fixed comments: unified blk2ctile; add test

include/ck/tensor_operation/gpu/device/device_gemm.hpp

@@ -76,9 +76,9 @@ template <typename AElementwiseOperation,
           typename CElementwiseOperation>
 struct DeviceGroupedGemm : public BaseOperator
 {
-    virtual std::unique_ptr<BaseArgument> MakeArgumentPointer(std::vector<const void*> p_a,
-                                                              std::vector<const void*> p_b,
-                                                              std::vector<void*> p_c,
+    virtual std::unique_ptr<BaseArgument> MakeArgumentPointer(std::vector<const void*>& p_a,
+                                                              std::vector<const void*>& p_b,
+                                                              std::vector<void*>& p_c,
                                                               std::vector<GemmShape>& gemm_shapes,
                                                               AElementwiseOperation a_element_op,
                                                               BElementwiseOperation b_element_op,

include/ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp

@@ -223,6 +223,35 @@ struct DeviceGroupedGemmXdl
         CThreadTransferDstScalarPerVector,
         NumPrefetch>;
 
+    struct GroupedGemmBlock2CTileMap
+    {
+        GroupedGemmBlock2CTileMap()
+        {
+            block_2_ctile_map_ = GridwiseGemm::MakeDefaultBlock2CTileMap(CGridDesc_M_N{}, 1, 1);
+            BlockStart_        = -1;
+        }
+
+        GroupedGemmBlock2CTileMap(const CGridDesc_M_N& c_grid_desc_m_n,
+                                  index_t M01,
+                                  index_t N01,
+                                  ck::index_t BlockStart)
+        {
+            block_2_ctile_map_ = GridwiseGemm::MakeDefaultBlock2CTileMap(c_grid_desc_m_n, M01, N01);
+            BlockStart_        = BlockStart;
+        }
+
+        template <typename TopIdx>
+        __host__ __device__ constexpr auto CalculateBottomIndex(const TopIdx& idx_top) const
+        {
+            return block_2_ctile_map_.CalculateBottomIndex(
+                make_multi_index(idx_top[I0] - BlockStart_));
+        }
+
+        private:
+        typename GridwiseGemm::DefaultBlock2CTileMap block_2_ctile_map_;
+        ck::index_t BlockStart_;
+    };
+
     struct GemmDescKernelArg
     {
         AGridDesc_K0_M_K1 a_grid_desc_k0_m_k1_;
@@ -232,21 +261,21 @@ struct DeviceGroupedGemmXdl
         typename GridwiseGemm::CGridDesc_M0_N0_M1_N1_M2_M3_M4_N2
             c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_;
 
-        typename GridwiseGemm::DefaultBlock2CTileMap block_2_ctile_map_;
+        GroupedGemmBlock2CTileMap grouped_gemm_block_2_ctile_map_;
 
         const ADataType* a_ptr;
         const BDataType* b_ptr;
         CDataType* c_ptr;
 
-        ck::index_t BlockStart, BlockEnd;
+        ck::index_t BlockStart_, BlockEnd_;
     };
 
     // Argument
     struct Argument : public BaseArgument
     {
-        Argument(std::vector<const void*> p_a,
-                 std::vector<const void*> p_b,
-                 std::vector<void*> p_c,
+        Argument(std::vector<const void*>& p_a,
+                 std::vector<const void*>& p_b,
+                 std::vector<void*>& p_c,
                  std::vector<GemmShape>& gemm_shapes,
                  index_t M01,
                  index_t N01,
@@ -301,15 +330,15 @@ struct DeviceGroupedGemmXdl
                     const auto c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_ =
                         GridwiseGemm::MakeCGridDescriptor_M0_N0_M1_N1_M2_M3_M4_N2(c_grid_desc_m_n_);
 
-                    const auto block_2_ctile_map_ =
-                        GridwiseGemm::MakeDefaultBlock2CTileMap(c_grid_desc_m_n_, M01, N01);
+                    const auto grouped_gemm_block_2_ctile_map_ =
+                        GroupedGemmBlock2CTileMap(c_grid_desc_m_n_, M01, N01, BlockStart);
 
                     gemm_desc_kernel_arg_.push_back(
                         GemmDescKernelArg{a_grid_desc_k0_m_k1_,
                                           b_grid_desc_k0_n_k1_,
                                           c_grid_desc_m_n_,
                                           c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_,
-                                          block_2_ctile_map_,
+                                          grouped_gemm_block_2_ctile_map_,
                                           static_cast<const ADataType*>(p_a[i]),
                                           static_cast<const BDataType*>(p_b[i]),
                                           static_cast<CDataType*>(p_c[i]),
@@ -470,9 +499,9 @@ struct DeviceGroupedGemmXdl
         return IsSupportedArgument(*dynamic_cast<const Argument*>(p_arg));
     }
 
-    static auto MakeArgument(std::vector<const void*> p_a,
-                             std::vector<const void*> p_b,
-                             std::vector<void*> p_c,
+    static auto MakeArgument(std::vector<const void*>& p_a,
+                             std::vector<const void*>& p_b,
+                             std::vector<void*>& p_c,
                              std::vector<GemmShape> gemm_shapes,
                              AElementwiseOperation a_element_op,
                              BElementwiseOperation b_element_op,
@@ -484,9 +513,9 @@ struct DeviceGroupedGemmXdl
     static auto MakeInvoker() { return Invoker{}; }
 
     // polymorphic
-    std::unique_ptr<BaseArgument> MakeArgumentPointer(std::vector<const void*> p_a,
-                                                      std::vector<const void*> p_b,
-                                                      std::vector<void*> p_c,
+    std::unique_ptr<BaseArgument> MakeArgumentPointer(std::vector<const void*>& p_a,
+                                                      std::vector<const void*>& p_b,
+                                                      std::vector<void*>& p_c,
                                                       std::vector<GemmShape>& gemm_shapes,
                                                       AElementwiseOperation a_element_op,
                                                       BElementwiseOperation b_element_op,

include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp

@@ -80,11 +80,10 @@ __global__ void
 
 #if 1
     static_for<0, MaxGroupCount, 1>{}([&](auto i) {
-        if(block_id >= gemm_desc_[i].BlockStart && block_id < gemm_desc_[i].BlockEnd &&
+        if(block_id >= gemm_desc_[i].BlockStart_ && block_id < gemm_desc_[i].BlockEnd_ &&
            i < group_count)
         {
             auto group_id = i;
-            const index_t block_id_grp = block_id - gemm_desc_[group_id].BlockStart;
 
             GridwiseGemm::template Run<HasMainK0BlockLoop>(
                 gemm_desc_[group_id].a_ptr,
@@ -97,8 +96,7 @@ __global__ void
                 a_element_op,
                 b_element_op,
                 c_element_op,
-                gemm_desc_[group_id].block_2_ctile_map_,
-                block_id_grp);
+                gemm_desc_[group_id].grouped_gemm_block_2_ctile_map_);
         }
     });
 #else
@@ -426,8 +424,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3
         const AElementwiseOperation& a_element_op,
         const BElementwiseOperation& b_element_op,
         const CElementwiseOperation& c_element_op,
-        const Block2CTileMap& block_2_ctile_map,
-        ck::index_t block_id = get_block_1d_id())
+        const Block2CTileMap& block_2_ctile_map)
     {
         const auto a_grid_buf = make_dynamic_buffer<AddressSpaceEnum_t::Global>(
             p_a_grid, a_grid_desc_k0_m_k1.GetElementSpaceSize());
@@ -440,7 +437,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3
 
         // divide block work by [M, N]
         const auto block_work_idx =
-            block_2_ctile_map.CalculateBottomIndex(make_multi_index(block_id));
+            block_2_ctile_map.CalculateBottomIndex(make_multi_index(get_block_1d_id()));
 
         // HACK: this force m/n_block_data_idx_on_grid into SGPR
         const index_t m_block_data_idx_on_grid =

test/CMakeLists.txt

@@ -35,6 +35,7 @@ add_subdirectory(space_filling_curve)
 add_subdirectory(conv_util)
 add_subdirectory(reference_conv_fwd)
 add_subdirectory(gemm)
+add_subdirectory(grouped_gemm)
 add_subdirectory(gemm_split_k)
 add_subdirectory(conv2d_fwd)
 add_subdirectory(convnd_fwd)

test/grouped_gemm/CMakeLists.txt

+add_test_executable(test_grouped_gemm_fp16 grouped_gemm_fp16.cpp)
+target_link_libraries(test_grouped_gemm_fp16 PRIVATE host_tensor)
+target_link_libraries(test_grouped_gemm_fp16 PRIVATE device_grouped_gemm_instance)

test/grouped_gemm/grouped_gemm_fp16.cpp

+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+#include <stdlib.h>
+#include <half.hpp>
+#include "config.hpp"
+#include "print.hpp"
+#include "device.hpp"
+#include "host_tensor.hpp"
+#include "host_tensor_generator.hpp"
+#include "host_gemm.hpp"
+#include "device_tensor.hpp"
+#include "device_grouped_gemm_xdl.hpp"
+#include "element_wise_operation.hpp"
+#include "reference_gemm.hpp"
+#include "gemm_specialization.hpp"
+#include "test_util.hpp"
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+using DeviceGroupedGemmPtr_ = ck::tensor_operation::device::DeviceGroupedGemmPtr<
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough>;
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace device_grouped_gemm_instance {
+void add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(
+    std::vector<DeviceGroupedGemmPtr_>&);
+}
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
+
+namespace {
+
+using ADataType   = ck::half_t;
+using BDataType   = ck::half_t;
+using CDataType   = ck::half_t;
+using AccDataType = float;
+
+using ALayout = ck::tensor_layout::gemm::RowMajor;
+using BLayout = ck::tensor_layout::gemm::ColumnMajor;
+using CLayout = ck::tensor_layout::gemm::RowMajor;
+
+template <typename T>
+static bool check_err(const Tensor<T>& ref, const Tensor<T>& result)
+{
+    float max_diff = 1e-2;
+
+    for(int i = 0; i < ref.mData.size(); ++i)
+    {
+        float diff = std::abs(double(ref.mData[i]) - double(result.mData[i]));
+        if(max_diff < diff)
+        {
+            std::cout << double(ref.mData[i]) << "," << double(result.mData[i]) << std::endl;
+            return false;
+        }
+    }
+
+    return true;
+}
+
+bool TestGroupedGemm(DeviceGroupedGemmPtr_& groupedGemmPtr)
+{
+    int group_count = 4;
+
+    // GEMM shape
+    std::vector<ck::tensor_operation::device::GemmShape> gemm_shapes;
+    std::vector<const void*> p_a, p_b;
+    std::vector<void*> p_c;
+
+    gemm_shapes.reserve(group_count);
+
+    for(int i = 0; i < group_count; i++)
+    {
+        int M = 256 + 256 * i;
+        int N = 128 + 128 * i;
+        int K = 64 + 64 * i;
+
+        int AStride = std::is_same<ck::tensor_layout::gemm::RowMajor, ALayout>::value ? K : M;
+        int BStride = std::is_same<ck::tensor_layout::gemm::RowMajor, BLayout>::value ? N : K;
+        int CStride = std::is_same<ck::tensor_layout::gemm::RowMajor, CLayout>::value ? N : M;
+
+        gemm_shapes.push_back({M, N, K, AStride, BStride, CStride});
+    }
+
+    auto f_host_tensor_descriptor =
+        [](std::size_t row, std::size_t col, std::size_t stride, auto layout) {
+            if(std::is_same<decltype(layout), ck::tensor_layout::gemm::RowMajor>::value)
+            {
+                return HostTensorDescriptor(std::vector<std::size_t>({row, col}),
+                                            std::vector<std::size_t>({stride, 1}));
+            }
+            else
+            {
+                return HostTensorDescriptor(std::vector<std::size_t>({row, col}),
+                                            std::vector<std::size_t>({1, stride}));
+            }
+        };
+
+    std::vector<Tensor<ADataType>> a_tensors;
+    ;
+    std::vector<Tensor<BDataType>> b_tensors;
+    std::vector<Tensor<CDataType>> c_host_tensors;
+    std::vector<Tensor<CDataType>> c_device_tensors;
+
+    a_tensors.reserve(group_count);
+    b_tensors.reserve(group_count);
+    c_host_tensors.reserve(group_count);
+    c_device_tensors.reserve(group_count);
+
+    using DeviceMemPtr = std::unique_ptr<DeviceMem>;
+
+    std::vector<DeviceMemPtr> a_tensors_device, b_tensors_device, c_tensors_device;
+
+    a_tensors_device.reserve(group_count);
+    b_tensors_device.reserve(group_count);
+    c_tensors_device.reserve(group_count);
+
+    std::size_t flop = 0, num_btype = 0;
+
+    for(int i = 0; i < gemm_shapes.size(); i++)
+    {
+        a_tensors.emplace_back(Tensor<ADataType>(f_host_tensor_descriptor(
+            gemm_shapes[i].M, gemm_shapes[i].K, gemm_shapes[i].StrideA, ALayout{})));
+        b_tensors.emplace_back(Tensor<BDataType>(f_host_tensor_descriptor(
+            gemm_shapes[i].K, gemm_shapes[i].N, gemm_shapes[i].StrideB, BLayout{})));
+        c_host_tensors.emplace_back(Tensor<CDataType>(f_host_tensor_descriptor(
+            gemm_shapes[i].M, gemm_shapes[i].N, gemm_shapes[i].StrideC, CLayout{})));
+        c_device_tensors.emplace_back(Tensor<CDataType>(f_host_tensor_descriptor(
+            gemm_shapes[i].M, gemm_shapes[i].N, gemm_shapes[i].StrideC, CLayout{})));
+
+        // std::cout << "gemm[" << i << "] a_m_k: " << a_tensors[i].mDesc
+        //<< " b_k_n: " << b_tensors[i].mDesc << " c_m_n: " << c_device_tensors[i].mDesc
+        //<< std::endl;
+
+        flop += std::size_t(2) * gemm_shapes[i].M * gemm_shapes[i].K * gemm_shapes[i].N;
+        num_btype += sizeof(ADataType) * a_tensors[i].mDesc.GetElementSize() +
+                     sizeof(BDataType) * b_tensors[i].mDesc.GetElementSize() +
+                     sizeof(CDataType) * c_device_tensors[i].mDesc.GetElementSize();
+
+        a_tensors[i].GenerateTensorValue(GeneratorTensor_3<ADataType>{0.0, 1.0});
+        b_tensors[i].GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
+    }
+
+    for(int i = 0; i < gemm_shapes.size(); i++)
+    {
+        a_tensors_device.emplace_back(
+            std::make_unique<DeviceMem>(sizeof(ADataType) * a_tensors[i].mDesc.GetElementSize()));
+        b_tensors_device.emplace_back(
+            std::make_unique<DeviceMem>(sizeof(BDataType) * b_tensors[i].mDesc.GetElementSize()));
+        c_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(CDataType) * c_device_tensors[i].mDesc.GetElementSize()));
+
+        a_tensors_device[i]->ToDevice(a_tensors[i].mData.data());
+        b_tensors_device[i]->ToDevice(b_tensors[i].mData.data());
+
+        p_a.push_back(a_tensors_device[i]->GetDeviceBuffer());
+        p_b.push_back(b_tensors_device[i]->GetDeviceBuffer());
+        p_c.push_back(c_tensors_device[i]->GetDeviceBuffer());
+    }
+
+    auto a_element_op = PassThrough{};
+    auto b_element_op = PassThrough{};
+    auto c_element_op = PassThrough{};
+
+    // do GEMM
+    auto invoker_ptr  = groupedGemmPtr->MakeInvokerPointer();
+    auto argument_ptr = groupedGemmPtr->MakeArgumentPointer(
+        p_a, p_b, p_c, gemm_shapes, a_element_op, b_element_op, c_element_op);
+
+    if(!groupedGemmPtr->IsSupportedArgument(argument_ptr.get()))
+    {
+        throw std::runtime_error(
+            "wrong! device_gemm with the specified compilation parameters does "
+            "not support this GEMM problem");
+    }
+
+    invoker_ptr->Run(argument_ptr.get());
+
+    for(int i = 0; i < gemm_shapes.size(); i++)
+    {
+        c_tensors_device[i]->FromDevice(c_device_tensors[i].mData.data());
+
+        using ReferenceGemmInstance = ck::tensor_operation::host::
+            ReferenceGemm<ADataType, BDataType, CDataType, PassThrough, PassThrough, PassThrough>;
+
+        auto ref_gemm    = ReferenceGemmInstance{};
+        auto ref_invoker = ref_gemm.MakeInvoker();
+
+        auto ref_argument = ref_gemm.MakeArgument(a_tensors[i],
+                                                  b_tensors[i],
+                                                  c_host_tensors[i],
+                                                  a_element_op,
+                                                  b_element_op,
+                                                  c_element_op);
+
+        ref_invoker.Run(ref_argument);
+
+        bool res = check_err(c_device_tensors[i], c_host_tensors[i]);
+
+        std::cout << "group_id: " << i << (res ? " SUCCESS" : " FAILURE") << std::endl;
+
+        if(!res)
+            return false;
+    }
+
+    return true;
+}
+
+} // anonymous namespace
+
+int main()
+{
+    std::vector<DeviceGroupedGemmPtr_> groupedGemmPtrs;
+    ck::tensor_operation::device::device_grouped_gemm_instance::
+        add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(groupedGemmPtrs);
+
+    bool res = true;
+
+    for(auto& gemmPtr : groupedGemmPtrs)
+    {
+        res &= TestGroupedGemm(gemmPtr);
+    }
+
+    std::cout << "TestGroupedGemm ..... " << (res ? "SUCCESS" : "FAILURE") << std::endl;
+}