adding tall and skinny gemm

example/53_gemv_splitk/gemv_splitk_fp16.cpp → composable_kernel/example/53_gemv_splitk/gemv_splitk_fp16.cpp

@@ -2,7 +2,7 @@
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 #include "common.hpp"
 
-#include "ck/tensor_operation/gpu/device/impl/device_gemv_splitk.hpp"
+#include "ck/tensor_operation/gpu/device/impl/device_tall_and_skinny_gemm_splitk.hpp"
 
 using ADataType   = ck::half_t;
 using BDataType   = ck::half_t;
@@ -25,7 +25,7 @@ static constexpr auto GemmMNPadding = ck::tensor_operation::device::GemmSpeciali
 #define B 64 // block-size:64
 
 // clang-format off
-using DeviceGemvInstance = ck::tensor_operation::device::deviceGemvDl/*
+using DeviceGemvInstance = ck::tensor_operation::device::deviceTsmmDl/*
 // ######|     AData|     BData|     CData|     AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer  |     ABlockTransfer|      ABlockTransfer  | BBlockTransfer|  BThreadTransfer|    BThreadTransfer|     CThreadTransfer| CThreadTransfer|    CThreadTransfer|
 // ######|      Type|      Type|      Type|        Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|SrcVectorTensorLengths|    SrcVectorTensor|DstVectorTensorLengths|      SrcAccess|     SrcVectorDim| SrcScalarPerVector|        SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
 // ######|          |          |          |            |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       | KBatch_K0_M0_M1_K1|   KBatch_K0_M0_M1_K1|   ArrangeOrder|          Order| KBatch_K0_M0_M1_K1   | ContiguousDimOrder| KBatch_K0_M0_M1_K1   |          Order|                 |                   |               Order|                |                   |

composable_kernel/example/54_tall_and_skinny_gemm_splitk/CMakeLists.txt

+list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
+set(target 0)
+foreach(gpu IN LISTS GPU_TARGETS)
+ if(gpu IN_LIST gpu_list AND target EQUAL 0)
+    add_custom_target(example_tall_and_skinny_gemm_splitk)
+    add_example_executable(example_tall_and_skinny_gemm_splitk_fp16 tall_and_skinny_gemm_splitk_fp16.cpp)
+    # set_source_files_properties(splitK_gemv_fp16.cpp PROPERTIES COMPILE_OPTIONS "--save-temps;-Wno-gnu-line-marker;-gline-tables-only")
+    add_dependencies(example_tall_and_skinny_gemm_splitk
+                example_tall_and_skinny_gemm_splitk_fp16)
+    set(target 1)
+ endif()
+ endforeach()
\ No newline at end of file

composable_kernel/example/54_tall_and_skinny_gemm_splitk/README.md

+# Instructions for ```example_gemv_splitk```
+
+## Run ```example_gemv_splitk```
+```bash
+#arg1: verification (0=no, 1=yes)
+#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
+#arg3: run kernel # of times (>1)
+#arg4: number of splitk batches
+./bin/example_tall_and_skinny_gemm_splitk_fp* 0 1 5 151
+
+```
+
+Result (MI250 @ 800Mhz, 181.05TFlops peak FP16)
+```
+a_m_k: dim 2, lengths {16, 1024}, strides {1024, 1}
+b_k_n: dim 2, lengths {1024, 16}, strides {16, 1}
+c_m_n: dim 2, lengths {16, 16}, strides {16, 1}
+Perf: 0.0684798 ms, 0.0076561 TFlops, 0.964489 GB/s, deviceGemvDl<64, 16, 128, 4, 2, 16, 2, 1>
+```

composable_kernel/example/54_tall_and_skinny_gemm_splitk/common.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <numeric>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/utility/data_type.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/fill.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/literals.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
+
+struct ProblemSize final // Default GEMV problem size
+{
+    ck::index_t M = 16;
+    ck::index_t N = 16;
+    ck::index_t K = 1024;
+    // ck::index_t M        = 2;
+    // ck::index_t N        = 256;
+    // ck::index_t K        = 256;
+    ck::index_t stride_A = K;
+    ck::index_t stride_B = N;//K;
+    ck::index_t stride_C = N;
+    ck::index_t k_batch  = 1;
+};
+
+struct ExecutionConfig final
+{
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+};
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+inline bool
+parse_cmd_args(int argc, char* argv[], ProblemSize& problem_size, ExecutionConfig& config)
+{
+    if(argc == 1)
+    {
+        // use default case
+    }
+    else if(argc == 5)
+    {
+        config.do_verification = std::stoi(argv[1]);
+        config.init_method     = std::stoi(argv[2]);
+        config.time_kernel     = std::stoi(argv[3]);
+
+        problem_size.k_batch = std::stoi(argv[4]);
+    }
+    else if(argc == 11)
+    {
+        config.do_verification = std::stoi(argv[1]);
+        config.init_method     = std::stoi(argv[2]);
+        config.time_kernel     = std::stoi(argv[3]);
+        problem_size.k_batch   = std::stoi(argv[4]);
+
+        problem_size.M = std::stoi(argv[5]);
+        problem_size.N = std::stoi(argv[6]);
+        problem_size.K = std::stoi(argv[7]);
+
+        problem_size.stride_A = std::stoi(argv[8]);
+        problem_size.stride_B = std::stoi(argv[9]);
+        problem_size.stride_C = std::stoi(argv[10]);
+    }
+    else
+    {
+        std::cerr << "arg1: verification (0=no, 1=yes)" << std::endl
+                  << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)"
+                  << std::endl
+                  << "arg3: time kernel (0=no, 1=yes)" << std::endl
+                  << "arg4 to 9: M (256x), N(128x), K(32x), StrideA, StrideB, StrideC" << std::endl;
+        return false;
+    }
+
+    return true;
+}

composable_kernel/example/54_tall_and_skinny_gemm_splitk/run_tall_and_skinny_gemm_splitk_example.inc

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+bool run_tall_and_skinny_gemm(const ProblemSize& problem_size, const ExecutionConfig& config)
+{
+#if defined(BUILD_INT4_EXAMPLE) && defined(CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4)
+    static_assert(sizeof(ck::int4_t) == sizeof(int8_t));
+#endif
+
+    using namespace ck::literals;
+
+    auto& [M, N, K, StrideA, StrideB, StrideC, k_batch] = problem_size; // //
+
+    auto f_host_tensor_descriptor =
+        [](std::size_t row, std::size_t col, std::size_t stride, auto layout) {
+            if constexpr(std::is_same_v<decltype(layout), ck::tensor_layout::gemm::RowMajor>)
+            {
+                return HostTensorDescriptor({row, col}, {stride, 1_uz});
+            }
+            else
+            {
+                return HostTensorDescriptor({row, col}, {1_uz, stride});
+            }
+        };
+
+    Tensor<ADataType> a_m_k(f_host_tensor_descriptor(M, K, StrideA, ALayout{}));
+    Tensor<BDataType> b_k_n(f_host_tensor_descriptor(K, N, StrideB, BLayout{}));
+
+    switch(config.init_method)
+    {
+    case 0: break;
+    case 1:
+        ck::utils::FillUniformDistributionIntegerValue<ADataType>{-5.f, 5.f}(a_m_k);
+        ck::utils::FillUniformDistributionIntegerValue<BDataType>{-5.f, 5.f}(b_k_n);
+        break;
+    default:
+        ck::utils::FillUniformDistribution<ADataType>{-1.f, 1.f}(a_m_k);
+        ck::utils::FillUniformDistribution<BDataType>{-1.f, 1.f}(b_k_n);
+    }
+
+    Tensor<CDataType> c_m_n_host_result(f_host_tensor_descriptor(M, N, StrideC, CLayout{}));
+    Tensor<CDataType> c_m_n_device_result(f_host_tensor_descriptor(M, N, StrideC, CLayout{}));
+
+    std::cout << "a_m_k: " << a_m_k.mDesc << std::endl;
+    std::cout << "b_k_n: " << b_k_n.mDesc << std::endl;
+    std::cout << "c_m_n: " << c_m_n_host_result.mDesc << std::endl;
+
+#ifdef BUILD_INT4_EXAMPLE
+    DeviceMem a_m_k_device_buf(sizeof(KernelADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(KernelBDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(KernelCDataType) *
+                               c_m_n_device_result.mDesc.GetElementSpaceSize());
+
+    const Tensor<KernelADataType> a_m_k_converted(a_m_k);
+    const Tensor<KernelBDataType> b_k_n_converted(b_k_n);
+
+    a_m_k_device_buf.ToDevice(a_m_k_converted.mData.data());
+    b_k_n_device_buf.ToDevice(b_k_n_converted.mData.data());
+#else
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
+
+    a_m_k_device_buf.ToDevice(a_m_k.mData.data());
+    b_k_n_device_buf.ToDevice(b_k_n.mData.data());
+#endif
+
+    auto a_element_op = AElementOp{};
+    auto b_element_op = BElementOp{};
+    auto c_element_op = CElementOp{};
+
+    // do GEMM
+    auto tsmm     = DeviceTSMMInstance{};
+    auto invoker  = tsmm.MakeInvoker();
+    auto argument = tsmm.MakeArgument(
+#ifdef BUILD_INT4_EXAMPLE
+        static_cast<KernelADataType*>(a_m_k_device_buf.GetDeviceBuffer()),
+        static_cast<KernelBDataType*>(b_k_n_device_buf.GetDeviceBuffer()),
+        static_cast<KernelCDataType*>(c_m_n_device_buf.GetDeviceBuffer()),
+#else
+        static_cast<ADataType*>(a_m_k_device_buf.GetDeviceBuffer()),
+        static_cast<BDataType*>(b_k_n_device_buf.GetDeviceBuffer()),
+        static_cast<CDataType*>(c_m_n_device_buf.GetDeviceBuffer()),
+#endif
+        M,
+        N,
+        K,
+        StrideA,
+        StrideB,
+        StrideC,
+        a_element_op,
+        b_element_op,
+        c_element_op,
+        k_batch); // //
+
+    // //
+    if(!tsmm.IsSupportedArgument(argument))
+    {
+        std::cerr << tsmm.GetTypeString() << " does not support this problem" << std::endl;
+
+        return true;
+    }
+
+    c_m_n_device_buf.SetZero();
+
+    
+
+    if(config.do_verification)
+    {
+        invoker.Run(argument, StreamConfig{nullptr, false}); // Run prior to verification
+        auto ref_tsmm    = ReferenceGemmInstance{};
+        auto ref_invoker = ref_tsmm.MakeInvoker();
+
+        auto ref_argument = ref_tsmm.MakeArgument(
+            a_m_k, b_k_n, c_m_n_host_result, a_element_op, b_element_op, c_element_op);
+
+        ref_invoker.Run(ref_argument);
+
+#ifdef BUILD_INT4_EXAMPLE
+        Tensor<CDataType> c_m_n_device_result_converted(c_m_n_host_result.mDesc);
+
+        c_m_n_device_buf.FromDevice(c_m_n_device_result_converted.mData.data());
+
+        c_m_n_device_result = c_m_n_device_result_converted.CopyAsType<CDataType>();
+
+#else
+        c_m_n_device_buf.FromDevice(c_m_n_device_result.mData.data());
+
+#endif
+    }
+
+    float ave_time = invoker.Run(
+        argument, StreamConfig{nullptr, config.time_kernel}); // Run to measure performance
+
+    std::size_t flop = 2_uz * M * N * K;
+    std::size_t num_btype =
+        sizeof(ADataType) * M * K + sizeof(BDataType) * K * N + sizeof(CDataType) * M * N;
+
+    float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
+
+    float gb_per_sec = num_btype / 1.E6 / ave_time;
+
+    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s, "
+              << tsmm.GetTypeString() << std::endl;
+
+#ifdef BUILD_INT4_EXAMPLE
+    return ck::utils::check_err(c_m_n_device_result_converted, c_m_n_host_result);
+#else
+    return ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
+#endif
+}
+
+bool run_tall_and_skinny_gemm_example(int argc, char* argv[])
+{
+    ProblemSize problem_size;
+    ExecutionConfig config;
+    if(argc == 1)
+    {
+        // use default case
+    }
+    else if(argc == 5)
+    {
+        config.do_verification = std::stoi(argv[1]);
+        config.init_method     = std::stoi(argv[2]);
+        config.time_kernel     = std::stoi(argv[3]);
+        problem_size.k_batch   = std::stoi(argv[4]);
+    }
+    else if(argc == 11)
+    {
+        config.do_verification = std::stoi(argv[1]);
+        config.init_method     = std::stoi(argv[2]);
+        config.time_kernel     = std::stoi(argv[3]);
+        problem_size.k_batch   = std::stoi(argv[4]);
+
+        problem_size.M = std::stoi(argv[5]);
+        problem_size.N = std::stoi(argv[6]);
+        problem_size.K = std::stoi(argv[7]);
+
+        problem_size.stride_A = std::stoi(argv[8]);
+        problem_size.stride_B = std::stoi(argv[9]);
+        problem_size.stride_C = std::stoi(argv[10]);
+    }
+    else
+    {
+        printf("arg1: verification (0=no, 1=yes)\n");
+        printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
+        printf("arg3: time kernel (0=no, 1=yes)\n");
+        printf("arg4: splitk\n");
+        printf("arg5 to 11: M (256x), N(128x), K(32x), StrideA, StrideB, StrideC\n");
+        exit(0);
+    }
+
+    return run_tall_and_skinny_gemm(problem_size, config);
+}

composable_kernel/example/54_tall_and_skinny_gemm_splitk/tall_and_skinny_gemm_splitk_fp16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
+#include "common.hpp"
+
+#include "ck/tensor_operation/gpu/device/impl/device_tall_and_skinny_gemm_splitk.hpp"
+
+using ADataType   = ck::half_t;
+using BDataType   = ck::half_t;
+using CDataType   = ck::half_t;
+using AccDataType = float;
+
+using ALayout = Row;
+using BLayout = Row;//Col;
+using CLayout = Row;
+
+using AElementOp = PassThrough;
+using BElementOp = PassThrough;
+using CElementOp = PassThrough;
+
+static constexpr auto GemmMNPadding = ck::tensor_operation::device::GemmSpecialization::MNPadding;
+
+#define K1 2
+#define K0 4
+#define N1 2
+#define B 64 // block-size:64
+#define M1 16
+
+// clang-format off
+using DeviceTSMMInstance = ck::tensor_operation::device::deviceTsmmDl/*
+// ######|     AData|     BData|     CData|     AccData| ALayout| BLayout| CLayout|           A|           B|           C|           GEMM| Block|  MPer|  NPer| K0Per| K1|      M1Per|      N1Per|   KPer|     ABlockTransfer|       ABlockTransfer| ABlockTransfer| ABlockTransfer|      ABlockTransfer  |     ABlockTransfer|      ABlockTransfer  | BBlockTransfer|  BThreadTransfer|    BThreadTransfer|     CThreadTransfer| CThreadTransfer|    CThreadTransfer|
+// ######|      Type|      Type|      Type|        Type|        |        |        | Elementwise| Elementwise| Elementwise| Spacialization|  Size| Block| Block| Block|   | ThreadM111| ThreadN111| Thread| ThreadSliceLengths| ThreadClusterLengths|  ThreadCluster|      SrcAccess|SrcVectorTensorLengths|    SrcVectorTensor|DstVectorTensorLengths|      SrcAccess|     SrcVectorDim| SrcScalarPerVector|        SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
+// ######|          |          |          |            |        |        |        |   Operation|   Operation|   Operation|               |      |      |      |      |   |           |           |       | KBatch_K0_M0_M1_K1|   KBatch_K0_M0_M1_K1|   ArrangeOrder|          Order| KBatch_K0_M0_M1_K1   | ContiguousDimOrder| KBatch_K0_M0_M1_K1   |          Order|                 |                   |               Order|                |                   |
+// ######|          |          |          |            |        |        |        |            |            |            |               |      |      |      |      |   |           |           |       |                   |                     |               |               |                      |                   |                      |               |                 |                   |                    |                |                   |
+       //< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,  GemmMNPadding,    64,     1,    64,    32,  2,          1,          1,      1,      S<1, 1, 1, 2>,      S<32, 1,  1, 1>,  S<1, 2, 0, 3>,  S<1, 2, 0, 3>,       S<1, 1, 1, 2>,      S<1, 2, 0, 3>,       S<1, 1, 1, 2>,    S<1, 2, 0, 3>,             3,               2,         S<0, 1, 2, 3, 4, 5>,               5,                  1>;*/
+         < ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,  GemmMNPadding,    B,     M1,   B*N1,   K0,  K1,         M1,         N1,     1,   S<1,1, 1, 1, K1>,   S<1,K0, 1,M1, 1>,  S<0,1,2,3,4>,   S<0,1,2,3,4>,      S<1,1, 1, 1, K1>,     S<0,1,2,3,4>,     S<1,1, 1, 1, K1>,      S<0,1,2,3,4>,                3,               N1,        S<0, 1, 2, 3, 4, 5>,             5,                  N1>;
+// clang-format on
+
+using ReferenceGemmInstance = ck::tensor_operation::host::
+    ReferenceGemm<ADataType, BDataType, CDataType, AccDataType, AElementOp, BElementOp, CElementOp>;
+
+#include "run_tall_and_skinny_gemm_splitk_example.inc"
+
+int main(int argc, char* argv[]) { return !run_tall_and_skinny_gemm_example(argc, argv); }