   Type|    Type|    Type|       Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess|  SrcDstVectorDim| DstScalarPerVector|
+        // #########|       |        |        |           |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                 |                   |
+        // #########|       |        |        |           |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                 |                   |
+        DeviceGemmDl< int8_t,  int8_t,  int8_t,    int32_t,     Col,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  4,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 4>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 4>,      S<2, 1, 4, 4>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 4>, S<0, 1, 2, 3, 4, 5>,                5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_i8_i8_i8_km_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_i8_i8_i8_km_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[k, m] * b[n, k] = c[m, n]
+using device_gemm_dl_i8_i8_i8_km_nk_mn_instances = std::tuple<
+    // clang-format off
+        // #########|  AData|   BData|   CData|    AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|      BBlockTransfer|     CThreadTransfer| CThreadTransfer|    CThreadTransfer|
+        // #########|   Type|    Type|    Type|       Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|     DstVectorTensor|        SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
+        // #########|       |        |        |           |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder| Lengths_K0_N0_N1_K1|               Order|                |                   |
+        // #########|       |        |        |           |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                    |                    |                |                   |
+        DeviceGemmDl< int8_t,  int8_t,  int8_t,    int32_t,     Col,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  4,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<2, 1, 4, 4>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 4>,      S<8, 1, 1, 4>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 4>,      S<1, 2, 0, 3>,       S<1, 1, 1, 4>, S<0, 1, 2, 3, 4, 5>,               5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_i8_i8_i8_km_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_i8_i8_i8_km_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
+using device_gemm_dl_i8_i8_i8_mk_kn_mn_instances = std::tuple<
+    // clang-format off
+        // #########|  AData|   BData|   CData|    AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|       BBlockTransfer|     CThreadTransfer| CThreadTransfer|    CThreadTransfer|
+        // #########|   Type|    Type|    Type|       Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor|        SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
+        // #########|       |        |        |           |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder|  Lengths_K0_N0_N1_K1|               Order|                |                   |
+        // #########|       |        |        |           |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                     |                    |                |                   |
+        DeviceGemmDl< int8_t,  int8_t,  int8_t,    int32_t,     Row,     Row,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  4,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 4>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 4>,      S<1, 2, 0, 3>,        S<1, 1, 1, 4>,      S<2, 1, 4, 4>,      S<8, 1,  32, 1>,  S<0, 3, 1, 2>,  S<0, 3, 1, 2>,       S<1, 1, 4, 1>,      S<0, 3, 1, 2>,        S<1, 1, 4, 4>, S<0, 1, 2, 3, 4, 5>,               5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_i8_i8_i8_mk_kn_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_i8_i8_i8_mk_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp

+#include <stdlib.h>
+#include "config.hpp"
+#include "device_gemm_dl.hpp"
+#include "element_wise_operation.hpp"
+#include "device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// Compilation parameters for a[m, k] * b[n, k] = c[m, n]
+using device_gemm_dl_i8_i8_i8_mk_nk_mn_instances = std::tuple<
+    // clang-format off
+        // #########|  AData|   BData|   CData|    AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|  M11N11Thread|  M11N11Thread|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer|     ABlockTransfer|       ABlockTransfer|     BBlockTransfer|       BBlockTransfer| BBlockTransfer| BBlockTransfer|      BBlockTransfer|     BBlockTransfer|      BBlockTransfer|     CThreadTransfer| CThreadTransfer|    CThreadTransfer|
+        // #########|   Type|    Type|    Type|       Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|      DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|     SrcVectorTensor|    SrcVectorTensor|     DstVectorTensor|        SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
+        // #########|       |        |        |           |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       |              |              |        K0_M0_M1_K1|          K0_M0_M1_K1|   ArrangeOrder|          Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder|  Lengths_K0_M0_M1_K1|        K0_N0_N1_K1|          K0_N0_N1_K1|   ArrangeOrder|          Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder| Lengths_K0_N0_N1_K1|               Order|                |                   |
+        // #########|       |        |        |           |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |              |              |                   |                     |               |               |                    |                   |                     |                   |                     |               |               |                    |                   |                    |                    |                |                   |
+        DeviceGemmDl< int8_t,  int8_t,  int8_t,    int32_t,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    GemmDefault,   256,   128,   128,    16,  4,          4,          4,      1,       S<8, 2>,       S<8, 2>,      S<8, 1, 1, 4>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 4>,      S<1, 2, 0, 3>,        S<1, 1, 1, 4>,      S<8, 1, 1, 4>,      S<2, 1, 128, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<4, 1, 1, 4>,      S<1, 2, 0, 3>,       S<1, 1, 1, 4>, S<0, 1, 2, 3, 4, 5>,               5,                  4>
+    // clang-format on
+    >;
+
+void add_device_gemm_dl_i8_i8_i8_mk_nk_mn_instances(
+    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+{
+    add_device_operation_instances(instances, device_gemm_dl_i8_i8_i8_mk_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp

 #include <stdlib.h>
 #include "config.hpp"
-#include "device_gemm_xdl_c_shuffle.hpp"
+#include "device_gemm_xdl_cshuffle.hpp"
 #include "element_wise_operation.hpp"
 #include "device_operation_instance.hpp"
 
@@ -20,26 +20,28 @@ using S = ck::Sequence<Is...>;
 
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
 // Compilation parameters for a[m, k] * b[n, k] = c[m, n]
 using device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instances = std::tuple<
     // clang-format off
-        //#####################| AData| BData| CData| AccData| CShuffle| ALayout| BLayout| CLayout|           A|           B|           C| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|     CBlockTransferClusterLengths|  CBlockTransfer|      Num|
-        //#####################|  Type|  Type|  Type|    Type| DataType|        |        |        | Elementwise| Elementwise| Elementwise|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MXdlPerWave_MWaveMPerXdl| ScalarPerVector| Prefetch|
-        //#####################|      |      |      |        |         |        |        |        |   Operation|   Operation|   Operation|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| _NBlock_NXdlPerWave_NWaveNPerXdl|   _NWaveNPerXdl|         |
-        //#####################|      |      |      |        |         |        |        |        |            |            |            |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                                 |                |         |
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 4>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 4>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 32, 1, 1, 4>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 8>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 4>,               8,        2>,
-        DeviceGemmXdl_C_Shuffle<   F16,   F16,   F16,     F32,      F16,     Row,     Col,     Row, PassThrough, PassThrough, PassThrough,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,             S<1, 1, 16, 1, 1, 4>,               8,        2>
+        //#####################| ALayout| BLayout| CLayout| AData| BData| CData| AccData| CShuffle|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //#####################|        |        |        |  Type|  Type|  Type|    Type| DataType| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //#####################|        |        |        |      |      |      |        |         |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //#####################|        |        |        |      |      |      |        |         |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemm_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,   F16,     F32,      F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        2,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
     // clang-format on
     >;
 

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instance.cpp → library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp

@@ -22,7 +22,7 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
 // Compilation parameters for a[k, m] * b[k, n] = c[m, n]
-using device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instances =
+using device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instances =
     std::tuple<
         // clang-format off
         //#####################| ALayout| BLayout| CLayout|  AData|  BData|  CData| AccData| CShuffle|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
@@ -48,11 +48,11 @@ using device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instances =
         // clang-format on
         >;
 
-void add_device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instances(
+void add_device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instances(
     std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
 {
     add_device_operation_instances(instances,
-                                   device_gemm_xdl_c_shuffle_int8_int8_int8_km_kn_mn_instances{});
+                                   device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instances{});
 }
 
 } // namespace device_gemm_instance