add mising file

example/35_splitK_gemm/CMakeLists.txt

@@ -10,15 +10,15 @@ foreach(gpu IN LISTS GPU_TARGETS)
    add_example_executable(example_splitK_gemm_xdl_fp16 splitK_gemm_xdl_fp16.cpp)
    add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16)
 
+   add_example_executable(example_splitK_gemm_xdl_fp16_fp8 splitK_gemm_xdl_fp16_fp8.cpp)
+   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16_fp8)
+
    add_example_executable(example_splitK_gemm_xdl_bf16 splitK_gemm_xdl_bf16.cpp)
    add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_bf16)
 
    add_example_executable(example_splitK_gemm_xdl_int8 splitK_gemm_xdl_int8.cpp)
    add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int8)
 
-   add_example_executable(example_splitK_gemm_xdl_fp16_fp8 splitK_gemm_xdl_fp16_fp8.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16_fp8)
-
    if(USE_BITINT_EXTENSION_INT4)
        add_example_executable(example_splitK_gemm_xdl_int4 splitK_gemm_xdl_int4.cpp)
        add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int4)

example/35_splitK_gemm/splitK_gemm_xdl_fp16_fp8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl_splitk_c_shuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
+#include "ck/library/utility/literals.hpp"
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+using ADataType   = F16;
+using BDataType   = F16;
+using AccDataType = F32;
+using CDataType   = F16;
+
+using ALayout = Row;
+using BLayout = Row;
+using CLayout = Row;
+
+using AElementOp = PassThrough;
+using BElementOp = PassThrough;
+using CElementOp = PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmXdlSplitKCShuffle
+    // clang-format off
+    // GemmXdlSplitKCShuffle_MNKPadding_RRR_B128_Vec8x4x4_32x128x4x8 LoopScheduler: Interwave, PipelineVersion: v1
+//######|     AData|     BData|     CData|     AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer|  KPer|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer|           ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|     CBlockTransferClusterLengths|  CBlockTransfer|
+//######|      Type|      Type|      Type|        Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|           DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MXdlPerWave_MWaveMPerXdl| ScalarPerVector|
+//######|          |          |          |            |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|           PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| _NBlock_NXdlPerWave_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|          |          |          |            |        |        |        |            |            |            |               |      |      |      |      |    |     |     |     |     |                |               |               |               |                           |               |          |                |               |               |              |               |               |          |            |            |                                 |                |
+      //< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,    GemmDefault,   256,   256,   128,     4,   8,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,                           8,              8,      true,  S<1, 4, 64, 1>,  S<0, 1, 3, 2>,  S<0, 1, 3, 2>,             3,              8,              8,      true,           1,           1,                   S<1, 32, 1, 8>,               8>;
+        < ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,    GemmDefault,   128,    32,   128,     4,   8,   32,   32,    1,    2,  S<1, 4, 32, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,                           8,              8,      true,  S<1, 4, 32, 1>,  S<0, 1, 3, 2>,  S<0, 1, 3, 2>,             2,              4,              8,      true,           1,           1,                   S<1, 16, 1, 8>,               4,  F16, ck::PipelineVersion::v1, ck::LoopScheduler::Interwave>;
+// clang-format on
+
+#include "run_splitK_gemm_example.inc"
+
+int main(int argc, char* argv[]) { return !run_splitK_gemm_example(argc, argv); }

include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp

@@ -16,7 +16,7 @@ namespace element_wise {
 extern "C" __device__ float __ocml_native_recip_f32(float);
 #endif
 
-struct PassThroughPack2
+struct PassThroughPacked2
 {
     template <typename Y, typename X>
     __host__ __device__ void operator()(Y& y, const X& x) const;
@@ -67,6 +67,7 @@ struct PassThroughPack2
     constexpr const static bool is_pack2_invocable = true;
 };
 
+#if 1
 struct PassThrough
 {
     template <typename Y, typename X>
@@ -283,6 +284,7 @@ struct PassThrough
         y = ck::type_convert<bf8_t>(x);
     }
 };
+#endif
 
 struct UnaryConvert
 {

script/cmake-ck-dev.sh

@@ -11,7 +11,7 @@ cmake
 -D CMAKE_CXX_FLAGS="-std=c++17 -O3 -ftemplate-backtrace-limit=0  -fPIE  -Wno-gnu-line-marker -save-temps=$PWD"     \
 -D CMAKE_BUILD_TYPE=Release                                                                       \
 -D BUILD_DEV=OFF                                                                                   \
--D GPU_TARGETS="gfx90a"                                                             \
+-D GPU_TARGETS="gfx942"                                                             \
 -D CMAKE_VERBOSE_MAKEFILE:BOOL=ON                                                                 \
 -D USE_BITINT_EXTENSION_INT4=OFF                                                                  \
 ${MY_PROJECT_SOURCE}