[CK TILE] Add gemm compute pipeline v3

include/ck_tile/ops/gemm.hpp

@@ -25,6 +25,9 @@
 #include "ck_tile/ops/gemm/kernel/gemm_kernel.hpp"
 #include "ck_tile/ops/gemm/kernel/gemm_tile_partitioner.hpp"
 #include "ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_mem.hpp"
+#if 0 // Enable after local prefetch implementation
+#include "ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v3.hpp"
+#endif
 #include "ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_scheduler.hpp"
 #include "ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1.hpp"
 #include "ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1_default_policy.hpp"

include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v3.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include "ck_tile/core.hpp"
+#include "ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1_default_policy.hpp"
+#include "ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_scheduler.hpp"
+
+namespace ck_tile {
+
+//  A Tile Window: global memory
+//  B Tile Window: global memory
+//  C Distributed tensor: register
+template <typename Problem>
+struct BaseGemmPipelineAgBgCrCompV3
+{
+    using ADataType      = remove_cvref_t<typename Problem::ADataType>;
+    using BDataType      = remove_cvref_t<typename Problem::BDataType>;
+    using BlockGemmShape = remove_cvref_t<typename Problem::BlockGemmShape>;
+
+    static constexpr index_t BlockSize = Problem::kBlockSize;
+    static constexpr index_t MPerBlock = BlockGemmShape::kM;
+    static constexpr index_t NPerBlock = BlockGemmShape::kN;
+    static constexpr index_t KPerBlock = BlockGemmShape::kK;
+
+    // TODO: Is this 32K value gfx9 arch specific?
+
+    static constexpr index_t PrefetchStages  = 2;
+    static constexpr index_t PrefillStages   = 1;
+    static constexpr index_t GlobalBufferNum = 1;
+
+    CK_TILE_HOST static constexpr bool BlockHasHotloop(index_t num_loop)
+    {
+        return num_loop > PrefetchStages;
+    }
+
+    CK_TILE_HOST static constexpr TailNumber GetBlockLoopTailNum(index_t num_loop)
+    {
+        ignore = num_loop;
+        return TailNumber::Full;
+    }
+};
+
+// Compute optimized pipeline
+// GlobalPrefetchStages: 2
+// LocalPreFillStages: 1
+// LocalPreFetchStages: 1
+// LocalSharedMemoryBuffer: 1
+template <typename Problem, typename Policy = GemmPipelineAGmemBGmemCRegV1DefaultPolicy>
+struct GemmPipelineAgBgCrCompV3 : public BaseGemmPipelineAgBgCrCompV3<Problem>
+{
+    using Base = BaseGemmPipelineAgBgCrCompV3<Problem>;
+
+    using ADataType      = remove_cvref_t<typename Problem::ADataType>;
+    using BDataType      = remove_cvref_t<typename Problem::BDataType>;
+    using CDataType      = remove_cvref_t<typename Problem::CDataType>;
+    using BlockGemmShape = remove_cvref_t<typename Problem::BlockGemmShape>;
+
+    using ALayout = remove_cvref_t<typename Problem::ALayout>;
+    using BLayout = remove_cvref_t<typename Problem::BLayout>;
+    using CLayout = remove_cvref_t<typename Problem::CLayout>;
+
+    using BlockGemm = remove_cvref_t<decltype(Policy::template GetBlockGemm<Problem>())>;
+    using I0        = number<0>;
+    using I1        = number<1>;
+
+    static constexpr index_t BlockSize = Problem::kBlockSize;
+    static constexpr index_t MPerBlock = BlockGemmShape::kM;
+    static constexpr index_t NPerBlock = BlockGemmShape::kN;
+    static constexpr index_t KPerBlock = BlockGemmShape::kK;
+
+    static constexpr index_t VectorSizeA = Problem::VectorSizeA;
+    static constexpr index_t VectorSizeB = Problem::VectorSizeB;
+    static constexpr index_t VectorSizeC = Problem::VectorSizeC;
+
+    static constexpr bool kPadA = Problem::kPadA;
+    static constexpr bool kPadB = Problem::kPadB;
+    static constexpr bool kPadC = Problem::kPadC;
+
+    // Where is the right place for HasHotLoop and TailNum ???
+    static constexpr bool HasHotLoop = Problem::HasHotLoop;
+    static constexpr auto TailNum    = Problem::TailNum;
+    static constexpr auto Scheduler  = Problem::Scheduler;
+
+    using Base::PrefetchStages;
+
+    CK_TILE_HOST_DEVICE static constexpr index_t GetSmemSize()
+    {
+        return Policy::template GetSmemSize<Problem>();
+    }
+
+    template <GemmPipelineScheduler Scheduler>
+    struct PipelineImpl
+    {
+    };
+
+    template <>
+    struct PipelineImpl<GemmPipelineScheduler::Intrawave>
+    {
+        template <typename DstBlockTile, typename SrcTileWindow>
+        CK_TILE_DEVICE void GlobalPrefetch(DstBlockTile& dst_block_tile,
+                                           SrcTileWindow& dram_tile_window) const
+        {
+            load_tile(dst_block_tile, dram_tile_window);
+            move_tile_window(dram_tile_window, {0, KPerBlock});
+        }
+
+        template <typename DstTileWindow, typename SrcBlockTile, typename ElementFunction>
+        CK_TILE_DEVICE void LocalPrefill(DstTileWindow& lds_tile_window,
+                                         const SrcBlockTile& src_block_tile,
+                                         const ElementFunction& element_func) const
+        {
+            const auto block_tile_tmp = tile_elementwise_in(element_func, src_block_tile);
+            store_tile(lds_tile_window, block_tile_tmp);
+        }
+
+        CK_TILE_DEVICE static constexpr auto HotLoopScheduler()
+        {
+            constexpr index_t MPerXDL = BlockGemmShape::WarpTile::at(number<0>{});
+            constexpr index_t NPerXDL = BlockGemmShape::WarpTile::at(number<1>{});
+            constexpr index_t KPerXDL = BlockGemmShape::WarpTile::at(number<2>{});
+
+            constexpr index_t WaveSize = 64;
+            constexpr index_t WaveNumM = BlockGemmShape::BlockWarps::at(number<0>{});
+            constexpr index_t WaveNumN = BlockGemmShape::BlockWarps::at(number<1>{});
+
+            constexpr index_t A_LDS_Read_Width = KPerXDL;
+            constexpr index_t B_LDS_Read_Width = KPerXDL;
+
+            constexpr index_t A_Buffer_Load_Inst_Num =
+                MPerBlock * KPerBlock / (BlockSize * VectorSizeA);
+            constexpr index_t B_Buffer_Load_Inst_Num =
+                NPerBlock * KPerBlock / (BlockSize * VectorSizeB);
+
+            constexpr index_t A_LDS_Write_Inst_Num = MPerBlock * KPerBlock / (BlockSize * KPerXDL);
+            constexpr index_t B_LDS_Write_Inst_Num = NPerBlock * KPerBlock / (BlockSize * KPerXDL);
+
+            constexpr index_t A_LDS_Read_Inst_Num =
+                WaveNumN * MPerBlock * KPerBlock / (BlockSize * KPerXDL);
+            constexpr index_t B_LDS_Read_Inst_Num =
+                WaveNumM * MPerBlock * KPerBlock / (BlockSize * KPerXDL);
+
+            constexpr index_t C_MFMA_Inst_Num = MPerBlock * NPerBlock * KPerBlock /
+                                                (BlockSize / WaveSize) /
+                                                (MPerXDL * NPerXDL * KPerXDL);
+
+            // A/B split schedule
+            // compiler is likely to use ds_read2 when instruction width smaller than 16bytes
+            constexpr auto num_ds_read_inst_a = A_LDS_Read_Width * sizeof(ADataType) == 16
+                                                    ? A_LDS_Read_Inst_Num
+                                                    : A_LDS_Read_Inst_Num / 2;
+            constexpr auto num_ds_read_inst_b = B_LDS_Read_Width * sizeof(BDataType) == 16
+                                                    ? B_LDS_Read_Inst_Num
+                                                    : B_LDS_Read_Inst_Num / 2;
+
+            constexpr auto num_ds_write_inst_a = A_LDS_Write_Inst_Num;
+            constexpr auto num_ds_write_inst_b = B_LDS_Write_Inst_Num;
+
+            constexpr auto num_buffer_load_inst_a = A_Buffer_Load_Inst_Num;
+            constexpr auto num_buffer_load_inst_b = B_Buffer_Load_Inst_Num;
+
+            constexpr auto num_mfma_inst = C_MFMA_Inst_Num;
+
+            constexpr auto mfma_cycle = NPerXDL == 16 ? 16 : 32;
+            constexpr auto ds_read_a_issue_cycle =
+                A_LDS_Read_Width * sizeof(ADataType) == 16 ? 8 : 4;
+            constexpr auto ds_read_b_issue_cycle =
+                B_LDS_Read_Width * sizeof(BDataType) == 16 ? 8 : 4;
+            constexpr auto ds_read_a_mfma_rate =
+                (mfma_cycle - 4 + 2 * ds_read_a_issue_cycle - 1) / (2 * ds_read_a_issue_cycle);
+            constexpr auto ds_read_b_mfma_rate =
+                (mfma_cycle - 4 + 2 * ds_read_b_issue_cycle - 1) / (2 * ds_read_b_issue_cycle);
+
+            constexpr auto num_dsread_a_mfma =
+                (num_ds_read_inst_a + ds_read_a_mfma_rate - 1) / ds_read_a_mfma_rate;
+            constexpr auto num_dsread_b_mfma =
+                (num_ds_read_inst_b + ds_read_b_mfma_rate - 1) / ds_read_b_mfma_rate;
+
+            // stage 1
+            // Separate this part?
+            // constexpr auto num_mfma_per_ds_read = sizeof(ComputeDataType) / sizeof(ADataType) >
+            //                                               sizeof(ComputeDataType) /
+            //                                               sizeof(BDataType)
+            //                                           ? sizeof(ComputeDataType) /
+            //                                           sizeof(ADataType) : sizeof(ComputeDataType)
+            //                                           / sizeof(BDataType);
+            constexpr auto num_mfma_stage1 =
+                num_mfma_inst - (num_dsread_a_mfma + num_dsread_b_mfma);
+            constexpr auto num_mfma_per_issue =
+                num_mfma_stage1 / (num_buffer_load_inst_a + num_buffer_load_inst_b);
+            constexpr auto num_dswrite_per_issue_a = num_ds_write_inst_a / num_buffer_load_inst_a;
+            constexpr auto num_dswrite_per_issue_b = num_ds_write_inst_b / num_buffer_load_inst_b;
+
+            static_for<0, num_buffer_load_inst_a, 1>{}([&](auto i) {
+                ignore = i;
+                static_for<0, num_dswrite_per_issue_a, 1>{}([&](auto idswrite) {
+                    ignore = idswrite;
+                    __builtin_amdgcn_sched_group_barrier(0x200, 1, 0); // DS write
+                    __builtin_amdgcn_sched_group_barrier(0x008, 1, 0); // MFMA
+                });
+                __builtin_amdgcn_sched_group_barrier(0x020, 1, 0); // VMEM read
+                __builtin_amdgcn_sched_group_barrier(
+                    0x008, num_mfma_per_issue - num_dswrite_per_issue_a, 0); // MFMA
+            });
+            static_for<0, num_buffer_load_inst_b, 1>{}([&](auto i) {
+                ignore = i;
+                static_for<0, num_dswrite_per_issue_b, 1>{}([&](auto idswrite) {
+                    ignore = idswrite;
+                    __builtin_amdgcn_sched_group_barrier(0x200, 1, 0); // DS write
+                    __builtin_amdgcn_sched_group_barrier(0x008, 1, 0); // MFMA
+                });
+                __builtin_amdgcn_sched_group_barrier(0x020, 1, 0); // VMEM read
+                __builtin_amdgcn_sched_group_barrier(
+                    0x008, num_mfma_per_issue - num_dswrite_per_issue_b, 0); // MFMA
+            });
+
+            // stage 2
+            static_for<0, num_dsread_a_mfma, 1>{}([&](auto i) {
+                if constexpr((num_ds_read_inst_a - (i + 1) * ds_read_a_mfma_rate) >=
+                             ds_read_a_mfma_rate)
+                {
+                    __builtin_amdgcn_sched_group_barrier(0x100, ds_read_a_mfma_rate, 0); // DS read
+                }
+                else
+                {
+                    __builtin_amdgcn_sched_group_barrier(
+                        0x100,
+                        num_ds_read_inst_a - (num_dsread_a_mfma - 1) * ds_read_a_mfma_rate,
+                        0); // DS read
+                }
+                __builtin_amdgcn_sched_group_barrier(0x008, 1, 0); // MFMA
+            });
+
+            static_for<0, num_dsread_b_mfma, 1>{}([&](auto i) {
+                if constexpr((num_ds_read_inst_b - (i + 1) * ds_read_b_mfma_rate) >=
+                             ds_read_b_mfma_rate)
+                {
+                    __builtin_amdgcn_sched_group_barrier(0x100, ds_read_b_mfma_rate, 0); // DS read
+                }
+                else
+                {
+                    __builtin_amdgcn_sched_group_barrier(
+                        0x100,
+                        num_ds_read_inst_b - (num_dsread_b_mfma - 1) * ds_read_b_mfma_rate,
+                        0); // DS read
+                }
+                __builtin_amdgcn_sched_group_barrier(0x008, 1, 0); // MFMA
+            });
+        }
+
+        template <bool HasHotLoop,
+                  TailNumber TailNum,
+                  typename ADramBlockWindowTmp,
+                  typename BDramBlockWindowTmp,
+                  typename AElementFunction,
+                  typename BElementFunction>
+        CK_TILE_DEVICE auto operator()(const ADramBlockWindowTmp& a_dram_block_window_tmp,
+                                       const AElementFunction& a_element_func,
+                                       const BDramBlockWindowTmp& b_dram_block_window_tmp,
+                                       const BElementFunction& b_element_func,
+                                       index_t num_loop,
+                                       void* p_smem) const
+        {
+            static_assert(
+                std::is_same_v<ADataType, remove_cvref_t<typename ADramBlockWindowTmp::DataType>> &&
+                    std::is_same_v<BDataType,
+                                   remove_cvref_t<typename BDramBlockWindowTmp::DataType>>,
+                "A/B Dram block window should have the same data type as appropriate "
+                "([A|B]DataType) defined in Problem definition!");
+
+            static_assert(MPerBlock == ADramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
+                              NPerBlock ==
+                                  BDramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
+                              KPerBlock == ADramBlockWindowTmp{}.get_window_lengths()[number<1>{}],
+                          "A/B block window appropriate sizes must be equal to MPerBlock/NPerblock"
+                          " or KPerBlock!");
+
+            // ------------------------------------------------------------------------------------
+            // Definitions of all needed tiles
+
+            // A tile in LDS
+            ADataType* p_a_lds              = static_cast<ADataType*>(p_smem);
+            constexpr auto a_lds_block_desc = Policy::template MakeALdsBlockDescriptor<Problem>();
+            auto a_lds_block = make_tensor_view<address_space_enum::lds>(p_a_lds, a_lds_block_desc);
+
+            // TODO: LDS alignment should come from Policy!
+            constexpr index_t a_lds_block_space_size_aligned =
+                integer_divide_ceil(sizeof(ADataType) * a_lds_block_desc.get_element_space_size(),
+                                    16) *
+                16;
+
+            // B tile in LDS
+            BDataType* p_b_lds = static_cast<BDataType*>(
+                static_cast<void*>(static_cast<char*>(p_smem) + a_lds_block_space_size_aligned));
+            constexpr auto b_lds_block_desc = Policy::template MakeBLdsBlockDescriptor<Problem>();
+            auto b_lds_block = make_tensor_view<address_space_enum::lds>(p_b_lds, b_lds_block_desc);
+
+            // A DRAM tile window for load
+            auto a_copy_dram_window =
+                make_tile_window(a_dram_block_window_tmp.get_bottom_tensor_view(),
+                                 make_tuple(number<MPerBlock>{}, number<KPerBlock>{}),
+                                 a_dram_block_window_tmp.get_window_origin(),
+                                 Policy::template MakeADramTileDistribution<Problem>());
+
+            // A LDS tile window for store
+            auto a_copy_lds_window =
+                make_tile_window(a_lds_block,
+                                 make_tuple(number<MPerBlock>{}, number<KPerBlock>{}),
+                                 {0, 0},
+                                 a_copy_dram_window.get_tile_distribution());
+            // B DRAM tile window for load
+            auto b_copy_dram_window =
+                make_tile_window(b_dram_block_window_tmp.get_bottom_tensor_view(),
+                                 make_tuple(number<NPerBlock>{}, number<KPerBlock>{}),
+                                 b_dram_block_window_tmp.get_window_origin(),
+                                 Policy::template MakeBDramTileDistribution<Problem>());
+
+            // B LDS tile window for store
+            auto b_copy_lds_window =
+                make_tile_window(b_lds_block,
+                                 make_tuple(number<NPerBlock>{}, number<KPerBlock>{}),
+                                 {0, 0},
+                                 b_copy_dram_window.get_tile_distribution());
+
+            // A LDS tile for block GEMM
+            auto a_lds_gemm_window = make_tile_window(
+                a_lds_block, make_tuple(number<MPerBlock>{}, number<KPerBlock>{}), {0, 0});
+            // B LDS tile for block GEMM
+            auto b_lds_gemm_window = make_tile_window(
+                b_lds_block, make_tuple(number<NPerBlock>{}, number<KPerBlock>{}), {0, 0});
+
+            // Block GEMM
+            constexpr auto block_gemm = BlockGemm();
+            auto c_block_tile         = block_gemm.MakeCBlockTile();
+
+            using ABlockTileDistr = decltype(a_copy_dram_window.get_tile_distribution());
+            using BBlockTileDistr = decltype(b_copy_dram_window.get_tile_distribution());
+
+            using ABlockTile =
+                decltype(make_static_distributed_tensor<ADataType>(ABlockTileDistr{}));
+            using BBlockTile =
+                decltype(make_static_distributed_tensor<BDataType>(BBlockTileDistr{}));
+
+            ABlockTile a_block_tile;
+            BBlockTile b_block_tile;
+
+            // -----------------------------------------------------------------------------------------
+            // Gemm pipeline start
+
+            // prefetch
+            // global read 0
+            GlobalPrefetch(a_block_tile, a_copy_dram_window);
+            GlobalPrefetch(b_block_tile, b_copy_dram_window);
+
+            // initialize C
+            tile_elementwise_inout([](auto& c) { c = 0; }, c_block_tile);
+
+            // LDS write 0
+            LocalPrefill(a_copy_lds_window, a_block_tile, a_element_func);
+            LocalPrefill(b_copy_lds_window, b_block_tile, b_element_func);
+
+            GlobalPrefetch(a_block_tile, a_copy_dram_window);
+            GlobalPrefetch(b_block_tile, b_copy_dram_window);
+
+            block_sync_lds();
+            block_gemm.LocalPrefetch(a_lds_gemm_window, b_lds_gemm_window);
+
+            __builtin_amdgcn_sched_barrier(0);
+
+            // main body
+            if constexpr(HasHotLoop)
+            {
+                index_t i = 0;
+                do
+                {
+                    block_sync_lds();
+
+                    LocalPrefill(a_copy_lds_window, a_block_tile, a_element_func);
+                    LocalPrefill(b_copy_lds_window, b_block_tile, b_element_func);
+
+                    GlobalPrefetch(a_block_tile, a_copy_dram_window);
+                    GlobalPrefetch(b_block_tile, b_copy_dram_window);
+
+                    block_gemm.template operator()<decltype(c_block_tile),
+                                                   decltype(a_lds_gemm_window),
+                                                   decltype(b_lds_gemm_window),
+                                                   false>(
+                        c_block_tile, a_lds_gemm_window, b_lds_gemm_window);
+
+                    block_sync_lds();
+                    block_gemm.LocalPrefetch(a_lds_gemm_window, b_lds_gemm_window);
+                    HotLoopScheduler();
+                    __builtin_amdgcn_sched_barrier(0);
+
+                    i += 1;
+                } while(i < (num_loop - 1));
+            }
+            // tail
+            if constexpr(TailNum == TailNumber::Full)
+            {
+                block_gemm.template operator()<decltype(c_block_tile),
+                                               decltype(a_lds_gemm_window),
+                                               decltype(b_lds_gemm_window),
+                                               false>(
+                    c_block_tile, a_lds_gemm_window, b_lds_gemm_window);
+            }
+            // Let's leak last MFMA block to epilogue region, cover the potential lds-shuffle
+            // latency
+            // __builtin_amdgcn_sched_barrier(0);
+            return c_block_tile;
+        }
+    };
+
+    template <typename ADramBlockWindowTmp,
+              typename BDramBlockWindowTmp,
+              typename AElementFunction,
+              typename BElementFunction>
+    CK_TILE_DEVICE auto operator()(const ADramBlockWindowTmp& a_dram_block_window_tmp,
+                                   const AElementFunction& a_element_func,
+                                   const BDramBlockWindowTmp& b_dram_block_window_tmp,
+                                   const BElementFunction& b_element_func,
+                                   index_t num_loop,
+                                   void* p_smem) const
+    {
+        return PipelineImpl<Scheduler>{}.template operator()<HasHotLoop, TailNum>(
+            a_dram_block_window_tmp,
+            a_element_func,
+            b_dram_block_window_tmp,
+            b_element_func,
+            num_loop,
+            p_smem);
+    }
+
+    template <typename ADramBlockWindowTmp, typename BDramBlockWindowTmp>
+    CK_TILE_DEVICE auto operator()(const ADramBlockWindowTmp& a_dram_block_window_tmp,
+                                   const BDramBlockWindowTmp& b_dram_block_window_tmp,
+                                   index_t num_loop,
+                                   void* p_smem) const
+    {
+        return PipelineImpl<Scheduler>{}.template operator()<HasHotLoop, TailNum>(
+            a_dram_block_window_tmp,
+            [](const ADataType& a) { return a; },
+            b_dram_block_window_tmp,
+            [](const BDataType& b) { return b; },
+            num_loop,
+            p_smem);
+    }
+};
+
+} // namespace ck_tile