Merge remote-tracking branch 'origin/develop' into rosenrodt/gemm-layernorm

2b27d5fc · Chao Liu · f689a155 · fa9a0a5c · 2b27d5fc · 2b27d5fc
Commit 2b27d5fc authored Jul 01, 2022 by Chao Liu
20 changed files
--- a/library/src/tensor_operation_instance/gpu/elementwise/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/elementwise/CMakeLists.txt
+set(DEVICE_ELEMENTWISE_INSTANCE_SOURCE
+    device_normalize_instance.cpp
+)
+
+add_instance_library(device_elementwise_instance ${DEVICE_ELEMENTWISE_INSTANCE_SOURCE})
+
+target_compile_features(device_elementwise_instance PUBLIC)
+set_target_properties(device_elementwise_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+clang_tidy_check(device_elementwise_instance)
--- a/library/src/tensor_operation_instance/gpu/elementwise/device_normalize_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/elementwise/device_normalize_instance.cpp
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp"
+#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using inputType      = F16;
+using MeanType       = F32;
+using SquareMeanType = F32;
+using GammaDataType  = F16;
+using BetaDataType   = F16;
+using outputType     = F16;
+
+using Normalize = ck::tensor_operation::element_wise::Normalize;
+using device_normalize_from_mean_squaremean_f16_f32_f32_f16_f16_instances = std::tuple<
+    // clang-format off
+    //###################|in | mean| square_mean| gamma| beta| out| ComputeDataType|  functor| NDim| MPerThread| in, mean, square_mean, gamma, beta, out ScalarPerVector|
+    //###################|in | mean| square_mean| gamma| beta| out| ComputeDataType|  functor| NDim| MPerThread| in, mean, square_mean, gamma, beta, out ScalarPerVector|
+    //###################|in | mean| square_mean| gamma| beta| out| ComputeDataType|  functor| NDim| MPerThread| in, mean, square_mean, gamma, beta, out ScalarPerVector|
+    //###################|in | mean| square_mean| gamma| beta| out| ComputeDataType|  functor| NDim| MPerThread| in, mean, square_mean, gamma, beta, out ScalarPerVector|
+    Device5AryElementwise<F16,  F32,         F32,   F16,  F16, F16,            F32, Normalize,    2,          8,  8,    1,           1,     8,    8,   8                >,
+    Device5AryElementwise<F16,  F32,         F32,   F16,  F16, F16,            F32, Normalize,    2,          4,  4,    1,           1,     4,    4,   4                >,
+    Device5AryElementwise<F16,  F32,         F32,   F16,  F16, F16,            F32, Normalize,    2,          2,  2,    1,           1,     2,    2,   2                >,
+    Device5AryElementwise<F16,  F32,         F32,   F16,  F16, F16,            F32, Normalize,    2,          1,  1,    1,           1,     1,    1,   1                >
+    // clang-format on
+    >;
+
+void add_device_normalize_from_mean_squaremean_f16_f32_f32_f16_f16_instances(
+    std::vector<DeviceElementwisePtr<5, 1, 2, Normalize>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_normalize_from_mean_squaremean_f16_f32_f32_f16_f16_instances{});
+}
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
@@ -28,14 +28,6 @@ set(DEVICE_GEMM_INSTANCE_SOURCE
   device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp;
   device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp;
   device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp;
-   device_gemm_xdl_splitk_f32_f32_f32_mk_kn_mn_instance.cpp;
-   device_gemm_xdl_splitk_f32_f32_f32_mk_nk_mn_instance.cpp;
-   device_gemm_xdl_splitk_f32_f32_f32_km_kn_mn_instance.cpp;
-   device_gemm_xdl_splitk_f32_f32_f32_km_nk_mn_instance.cpp;
-   device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp;
-   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;

--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/CMakeLists.txt
-set(DEVICE_GEMM_REDUCE_INSTANCE_SOURCE
-    device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
-    device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
-    device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
-    device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
+set(DEVICE_GEMM_BIAS_ADD_REDUCE_INSTANCE_SOURCE
+    device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
+    device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
+    device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
+    device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
 )

-add_instance_library(device_gemm_bias_add_reduce_instance ${DEVICE_GEMM_REDUCE_INSTANCE_SOURCE})
-rocm_install(TARGETS device_gemm_bias_add_reduce_instance)
+add_library(device_gemm_bias_add_reduce_instance OBJECT ${DEVICE_GEMM_BIAS_ADD_REDUCE_INSTANCE_SOURCE})
+
+target_compile_features(device_gemm_bias_add_reduce_instance PUBLIC)
+set_target_properties(device_gemm_bias_add_reduce_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
 clang_tidy_check(device_gemm_bias_add_reduce_instance)
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/utility/reduction_operator.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;
+
+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;
+using ReduceSum   = ck::reduce::Add;
+using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
+
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;
+
+using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
+                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// c[m, n] = a[k, m] * b[k, n]
+using device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|          C1|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |            |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        // clang-format on
+        >;
+
+void add_device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances(
+    std::vector<DeviceGemmReducePtr<1, ReduceOps::Size()>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/utility/reduction_operator.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;
+
+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;
+using ReduceSum   = ck::reduce::Add;
+using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
+
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;
+
+using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
+                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// c[m, n] = a[k, m] * b[n, k]
+using device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|          C1|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |            |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        // clang-format on
+        >;
+
+void add_device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances(
+    std::vector<DeviceGemmReducePtr<1, ReduceOps::Size()>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/utility/reduction_operator.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_gemm_instance {
+
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;
+
+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;
+using ReduceSum   = ck::reduce::Add;
+using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
+
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;
+
+using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
+                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// c[m, n] = a[m, k] * b[n, k]
+using device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|          C1|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //##################################|        |        |        |     |      |      |      |      |         |         |          |                   |            |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        // clang-format on
+        >;
+
+void add_device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances(
+    std::vector<DeviceGemmReducePtr<1, ReduceOps::Size()>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances{});
+}
+
+} // namespace device_gemm_instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
@@ -16,9 +16,9 @@ namespace tensor_operation {
 namespace device {
 namespace device_gemm_instance {

-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;

 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
@@ -30,11 +30,11 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ReduceSum   = ck::reduce::Add;
 using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;

-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;

 using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
@@ -42,40 +42,35 @@ using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperat
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;

 // c[m, n] = a[m, k] * b[n, k]
-using device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances =
+using device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances =
    std::tuple<
        // clang-format off
        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|          C1|         Dxs|    DxsInEleOp|    DxsAccEleOp|           D|           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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise|      Reduce|              |               |  MemoryData| 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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |   Operation|   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |            |            |            |            |            |              |               |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
+        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>
        // clang-format on
        >;

-void add_device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances(
-    std::vector<DeviceGemmBiasAddReducePtr<PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           DInElementOps,
-                                           DOutElementOps>>& instances)
+void add_device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances(
+    std::vector<DeviceGemmReducePtr<1, ReduceOps::Size()>>& instances)
 {
    add_device_operation_instances(
        instances,
-        device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances{});
+        device_gemm_bias_add_mean_squaremean_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_nk_mn_instances{});
 }

 } // namespace device_gemm_instance

--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/utility/reduction_operator.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace device_gemm_instance {
-
-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
-
-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;
-using ReduceSum   = ck::reduce::Add;
-using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
-
-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
-
-using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
-                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// c[m, n] = a[k, m] * b[k, n]
-using device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances =
-    std::tuple<
-        // clang-format off
-        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|          C1|       Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise|    Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |   Operation|   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |            |            |            |            |          |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
-        // clang-format on
-        >;
-
-void add_device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances(
-    std::vector<DeviceGemmBiasAddReducePtr<PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           DInElementOps,
-                                           DOutElementOps>>& instances)
-{
-    add_device_operation_instances(
-        instances,
-        device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_kn_mn_instances{});
-}
-
-} // namespace device_gemm_instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/utility/reduction_operator.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace device_gemm_instance {
-
-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
-
-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;
-using ReduceSum   = ck::reduce::Add;
-using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
-
-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
-
-using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
-                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// c[m, n] = a[k, m] * b[n, k]
-using device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances =
-    std::tuple<
-        // clang-format off
-        //##################################| ALayout| BLayout| CLayout|AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|          C1|       Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //##################################|        |        |        | Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise|    Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |   Operation|   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //##################################|        |        |        |     |      |      |      |      |         |         |          |              |            |            |            |            |          |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
-        // clang-format on
-        >;
-
-void add_device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances(
-    std::vector<DeviceGemmBiasAddReducePtr<PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           DInElementOps,
-                                           DOutElementOps>>& instances)
-{
-    add_device_operation_instances(
-        instances,
-        device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_km_nk_mn_instances{});
-}
-
-} // namespace device_gemm_instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_bias_add_reduce/device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/utility/reduction_operator.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace device_gemm_instance {
-
-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
-
-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;
-using ReduceSum   = ck::reduce::Add;
-using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;
-
-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
-
-using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
-                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// c[m, n] = a[m, k] * b[n, k]
-using device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances =
-    std::tuple<
-        // clang-format off
-        //##################################| ALayout| BLayout| CLayout| AData| BData| CData|C0Data|C1Data|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|          C1|         Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //##################################|        |        |        |  Type|  Type|  Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise| Elementwise|      Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //##################################|        |        |        |      |      |      |      |      |         |         |          |              |   Operation|   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //##################################|        |        |        |      |      |      |      |      |         |         |          |              |            |            |            |            |            |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmBiasAddReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
-        // clang-format on
-        >;
-
-void add_device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances(
-    std::vector<DeviceGemmBiasAddReducePtr<PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           PassThrough,
-                                           DInElementOps,
-                                           DOutElementOps>>& instances)
-{
-    add_device_operation_instances(
-        instances,
-        device_gemm_bias_add_reduce_xdl_cshuffle_f16_f16_f16_f16_f16_f32_f32_mk_kn_mn_instances{});
-}
-
-} // namespace device_gemm_instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
--- a/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instance.cpp
@@ -16,9 +16,9 @@ namespace tensor_operation {
 namespace device {
 namespace device_gemm_instance {

-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;

 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
@@ -30,11 +30,11 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ReduceSum   = ck::reduce::Add;
 using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;

-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;

 using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
@@ -44,33 +44,31 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // c[m, n] = a[k, m] * b[k, n]
 using device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instances = std::tuple<
    // clang-format off
-        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|       Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise|    Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //###########################|        |        |        |     |      |      |         |         |          |              |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //###########################|        |        |        |     |      |      |         |         |          |              |            |            |            |          |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
    // clang-format on
    >;

 void add_device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instances(
-    std::vector<
-        DeviceGemmReducePtr<PassThrough, PassThrough, PassThrough, DInElementOps, DOutElementOps>>&
-        instances)
+    std::vector<DeviceGemmReducePtr<0, ReducePtrsGlobal::Size()>>& instances)
 {
    add_device_operation_instances(
        instances, device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_kn_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instance.cpp
@@ -16,9 +16,9 @@ namespace tensor_operation {
 namespace device {
 namespace device_gemm_instance {

-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;

 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
@@ -30,11 +30,11 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ReduceSum   = ck::reduce::Add;
 using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;

-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;

 using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
@@ -44,33 +44,31 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // c[m, n] = a[k, m] * b[n, k]
 using device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instances = std::tuple<
    // clang-format off
-        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|       Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise|    Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //###########################|        |        |        |     |      |      |         |         |          |              |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //###########################|        |        |        |     |      |      |         |         |          |              |            |            |            |          |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,     false,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Col,      Col,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
    // clang-format on
    >;

 void add_device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instances(
-    std::vector<
-        DeviceGemmReducePtr<PassThrough, PassThrough, PassThrough, DInElementOps, DOutElementOps>>&
-        instances)
+    std::vector<DeviceGemmReducePtr<0, ReducePtrsGlobal::Size()>>& instances)
 {
    add_device_operation_instances(
        instances, device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_km_nk_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instance.cpp
@@ -16,9 +16,9 @@ namespace tensor_operation {
 namespace device {
 namespace device_gemm_instance {

-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;

 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
@@ -30,11 +30,11 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ReduceSum   = ck::reduce::Add;
 using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;

-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;

 using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
@@ -44,33 +44,31 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // c[m, n] = a[m, k] * b[n, k]
 using device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instances = std::tuple<
    // clang-format off
-        //###########################| ALayout| BLayout| CLayout| AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|         Dxs|    DxsInEleOp|    DxsAccEleOp|           D|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //###########################|        |        |        |  Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise|      Reduce|              |               |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //###########################|        |        |        |      |      |      |         |         |          |              |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //###########################|        |        |        |      |      |      |         |         |          |              |            |            |            |            |              |               |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
+        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|          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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData|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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |            |            |            |          |                   |                    |            |              |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   256,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   128,    64,   128,    32,   8,   2,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,      true,           1,           1,               S<1, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,      true,           1,           1,               S<1, 16, 1, 4>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,     false,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,      Row,    Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,   GemmDefault,        1,   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<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,      true,           1,           1,               S<1, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>
    // clang-format on
    >;

 void add_device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instances(
-    std::vector<
-        DeviceGemmReducePtr<PassThrough, PassThrough, PassThrough, DInElementOps, DOutElementOps>>&
-        instances)
+    std::vector<DeviceGemmReducePtr<0, ReducePtrsGlobal::Size()>>& instances)
 {
    add_device_operation_instances(
        instances, device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_kn_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_reduce/device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instance.cpp
@@ -16,9 +16,9 @@ namespace tensor_operation {
 namespace device {
 namespace device_gemm_instance {

-using F16         = ck::half_t;
-using F32         = float;
-using DPtrsGlobal = ck::Tuple<F32*, F32*>;
+using F16              = ck::half_t;
+using F32              = float;
+using ReducePtrsGlobal = ck::Tuple<F32*, F32*>;

 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
@@ -30,11 +30,11 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ReduceSum   = ck::reduce::Add;
 using ReduceOps   = ck::Tuple<ReduceSum, ReduceSum>;

-using Div            = ck::tensor_operation::element_wise::UnaryDivide;
-using Identity       = ck::tensor_operation::element_wise::PassThrough;
-using Square         = ck::tensor_operation::element_wise::UnarySquare;
-using DInElementOps  = ck::Tuple<Identity, Square>;
-using DOutElementOps = ck::Tuple<Div, Div>;
+using Div                 = ck::tensor_operation::element_wise::UnaryDivide;
+using Identity            = ck::tensor_operation::element_wise::PassThrough;
+using Square              = ck::tensor_operation::element_wise::UnarySquare;
+using ReduceInElementOps  = ck::Tuple<Identity, Square>;
+using ReduceOutElementOps = ck::Tuple<Div, Div>;

 using ReduceMemOp = ck::InMemoryDataOperationEnumSequence<ck::InMemoryDataOperationEnum::AtomicAdd,
                                                          ck::InMemoryDataOperationEnum::AtomicAdd>;
@@ -44,30 +44,28 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // c[m, n] = a[m, k] * b[n, k]
 using device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instances = std::tuple<
    // clang-format off
-        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         DData|           A|           B|           C|         Dxs|    DxsInEleOp|    DxsAccEleOp|           D|           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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
-        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|    Type Tuple| Elementwise| Elementwise| Elementwise|      Reduce|              |               |  MemoryData| 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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
-        //###########################|        |        |        |     |      |      |         |         |          |              |   Operation|   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
-        //###########################|        |        |        |     |      |      |         |         |          |              |            |            |            |            |              |               |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
-        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   DPtrsGlobal, PassThrough, PassThrough, PassThrough,   ReduceOps, DInElementOps, DOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>
+        //###########################| ALayout| BLayout| CLayout|AData| BData| CData|  GemmAcc| CShuffle| ReduceAcc|         ReduceData|           A|           B|           C|    Reduce|      ReduceInEleOp|      ReduceAccEleOp|      Reduce|           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|              CReduce| CReduceThreadLds2VGprCopy| CReduceThreadVgpr2GlobalCopy|
+        //###########################|        |        |        | Type|  Type|  Type| DataType| DataType|  DataType|         Type Tuple| Elementwise| Elementwise| Elementwise| Operation|                   |                    |  MemoryData| 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_MPerBlock| ScalarPerVector| ThreadClusterLengths|     SrcDstScalarPerVector|        SrcDstScalarPerVector|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |   Operation|   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_NPerBlock|      _NPerBlock| _MPerBlock_NPerBlock|                _NPerBlock|                   _MPerBlock|
+        //###########################|        |        |        |     |      |      |         |         |          |                   |            |            |            |          |                   |                    |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |                     |                          |                             |
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 8>,               8,             S<64, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 32, 1, 4>,               8,             S<64, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,   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, 16, 1, 8>,               8,             S<32, 4>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>,
+        DeviceGemmReduce_Xdl_CShuffle<     Row,     Col,     Row,  F16,   F16,   F16,      F32,      F32,       F32,   ReducePtrsGlobal, PassThrough, PassThrough, PassThrough, ReduceOps, ReduceInElementOps, ReduceOutElementOps, ReduceMemOp,    GemmDefault,        1,    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, 16, 1, 4>,               8,             S<32, 2>,                         4,                            1>
    // clang-format on
    >;

 void add_device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instances(
-    std::vector<
-        DeviceGemmReducePtr<PassThrough, PassThrough, PassThrough, DInElementOps, DOutElementOps>>&
-        instances)
+    std::vector<DeviceGemmReducePtr<0, ReducePtrsGlobal::Size()>>& instances)
 {
    add_device_operation_instances(
        instances, device_gemm_reduce_xdl_cshuffle_f16_f16_f16_f32_f32_mk_nk_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm_splitk/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm_splitk/CMakeLists.txt
+set(DEVICE_GEMM_SPLITK_INSTANCE_SOURCE
+   device_gemm_xdl_splitk_f32_f32_f32_mk_kn_mn_instance.cpp;
+   device_gemm_xdl_splitk_f32_f32_f32_mk_nk_mn_instance.cpp;
+   device_gemm_xdl_splitk_f32_f32_f32_km_kn_mn_instance.cpp;
+   device_gemm_xdl_splitk_f32_f32_f32_km_nk_mn_instance.cpp;
+   device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp;
+   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;
+)
+
+add_library(device_gemm_splitk_instance OBJECT ${DEVICE_GEMM_SPLITK_INSTANCE_SOURCE})
+
+target_compile_features(device_gemm_splitk_instance PUBLIC)
+set_target_properties(device_gemm_splitk_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instance.cpp
@@ -46,7 +46,7 @@ using device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instances = std::tuple<
    >;

 void add_device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instances(
-    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<DeviceGemmSplitKPtr<PassThrough, PassThrough, PassThrough>>& instances)
 {
    add_device_operation_instances(instances,
                                   device_gemm_xdl_splitk_f16_f16_f16_km_kn_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instance.cpp
@@ -46,7 +46,7 @@ using device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instances = std::tuple<
    >;

 void add_device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instances(
-    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<DeviceGemmSplitKPtr<PassThrough, PassThrough, PassThrough>>& instances)
 {
    add_device_operation_instances(instances,
                                   device_gemm_xdl_splitk_f16_f16_f16_km_nk_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instance.cpp
@@ -46,7 +46,7 @@ using device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instances = std::tuple<
    >;

 void add_device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instances(
-    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<DeviceGemmSplitKPtr<PassThrough, PassThrough, PassThrough>>& instances)
 {
    add_device_operation_instances(instances,
                                   device_gemm_xdl_splitk_f16_f16_f16_mk_kn_mn_instances{});

--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instance.cpp
@@ -83,7 +83,7 @@ using device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instances = std::tuple<
 //     >;

 void add_device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instances(
-    std::vector<DeviceGemmPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<DeviceGemmSplitKPtr<PassThrough, PassThrough, PassThrough>>& instances)
 {
    add_device_operation_instances(instances,
                                   device_gemm_xdl_splitk_f16_f16_f16_mk_nk_mn_instances{});