gemm_xdl_fp16_fp8_streamk_v3.cpp added

example/01_gemm/CMakeLists.txt

@@ -28,11 +28,17 @@ add_example_executable(example_gemm_xdl_fp16_v3 gemm_xdl_fp16_v3.cpp)
 add_example_dependencies(example_gemm_xdl example_gemm_xdl_fp16_v3)
 add_example_executable(example_gemm_xdl_fp8_v3 gemm_xdl_fp8_v3.cpp)
 add_example_dependencies(example_gemm_xdl example_gemm_xdl_fp8_v3)
+
 add_example_executable(example_gemm_xdl_fp16_fp8_v3 gemm_xdl_fp16_fp8_v3.cpp)
+add_example_dependencies(example_gemm_xdl example_gemm_xdl_fp16_fp8_v3)
+
+add_example_executable(example_gemm_xdl_fp16_fp8_streamk_v3 gemm_xdl_fp16_fp8_streamk_v3.cpp)
+add_example_dependencies(example_gemm_xdl example_gemm_xdl_fp16_fp8_streamk_v3)
+
 add_example_executable(example_gemm_xdl_fp16_pk_i4_v3 gemm_xdl_fp16_pk_i4_v3.cpp)
 add_example_executable(example_gemm_xdl_fp16_pk_i4_v3_b_scale gemm_xdl_fp16_pk_i4_v3_b_scale.cpp)
 add_example_executable(example_gemm_xdl_bf16_pk_i4_v3 gemm_xdl_bf16_pk_i4_v3.cpp)
-add_example_dependencies(example_gemm_xdl example_gemm_xdl_fp16_fp8_v3)
+
 add_example_executable(example_gemm_xdl_bf16_v3 gemm_xdl_bf16_v3.cpp)
 add_example_dependencies(example_gemm_xdl example_gemm_xdl_bf16_v3)
 

example/01_gemm/gemm_xdl_fp16_fp8_streamk_v3.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "common.hpp"
+
+#include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl_cshuffle_streamk_v3.hpp"
+
+using ADataType        = ck::half_t;
+using BDataType        = ck::f8_t;
+using AccDataType      = float;
+using CShuffleDataType = ck::half_t;
+using CDataType        = ck::half_t;
+
+using ALayout = Row;
+using BLayout = Col;
+using CLayout = Row;
+
+using AElementOp = PassThrough;
+using BElementOp = PassThrough;
+using CElementOp = PassThrough;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// clang-format off
+using DeviceGemmV2_Streamk_Instance = 
+    ck::tensor_operation::device::DeviceGemm_Xdl_CShuffle_Streamk_V3<
+        ALayout,   BLayout,  CLayout,   
+        ADataType, BDataType, CDataType, AccDataType, CShuffleDataType, 
+        AElementOp, BElementOp, CElementOp, GemmDefault, 
+        64,
+        16, 16, 
+        256, 8, 16,
+        16,   16,
+        1,    1, 
+        S<32, 2, 1>,  S<1, 0, 2>,  S<1, 0, 2>,
+        2, 8, 8, 0,
+        S<16, 4, 1>,  S<1, 0, 2>,  S<1, 0, 2>,
+        2, 16, 16, 0,
+        1, 1, S<1, 16, 1, 4>, 4,
+        ck::BlockGemmPipelineScheduler::Interwave, ck::BlockGemmPipelineVersion::v1>;
+// clang-format on
+
+
+
+using ReferenceGemmInstanceGPU = ck::tensor_operation::device::ReferenceGemm<ALayout,
+                                                                             BLayout,
+                                                                             CLayout,
+                                                                             ADataType,
+                                                                             BDataType,
+                                                                             CDataType,
+                                                                             AccDataType,
+                                                                             AElementOp,
+                                                                             BElementOp,
+                                                                             CElementOp>;
+
+using ReferenceGemmInstance = ck::tensor_operation::host::ReferenceGemm<ADataType,
+                                                                        BDataType,
+                                                                        CDataType,
+                                                                        AccDataType,
+                                                                        PassThrough,
+                                                                        PassThrough,
+                                                                        PassThrough>;
+
+#include "run_gemm_example_streamk_v2.inc"
+
+int main(int argc, char* argv[]) { return !run_gemm_universal_streamk_example(argc, argv); }