mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-04-19 22:39:03 +00:00
4d041837ad
* Adding RapidJson Library * Adding Json Dumps in all CK_Tile Examples Not verified yet * Adding json to cktile Batched Transpose * adding json dumps to layernorm2d_fwd * Adding json dump to flatmm_basic * Adding RapidJson Library * Adding Json Dumps in all CK_Tile Examples Not verified yet * Adding json to cktile Batched Transpose * adding json dumps to layernorm2d_fwd * Adding json dump to flatmm_basic * Adding json in 03_gemm * Add json dump to 16_batched_gemm * Add json dump to gemm_multi_d_fp16 * Add json dump to grouped_gemm * fix fmha_bwd/fwd * Fix clang-format errors exclude include/rapidjson in jenkins as its a third-party library * Saparating function and defination. * Update Documentation of 03_gemm * Refactoring as per code review * Disable fp8 instances on unsupported targets (#2592) * Restrict building of gemm_universal_preshuffle_f8 instances to specific targets in CMakeLists.txt * Add condition to skip gemm_xdl_universal_preshuffle_f8 instances for unsupported targets in CMakeLists.txt * Add conditions to skip unsupported targets for gemm_universal_preshuffle_f8 and gemm_xdl_universal_preshuffle_f8 instances in CMakeLists.txt * Refine conditions to exclude gemm_universal_preshuffle_f8 instances for unsupported targets in CMakeLists.txt --------- Co-authored-by: AviralGoelAMD <aviralgoel@amd.com> * fix clang format * remove duplicate lines of code from library/src/tensor_operation_instance/gpu/CMakeLists.txt * Fixing Readme and unifying jsondumps * adding moe_smoothquant * adding fused_moe * Fixing Readme for batched_gemm * Fixing Readme for grouped_gemm * adding flatmm * adding gemm_multi_d_fp16 * adding elementwise * adding File name when json is dumped * Fixing Reduce after merge * adding batched_transpose * Adding Warptile in Gemm * Fixing Clang Format --------- Co-authored-by: Aviral Goel <aviral.goel@amd.com> Co-authored-by: AviralGoelAMD <aviralgoel@amd.com> Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
204 lines
6.6 KiB
C++
204 lines
6.6 KiB
C++
|
|
// SPDX-License-Identifier: MIT
|
|
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
|
|
|
|
#pragma once
|
|
|
|
#include <string>
|
|
|
|
#include "ck_tile/core.hpp"
|
|
#include "ck_tile/host/kernel_launch.hpp"
|
|
#include "ck_tile/ops/epilogue.hpp"
|
|
#include "ck_tile/ops/flatmm.hpp"
|
|
#include "ck_tile/ops/gemm.hpp"
|
|
|
|
// GEMM config with 32x132 warp tile
|
|
template <typename DataType>
|
|
struct FlatmmConfig32
|
|
{
|
|
static constexpr ck_tile::index_t M_Tile = 128;
|
|
static constexpr ck_tile::index_t N_Tile = 128;
|
|
static constexpr ck_tile::index_t K_Tile = 128 / sizeof(DataType);
|
|
|
|
static constexpr ck_tile::index_t M_Warp = 1;
|
|
static constexpr ck_tile::index_t N_Warp = 4;
|
|
static constexpr ck_tile::index_t K_Warp = 1;
|
|
|
|
static constexpr ck_tile::index_t M_Warp_Tile = 32;
|
|
static constexpr ck_tile::index_t N_Warp_Tile = 32;
|
|
static constexpr ck_tile::index_t K_Warp_Tile = sizeof(DataType) == 2 ? 16 : 32;
|
|
|
|
static constexpr bool kPadM = false;
|
|
static constexpr bool kPadN = false;
|
|
static constexpr bool kPadK = false;
|
|
|
|
static constexpr bool TransposeC = false;
|
|
static constexpr bool UseStructuredSparsity = false;
|
|
|
|
static constexpr int kBlockPerCu = 2;
|
|
static constexpr int TileParitionerGroupNum = 8;
|
|
static constexpr int TileParitionerM01 = 4;
|
|
static constexpr auto Scheduler = ck_tile::GemmPipelineScheduler::Default;
|
|
static constexpr ck_tile::index_t NumWaveGroups = 1;
|
|
static constexpr bool DoubleSmemBuffer = false;
|
|
};
|
|
|
|
template <typename DataType>
|
|
struct FlatmmConfig32_950 : public FlatmmConfig32<DataType>
|
|
{
|
|
static constexpr ck_tile::index_t K_Warp_Tile = sizeof(DataType) == 2 ? 16 : 64;
|
|
};
|
|
|
|
// GEMM config with 16x16 warp tile
|
|
template <typename DataType>
|
|
struct FlatmmConfig16
|
|
{
|
|
static constexpr ck_tile::index_t M_Tile = 128;
|
|
static constexpr ck_tile::index_t N_Tile = 128;
|
|
static constexpr ck_tile::index_t K_Tile = 128 / sizeof(DataType);
|
|
|
|
static constexpr ck_tile::index_t M_Warp = 1;
|
|
static constexpr ck_tile::index_t N_Warp = 4;
|
|
static constexpr ck_tile::index_t K_Warp = 1;
|
|
|
|
static constexpr ck_tile::index_t M_Warp_Tile = 16;
|
|
static constexpr ck_tile::index_t N_Warp_Tile = 16;
|
|
static constexpr ck_tile::index_t K_Warp_Tile = sizeof(DataType) == 2 ? 32 : 64;
|
|
|
|
static constexpr bool kPadM = false;
|
|
static constexpr bool kPadN = false;
|
|
static constexpr bool kPadK = false;
|
|
|
|
static constexpr bool TransposeC = false;
|
|
static constexpr bool UseStructuredSparsity = false;
|
|
|
|
static constexpr int kBlockPerCu = 2;
|
|
static constexpr int TileParitionerGroupNum = 8;
|
|
static constexpr int TileParitionerM01 = 4;
|
|
static constexpr auto Scheduler = ck_tile::GemmPipelineScheduler::Default;
|
|
static constexpr ck_tile::index_t NumWaveGroups = 1;
|
|
static constexpr bool DoubleSmemBuffer = false;
|
|
};
|
|
|
|
template <typename DataType>
|
|
struct FlatmmConfig16_950 : public FlatmmConfig16<DataType>
|
|
{
|
|
static constexpr ck_tile::index_t K_Warp_Tile = sizeof(DataType) == 2 ? 32 : 128;
|
|
};
|
|
|
|
template <typename ADataType>
|
|
struct GemmBasicTypeConfig;
|
|
|
|
template <>
|
|
struct GemmBasicTypeConfig<ck_tile::half_t>
|
|
{
|
|
using ADataType = ck_tile::half_t;
|
|
using BDataType = ck_tile::half_t;
|
|
using AccDataType = float;
|
|
using CDataType = ck_tile::half_t;
|
|
// ToDo: Add more bias config to support different categories of GEMM.
|
|
};
|
|
|
|
template <>
|
|
struct GemmBasicTypeConfig<ck_tile::bf16_t>
|
|
{
|
|
using ADataType = ck_tile::bf16_t;
|
|
using BDataType = ck_tile::bf16_t;
|
|
using AccDataType = float;
|
|
using CDataType = ck_tile::bf16_t;
|
|
};
|
|
template <>
|
|
struct GemmBasicTypeConfig<ck_tile::fp8_t>
|
|
{
|
|
using ADataType = ck_tile::fp8_t;
|
|
using BDataType = ck_tile::fp8_t;
|
|
using AccDataType = float;
|
|
using CDataType = ck_tile::half_t;
|
|
// ToDo: Add more bias config to support different categories of GEMM.
|
|
};
|
|
|
|
template <>
|
|
struct GemmBasicTypeConfig<ck_tile::bf8_t>
|
|
{
|
|
using ADataType = ck_tile::bf8_t;
|
|
using BDataType = ck_tile::bf8_t;
|
|
using AccDataType = float;
|
|
using CDataType = ck_tile::half_t;
|
|
};
|
|
|
|
template <typename T>
|
|
struct DataTypeTraits;
|
|
|
|
template <>
|
|
struct DataTypeTraits<ck_tile::fp8_t>
|
|
{
|
|
static constexpr const char* name = "fp8";
|
|
};
|
|
|
|
template <>
|
|
struct DataTypeTraits<ck_tile::bf8_t>
|
|
{
|
|
static constexpr const char* name = "bf8";
|
|
};
|
|
template <>
|
|
struct DataTypeTraits<float>
|
|
{
|
|
static constexpr const char* name = "fp32";
|
|
};
|
|
|
|
template <>
|
|
struct DataTypeTraits<double>
|
|
{
|
|
static constexpr const char* name = "fp64";
|
|
};
|
|
|
|
template <>
|
|
struct DataTypeTraits<ck_tile::half_t>
|
|
{
|
|
static constexpr const char* name = "fp16";
|
|
};
|
|
|
|
template <typename T>
|
|
struct is_8bit_type
|
|
: std::bool_constant<std::is_same_v<T, ck_tile::fp8_t> || std::is_same_v<T, ck_tile::bf8_t>>
|
|
{
|
|
};
|
|
|
|
auto create_args(int argc, char* argv[])
|
|
{
|
|
ck_tile::ArgParser arg_parser;
|
|
arg_parser.insert("m", "256", "m dimension")
|
|
.insert("n", "256", "n dimension")
|
|
.insert("k", "128", "k dimension")
|
|
.insert("a_layout", "R", "A tensor data layout - Row by default")
|
|
.insert("b_layout", "C", "B tensor data layout - Row by default")
|
|
.insert("c_layout", "R", "C tensor data layout - Row by default")
|
|
.insert("stride_a", "0", "Tensor A stride")
|
|
.insert("stride_b", "0", "Tensor B stride")
|
|
.insert("stride_c", "0", "Tensor C stride")
|
|
.insert("v", "1", "0. No validation, 1. Validation on CPU, 2. Validation on GPU")
|
|
.insert("prec", "fp16", "data type. fp16/bf16/fp8/bf8")
|
|
.insert("warmup", "50", "number of iterations before benchmark the kernel")
|
|
.insert("repeat", "100", "number of iterations to benchmark the kernel")
|
|
.insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer")
|
|
.insert("split_k", "1", "splitK value")
|
|
.insert("init", "0", "0:random, 1:linear, 2:constant(1)")
|
|
.insert(
|
|
"warp_tile", "0", "0: 16x16, 1: 32x32, 2: 16x16x128 (950 only), 3: 32x32x64 (950 only)")
|
|
.insert("json", "0", "0: No Json, 1: Dump Results in Json format")
|
|
.insert("jsonfile", "flatmm_basic.json", "json file name to dump results");
|
|
bool result = arg_parser.parse(argc, argv);
|
|
return std::make_tuple(result, arg_parser);
|
|
}
|
|
|
|
// host API
|
|
template <typename ADataType,
|
|
typename BDataType,
|
|
typename AccDataType,
|
|
typename CDataType,
|
|
typename FlatmmConfig,
|
|
typename ALayout,
|
|
typename BLayout,
|
|
typename CLayout>
|
|
float flatmm_calc(const ck_tile::FlatmmHostArgs<>& args, const ck_tile::stream_config& s);
|