Skip to content

GitLab

Commit fb9f0757 authored by Jun Liu's avatar Jun Liu
Browse files

Merge branch 'amd-develop' into amd-master

parents 4d914af3 c5ad2e80
Hide whitespace changes
Inline Side-by-side
Showing with 1145 additions and 256 deletions
include/ck_tile/ops/fmha/pipeline/block_fmha_pipeline_qx_ks_vs_custom_policy.hpp
View file @ fb9f0757
......@@ -863,6 +863,8 @@ struct BlockFmhaPipelineQXKSVSCustomPolicy : BlockFmhaPipelineQXCustomPolicy<QLo
constexpr index_t K0 = kKPerBlock / K1;
constexpr index_t N2 = get_warp_size() / K0;
constexpr index_t N1 = kBlockSize / get_warp_size();
static_assert(N2 != 0, "N2 is zero, which will lead to a division by zero error.");
static_assert(N1 != 0, "N1 is zero, which will lead to a division by zero error.");
constexpr index_t N0 = kNPerBlock / (N2 * N1);
static_assert(N0 != 0);
......
include/ck_tile/ops/fused_moe/kernel/moe_sorting_kernel.hpp 0 → 100644
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
#include "ck_tile/ops/common.hpp"
#include "ck_tile/ops/elementwise.hpp"
#include "ck_tile/host/hip_check_error.hpp"
#include <string>
#include <type_traits>
namespace ck_tile {
struct MoeSortingHostArgs
{
const void* p_topk_ids;
const void* p_weights;
void* p_sorted_token_ids;
void* p_sorted_weights;
void* p_sorted_expert_ids;
void* p_total_tokens_post_pad;
void* p_moe_buf;
index_t tokens;
index_t unit_size;
index_t num_experts;
index_t topk;
index_t moe_buf_bytes;
};
template <typename Problem_>
struct MoeSortingKernel
{
using Problem = remove_cvref_t<Problem_>;
using IndexType = typename Problem::IndexType;
using WeightType = typename Problem::WeightType;
typedef MoeSortingHostArgs MoeSortingKargs;
using Hargs = MoeSortingHostArgs;
struct Kargs
{
const void* p_topk_ids;
const void* p_weights;
void* p_sorted_token_ids;
void* p_sorted_weights;
void* p_sorted_expert_ids;
void* p_total_tokens_post_pad;
void* p_moe_buf;
index_t tokens;
index_t num_experts;
index_t moe_buf_bytes;
index_t tokens_per_thread;
mdiv unit_size_mdiv;
mdiv topk_mdiv;
};
CK_TILE_HOST static constexpr auto GridSize(const Hargs& h)
{
// TODO: assume num-experts not too much
return dim3(1 + ck_tile::integer_divide_ceil(h.moe_buf_bytes, BlockSize(h).x * 16));
}
CK_TILE_HOST static constexpr auto BlockSize(const Hargs& h)
{
return dim3(ck_tile::integer_least_multiple(h.num_experts, ck_tile::get_warp_size()));
}
// in byte
CK_TILE_HOST static constexpr auto GetSmemSize(const Hargs& h)
{
const auto blocks = BlockSize(h);
return ((blocks.x + 1) * h.num_experts + (h.num_experts + 1)) * sizeof(index_t);
}
CK_TILE_HOST static constexpr auto MakeKargs(const Hargs& h)
{
Kargs k;
k.p_topk_ids = h.p_topk_ids;
k.p_weights = h.p_weights;
k.p_sorted_token_ids = h.p_sorted_token_ids;
k.p_sorted_weights = h.p_sorted_weights;
k.p_sorted_expert_ids = h.p_sorted_expert_ids;
k.p_moe_buf = h.p_moe_buf;
k.p_total_tokens_post_pad = h.p_total_tokens_post_pad;
k.tokens = h.tokens;
k.num_experts = h.num_experts;
k.moe_buf_bytes = h.moe_buf_bytes;
const auto blocks = BlockSize(h);
k.tokens_per_thread = integer_divide_ceil(h.tokens * h.topk, blocks.x);
k.unit_size_mdiv = mdiv{static_cast<uint32_t>(h.unit_size)};
k.topk_mdiv = mdiv{static_cast<uint32_t>(h.topk)};
return k;
}
CK_TILE_DEVICE index_t calc_index(index_t total_col, index_t row, index_t col) const
{
return row * total_col + col;
}
CK_TILE_DEVICE void moe_buf_set_zero_kernel(uint8x16_t* buf, index_t buf_bytes) const
{
const index_t offset = (blockIdx.x - 1) * blockDim.x + threadIdx.x;
if(offset < buf_bytes / 16)
{
buf[offset] = uint8x16_t{0};
}
}
CK_TILE_DEVICE void moe_align_block_size_kernel(const IndexType* __restrict__ topk_id,
const WeightType* __restrict__ weights,
index_t* p_sorted_token_ids,
WeightType* p_sorted_weights,
index_t* p_sorted_expert_ids,
index_t* p_total_tokens_post_pad,
const index_t num_experts,
const index_t tokens_per_thread,
const index_t numel,
const mdiv unit_size_mdiv,
const mdiv topk_mdiv,
void* smem) const
{
const index_t tid = static_cast<index_t>(threadIdx.x);
const index_t start_idx = tid * tokens_per_thread;
index_t* shared_mem = reinterpret_cast<index_t*>(smem);
index_t* tokens_cnts = shared_mem; // 2d: (blockDim.x + 1, num_experts)
index_t* cumsum = shared_mem + (blockDim.x + 1) * num_experts; // 1: (num_experts + 1)
for(int i = 0; i < num_experts; ++i)
{
tokens_cnts[calc_index(num_experts, tid + 1, i)] = 0;
}
#pragma unroll Problem_::InternalLoadUnroll
for(int i = start_idx; i < numel && i < start_idx + tokens_per_thread; ++i)
{
++tokens_cnts[calc_index(num_experts, tid + 1, topk_id[i])];
}
__syncthreads();
if(tid < num_experts)
{
tokens_cnts[calc_index(num_experts, 0, tid)] = 0;
for(int i = 1; i <= static_cast<index_t>(blockDim.x); ++i)
{
tokens_cnts[calc_index(num_experts, i, tid)] +=
tokens_cnts[calc_index(num_experts, i - 1, tid)];
}
}
// __syncthreads();
if(tid == 0)
{
cumsum[0] = 0;
for(int i = 1; i <= num_experts; ++i)
{
auto current_units = [&]() {
index_t x_ = tokens_cnts[calc_index(num_experts, blockDim.x, i - 1)] +
unit_size_mdiv.divisor - 1;
index_t y_ = unit_size_mdiv.div(x_);
return max(y_, 1) * unit_size_mdiv.divisor;
}();
cumsum[i] = cumsum[i - 1] + current_units;
}
*p_total_tokens_post_pad = cumsum[num_experts];
}
__syncthreads();
if(tid < num_experts)
{
for(int i = cumsum[tid]; i < cumsum[tid + 1]; i += unit_size_mdiv.divisor)
{
p_sorted_expert_ids[unit_size_mdiv.div(i)] = tid;
}
}
#pragma unroll Problem_::InternalLoadUnroll
for(int i = start_idx; i < numel && i < start_idx + tokens_per_thread; ++i)
{
index_t expert_id = topk_id[i];
index_t rank_post_pad =
tokens_cnts[calc_index(num_experts, tid, expert_id)] + cumsum[expert_id];
p_sorted_token_ids[rank_post_pad] = topk_mdiv.div(i);
p_sorted_weights[rank_post_pad] = weights[i];
++tokens_cnts[calc_index(num_experts, tid, expert_id)];
}
const index_t prefill_token = topk_mdiv.div(numel);
if(tid < num_experts)
{
index_t expert_offset =
cumsum[tid] + tokens_cnts[calc_index(num_experts, blockDim.x, tid)];
while(expert_offset < cumsum[tid + 1])
{
p_sorted_token_ids[expert_offset] = prefill_token;
p_sorted_weights[expert_offset] = static_cast<WeightType>(0.0);
expert_offset++;
}
}
}
CK_TILE_DEVICE void operator()(Kargs kargs) const
{
if(blockIdx.x > 0)
{
if(kargs.p_moe_buf)
{
moe_buf_set_zero_kernel(reinterpret_cast<uint8x16_t*>(kargs.p_moe_buf),
kargs.moe_buf_bytes);
}
return;
}
const size_t numel = kargs.tokens * kargs.topk_mdiv.divisor;
extern __shared__ char smem[];
return moe_align_block_size_kernel(static_cast<const IndexType*>(kargs.p_topk_ids),
static_cast<const WeightType*>(kargs.p_weights),
static_cast<IndexType*>(kargs.p_sorted_token_ids),
static_cast<WeightType*>(kargs.p_sorted_weights),
static_cast<IndexType*>(kargs.p_sorted_expert_ids),
static_cast<IndexType*>(kargs.p_total_tokens_post_pad),
kargs.num_experts,
kargs.tokens_per_thread,
numel,
kargs.unit_size_mdiv,
kargs.topk_mdiv,
smem);
}
};
} // namespace ck_tile
include/ck_tile/ops/fused_moe/pipeline/moe_sorting_pipeline.hpp 0 → 100644
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
#include "ck_tile/ops/fused_moe/pipeline/moe_sorting_policy.hpp"
#include <string>
#include <type_traits>
#ifndef TOPK_SOFTMAX_USE_RAW_TILE_WINDOW
#define TOPK_SOFTMAX_USE_RAW_TILE_WINDOW 0
#endif
namespace ck_tile {
// template <typename Problem_, typename Policy_ = MoeSortingPolicy>
// struct MoeSortingPipeline
// {
// // TODO: this kernel only support warp per row
// using Problem = remove_cvref_t<Problem_>;
// using Policy = remove_cvref_t<Policy_>;
// using WeightType = typename Problem::WeightType;
// template <typename TopkIdWindow, typename WeightWindow>
// CK_TILE_DEVICE auto operator()(const TopkIdWindow& topk_id_window,
// const WeightWindow& weight_window,
// index_t* p_sorted_token_ids,
// WeightType* p_sorted_weights,
// index_t* p_sorted_expert_ids,
// index_t* p_total_tokens_post_pad,
// const index_t num_experts,
// const index_t unit_size,
// const size_t numel,
// const index_t topk)
// {
// }
// };
} // namespace ck_tile
include/ck_tile/ops/fused_moe/pipeline/moe_sorting_policy.hpp 0 → 100644
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
#include "ck_tile/ops/softmax.hpp"
#include "ck_tile/ops/topk.hpp"
namespace ck_tile {
struct MoeSortingPolicy
{
};
} // namespace ck_tile
include/ck_tile/ops/fused_moe/pipeline/moe_sorting_problem.hpp 0 → 100644
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
#include <string>
#include <type_traits>
namespace ck_tile {
template <typename IndexType_, typename WeightType_, index_t InternalLoadUnroll_>
struct MoeSortingProblem
{
// TODO: this kernel only support warp per row
using WeightType = remove_cvref_t<WeightType_>;
using IndexType = remove_cvref_t<IndexType_>;
static constexpr index_t WarpSize = get_warp_size();
static constexpr index_t WarpsPerBlock = 1;
static constexpr index_t InternalLoadUnroll = InternalLoadUnroll_;
};
} // namespace ck_tile
include/ck_tile/ops/gemm/kernel/gemm_kernel.hpp
View file @ fb9f0757
......@@ -115,12 +115,22 @@ struct GemmKernel
}
}();
auto a_pad_view = pad_tensor_view(
a_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kK>{}),
// somehow clang-format is splitting below line into multiple.
// clang-format off
sequence<false, GemmPipeline::kPadA>{});
auto a_pad_view = [&]() {
if constexpr(std::is_same_v<ALayout, tensor_layout::gemm::RowMajor>)
{
return pad_tensor_view(
a_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kK>{}),
sequence<false, GemmPipeline::kPadK>{});
}
else
{
return pad_tensor_view(
a_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kK>{}),
sequence<GemmPipeline::kPadM, false>{});
}
}();
// clang-format on
auto a_block_window = make_tile_window(
......@@ -128,12 +138,22 @@ struct GemmKernel
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kK>{}),
{i_m, 0});
auto b_pad_view = pad_tensor_view(
b_tensor_view,
make_tuple(number<TilePartitioner::kN>{}, number<TilePartitioner::kK>{}),
// clang-format off
sequence<false, GemmPipeline::kPadB>{});
// clang-format on
auto b_pad_view = [&]() {
if constexpr(std::is_same_v<BLayout, tensor_layout::gemm::ColumnMajor>)
{
return pad_tensor_view(
b_tensor_view,
make_tuple(number<TilePartitioner::kN>{}, number<TilePartitioner::kK>{}),
sequence<false, GemmPipeline::kPadK>{});
}
else
{
return pad_tensor_view(
b_tensor_view,
make_tuple(number<TilePartitioner::kN>{}, number<TilePartitioner::kK>{}),
sequence<GemmPipeline::kPadN, false>{});
}
}();
auto b_block_window = make_tile_window(
b_pad_view,
......@@ -171,18 +191,28 @@ struct GemmKernel
}
}();
auto c_pad_view = pad_tensor_view(
c_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kN>{}),
// clang-format off
sequence<false, GemmPipeline::kPadC>{});
// clang-format on
auto c_block_window = make_tile_window(
auto c_pad_view = [&]() {
if constexpr(std::is_same_v<CLayout, tensor_layout::gemm::RowMajor>)
{
return pad_tensor_view(
c_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kN>{}),
sequence<false, GemmPipeline::kPadN>{});
}
else
{
return pad_tensor_view(
c_tensor_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kN>{}),
sequence<GemmPipeline::kPadM, false>{});
}
}();
auto CBlockWindow_pad = make_tile_window(
c_pad_view,
make_tuple(number<TilePartitioner::kM>{}, number<TilePartitioner::kN>{}),
{i_m, i_n});
EpiloguePipeline{}(c_block_window, c_block_tile);
EpiloguePipeline{}(CBlockWindow_pad, c_block_tile);
}
};
......
include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_mem.hpp
View file @ fb9f0757
......@@ -113,9 +113,9 @@ struct GemmPipelineAgBgCrMem : public BaseGemmPipelineAgBgCrMem<Problem>
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;
static constexpr bool kPadM = Problem::kPadM;
static constexpr bool kPadN = Problem::kPadN;
static constexpr bool kPadK = Problem::kPadK;
// Where is the right place for HasHotLoop and TailNum ???
static constexpr bool HasHotLoop = Problem::HasHotLoop;
......
include/ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1.hpp
View file @ fb9f0757
......@@ -33,9 +33,9 @@ struct GemmPipelineAGmemBGmemCRegV1
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;
static constexpr bool kPadM = Problem::kPadM;
static constexpr bool kPadN = Problem::kPadN;
static constexpr bool kPadK = Problem::kPadK;
CK_TILE_HOST_DEVICE static constexpr index_t GetStaticLdsSize()
{
......@@ -101,11 +101,8 @@ struct GemmPipelineAGmemBGmemCRegV1
Policy::template MakeADramTileDistribution<Problem>());
// A LDS tile window for store
auto a_copy_lds_window =
make_tile_window(a_lds_block,
make_tuple(number<kMPerBlock>{}, number<kKPerBlock>{}),
{0, 0},
a_copy_dram_window.get_tile_distribution());
auto a_copy_lds_window = make_tile_window(
a_lds_block, make_tuple(number<kMPerBlock>{}, number<kKPerBlock>{}), {0, 0});
// B DRAM tile window for load
auto b_copy_dram_window =
......@@ -115,11 +112,8 @@ struct GemmPipelineAGmemBGmemCRegV1
Policy::template MakeBDramTileDistribution<Problem>());
// B LDS tile window for store
auto b_copy_lds_window =
make_tile_window(b_lds_block,
make_tuple(number<kNPerBlock>{}, number<kKPerBlock>{}),
{0, 0},
b_copy_dram_window.get_tile_distribution());
auto b_copy_lds_window = make_tile_window(
b_lds_block, make_tuple(number<kNPerBlock>{}, number<kKPerBlock>{}), {0, 0});
// A LDS tile for block GEMM
auto a_lds_gemm_window = make_tile_window(
......@@ -149,12 +143,32 @@ struct GemmPipelineAGmemBGmemCRegV1
tile_elementwise_inout([](auto& c) { c = 0; }, c_block_tile);
// LDS write 0
const auto a_block_tile_tmp = tile_elementwise_in(a_element_func, a_block_tile);
store_tile(a_copy_lds_window, a_block_tile_tmp);
if constexpr(std::is_same_v<ALayout, tensor_layout::gemm::ColumnMajor>)
{
auto a_shuffle_tmp = make_static_distributed_tensor<ADataType>(
Policy::template MakeShuffledARegBlockDescriptor<Problem>());
shuffle_tile(a_shuffle_tmp, a_block_tile);
const auto a_block_tile_tmp = tile_elementwise_in(a_element_func, a_shuffle_tmp);
store_tile(a_copy_lds_window, a_block_tile_tmp);
}
else
{
store_tile(a_copy_lds_window, tile_elementwise_in(a_element_func, a_block_tile));
}
// LDS write 0
const auto b_block_tile_tmp = tile_elementwise_in(b_element_func, b_block_tile);
store_tile(b_copy_lds_window, b_block_tile_tmp);
if constexpr(std::is_same_v<BLayout, tensor_layout::gemm::RowMajor>)
{
auto b_shuffle_tmp = make_static_distributed_tensor<BDataType>(
Policy::template MakeShuffledBRegBlockDescriptor<Problem>());
shuffle_tile(b_shuffle_tmp, b_block_tile);
const auto b_block_tile_tmp = tile_elementwise_in(b_element_func, b_shuffle_tmp);
store_tile(b_copy_lds_window, b_block_tile_tmp);
}
else
{
store_tile(b_copy_lds_window, tile_elementwise_in(b_element_func, b_block_tile));
}
}
index_t iCounter = num_loop - 1;
......@@ -180,8 +194,19 @@ struct GemmPipelineAGmemBGmemCRegV1
store_tile(a_copy_lds_window, a_block_tile_tmp);
// LDS write i + 1
const auto b_block_tile_tmp = tile_elementwise_in(b_element_func, b_block_tile);
store_tile(b_copy_lds_window, b_block_tile_tmp);
if constexpr(std::is_same_v<BLayout, tensor_layout::gemm::RowMajor>)
{
auto b_shuffle_tmp_loop = make_static_distributed_tensor<BDataType>(
Policy::template MakeShuffledBRegBlockDescriptor<Problem>());
shuffle_tile(b_shuffle_tmp_loop, b_block_tile);
store_tile(b_copy_lds_window,
tile_elementwise_in(b_element_func, b_shuffle_tmp_loop));
}
else
{
const auto b_block_tile_tmp = tile_elementwise_in(b_element_func, b_block_tile);
store_tile(b_copy_lds_window, b_block_tile_tmp);
}
iCounter--;
}
......
include/ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1_default_policy.hpp
View file @ fb9f0757
......@@ -11,6 +11,7 @@ namespace ck_tile {
// Default policy class should not be templated, put template on member functions instead
struct GemmPipelineAGmemBGmemCRegV1DefaultPolicy
{
#if 0
// 2d
template <typename Problem>
......@@ -116,6 +117,20 @@ struct GemmPipelineAGmemBGmemCRegV1DefaultPolicy
return smem_size;
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetSmemPackA()
{
using ADataType = remove_cvref_t<typename Problem::ADataType>;
return Problem::VectorLoadSize / sizeof(ADataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetSmemPackB()
{
using BDataType = remove_cvref_t<typename Problem::BDataType>;
return Problem::VectorLoadSize / sizeof(BDataType);
}
#elif 1
// fake XOR
template <typename Problem>
......@@ -192,80 +207,269 @@ struct GemmPipelineAGmemBGmemCRegV1DefaultPolicy
CK_TILE_HOST_DEVICE static constexpr auto MakeADramTileDistribution()
{
using ADataType = remove_cvref_t<typename Problem::ADataType>;
constexpr index_t kBlockSize = Problem::kBlockSize;
constexpr index_t kMPerBlock = Problem::BlockGemmShape::kM;
constexpr index_t kKPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = 16 / sizeof(ADataType);
constexpr index_t K0 = kKPerBlock / K1;
constexpr index_t M2 = get_warp_size() / K0;
#if 1 // coalesce reading for each blocks
constexpr index_t M1 = kBlockSize / get_warp_size();
static_assert(M2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(M1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t M0 = kMPerBlock / (M2 * M1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
#else // coalesce reading for each warps
constexpr index_t M0 = kBlockSize / get_warp_size();
constexpr index_t M1 = kMPerBlock / (M2 * M0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
#endif
using ALayout = remove_cvref_t<typename Problem::ALayout>;
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t MPerBlock = Problem::BlockGemmShape::kM;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
if constexpr(std::is_same_v<ALayout, ck_tile::tensor_layout::gemm::ColumnMajor>)
{
constexpr index_t M1 = Problem::VectorLoadSize / sizeof(ADataType);
constexpr index_t M0 = MPerBlock / M1;
constexpr index_t total_pixels = MPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % M1 == 0);
constexpr index_t K3 = total_pixels / M1;
constexpr index_t KPack = GetSmemPackA<Problem>();
static_assert(KPack % K3 == 0);
constexpr index_t K2 = KPack / K3;
if constexpr(get_warp_size() % (K2 * M0))
{
constexpr index_t K1 = get_warp_size() / (K2 * M0);
constexpr index_t K0 = BlockSize / get_warp_size();
static_assert(KPerBlock == K0 * K1 * K2 * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
else
{
constexpr index_t K1 = (K2 * M0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
}
else
{
constexpr index_t K1 = 16 / sizeof(ADataType);
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t M2 = get_warp_size() / K0;
// coalesce reading for each blocks
if constexpr(get_warp_size() % (M2 * K0) == 0)
{
constexpr index_t M1 = BlockSize / get_warp_size();
static_assert(M2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(M1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t M0 = MPerBlock / (M2 * M1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
}
else
{
constexpr index_t M0 = BlockSize / get_warp_size();
constexpr index_t M1 = MPerBlock / (M2 * M0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
}
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeBDramTileDistribution()
{
using BDataType = remove_cvref_t<typename Problem::BDataType>;
using BLayout = remove_cvref_t<typename Problem::BLayout>;
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t NPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
if constexpr(std::is_same_v<BLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
constexpr index_t N1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t N0 = NPerBlock / N1;
constexpr index_t total_pixels = NPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % N1 == 0);
constexpr index_t K3 = total_pixels / N1;
constexpr index_t KPack = GetSmemPackB<Problem>();
static_assert(KPack % K3 == 0);
constexpr index_t K2 = KPack / K3;
if constexpr(get_warp_size() % (K2 * N0) == 0)
{
constexpr index_t K1 = get_warp_size() / (K2 * N0);
constexpr index_t K0 = BlockSize / get_warp_size();
static_assert(KPerBlock == K0 * K1 * K2 * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
else
{
constexpr index_t K1 = (K2 * N0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
}
else
{
constexpr index_t K1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t N2 = get_warp_size() / K0;
// coalesce reading for each blocks
if constexpr(get_warp_size() % (N2 * K0) == 0)
{
constexpr index_t N1 = BlockSize / get_warp_size();
static_assert(N2 != 0, "N2 is zero, which will lead to a division by zero error.");
static_assert(N1 != 0, "N1 is zero, which will lead to a division by zero error.");
constexpr index_t N0 = NPerBlock / (N2 * N1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
}
// coalesce reading for each warps
else
{
constexpr index_t N0 = BlockSize / get_warp_size();
constexpr index_t N1 = NPerBlock / (N2 * N0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
}
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeShuffledBRegBlockDescriptor()
{
using BLayout = remove_cvref_t<typename Problem::BLayout>;
using BDataType = remove_cvref_t<typename Problem::BDataType>;
static_assert(std::is_same_v<BLayout, ck_tile::tensor_layout::gemm::RowMajor>);
constexpr index_t kBlockSize = Problem::kBlockSize;
constexpr index_t kNPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t kKPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = 16 / sizeof(BDataType);
constexpr index_t K0 = kKPerBlock / K1;
constexpr index_t N2 = get_warp_size() / K0;
#if 1 // coalesce reading for each blocks
constexpr index_t N1 = kBlockSize / get_warp_size();
static_assert(N2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(N1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t N0 = kNPerBlock / (N2 * N1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
#else // coalesce reading for each warps
constexpr index_t N0 = kBlockSize / get_warp_size();
constexpr index_t N1 = kNPerBlock / (N2 * N0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
#endif
constexpr index_t N1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t N0 = kNPerBlock / N1;
constexpr index_t total_pixels = kNPerBlock * kKPerBlock / kBlockSize;
static_assert(total_pixels % N1 == 0);
constexpr index_t K3 = total_pixels / N1;
constexpr index_t kKPack = GetSmemPackB<Problem>();
static_assert(kKPack % K3 == 0);
constexpr index_t K2 = kKPack / K3; // TODO: this dimention could be outside single wave
constexpr index_t warp_size = get_warp_size();
if constexpr(warp_size % (K2 * N0) == 0)
{
constexpr index_t K1 = warp_size / (K2 * N0);
constexpr index_t K0 = kBlockSize / warp_size;
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
else
{
constexpr index_t K1 = (K2 * N0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = kBlockSize / get_warp_size() / K1;
static_assert(kKPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeShuffledARegBlockDescriptor()
{
using ALayout = remove_cvref_t<typename Problem::ALayout>;
using ADataType = remove_cvref_t<typename Problem::ADataType>;
static_assert(std::is_same_v<ALayout, ck_tile::tensor_layout::gemm::RowMajor>);
constexpr index_t kBlockSize = Problem::kBlockSize;
constexpr index_t kMPerBlock = Problem::BlockGemmShape::kM;
constexpr index_t kKPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t M1 = Problem::VectorLoadSize / sizeof(ADataType);
constexpr index_t M0 = kMPerBlock / M1;
constexpr index_t total_pixels = kMPerBlock * kKPerBlock / kBlockSize;
static_assert(total_pixels % M1 == 0);
constexpr index_t K3 = total_pixels / M1;
constexpr index_t kKPack = GetSmemPackA<Problem>();
static_assert(kKPack % K3 == 0);
constexpr index_t K2 = kKPack / K3; // TODO: this dimention could be outside single wave
constexpr index_t warp_size = get_warp_size();
if constexpr(warp_size % (K2 * M0) == 0)
{
constexpr index_t K1 = warp_size / (K2 * M0);
constexpr index_t K0 = kBlockSize / warp_size;
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
else
{
constexpr index_t K1 = (K2 * M0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = kBlockSize / get_warp_size() / K1;
static_assert(kKPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
}
template <typename Problem>
......
include/ck_tile/ops/gemm/pipeline/gemm_pipeline_problem.hpp
View file @ fb9f0757
......@@ -3,40 +3,133 @@
#pragma once
#include "ck_tile/core.hpp"
#include "ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_scheduler.hpp"
namespace ck_tile {
static constexpr int _VectorSize = 16;
template <typename ADataType_,
typename BDataType_,
typename CDataType_,
typename BlockGemmShape_,
typename TileGemmTraits_>
struct GemmPipelineProblem
struct GemmPipelineProblemBase
{
using ADataType = remove_cvref_t<ADataType_>;
using BDataType = remove_cvref_t<BDataType_>;
using CDataType = remove_cvref_t<CDataType_>;
using GemmTraits = remove_cvref_t<TileGemmTraits_>;
using ADataType = remove_cvref_t<ADataType_>;
using BDataType = remove_cvref_t<BDataType_>;
using CDataType = remove_cvref_t<CDataType_>;
using BlockGemmShape = remove_cvref_t<BlockGemmShape_>;
using GemmTraits = remove_cvref_t<TileGemmTraits_>;
using ALayout = remove_cvref_t<typename GemmTraits::ALayout>;
using BLayout = remove_cvref_t<typename GemmTraits::BLayout>;
using CLayout = remove_cvref_t<typename GemmTraits::CLayout>;
static constexpr index_t kBlockSize = BlockGemmShape::NumWarps * get_warp_size();
static constexpr bool kPadA = GemmTraits::kPadA;
static constexpr bool kPadB = GemmTraits::kPadB;
static constexpr bool kPadC = GemmTraits::kPadC;
static constexpr index_t VectorLoadSize = GemmTraits::_VectorSize;
static constexpr index_t kBlockSize = BlockGemmShape::NumWarps * get_warp_size();
static constexpr bool kPadM = GemmTraits::kPadM;
static constexpr bool kPadN = GemmTraits::kPadN;
static constexpr bool kPadK = GemmTraits::kPadK;
CK_TILE_HOST_DEVICE static constexpr auto GetAlignmentA()
{
if constexpr(std::is_same_v<ALayout, ck_tile::tensor_layout::gemm::ColumnMajor>)
{
constexpr index_t pixels_per_thread =
BlockGemmShape::kM * BlockGemmShape::kK / kBlockSize;
return pixels_per_thread < VectorLoadSize / sizeof(ADataType)
? pixels_per_thread
: VectorLoadSize / sizeof(ADataType);
}
else
{
return VectorLoadSize / sizeof(ADataType);
}
}
CK_TILE_HOST_DEVICE static constexpr auto GetAlignmentB()
{
if constexpr(std::is_same_v<BLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
constexpr index_t pixels_per_thread =
BlockGemmShape::kN * BlockGemmShape::kK / kBlockSize;
return pixels_per_thread < VectorLoadSize / sizeof(BDataType)
? pixels_per_thread
: VectorLoadSize / sizeof(BDataType);
}
else
{
return VectorLoadSize / sizeof(BDataType);
}
}
CK_TILE_HOST_DEVICE static constexpr auto GetAlignmentC()
{
if constexpr(std::is_same_v<CLayout, ck_tile::tensor_layout::gemm::ColumnMajor>)
{
constexpr index_t N1 = kBlockSize / get_warp_size();
constexpr index_t N2 = std::min(BlockGemmShape::kN / N1, get_warp_size());
constexpr index_t M0 = get_warp_size() / N2;
constexpr index_t M1 = BlockGemmShape::kM / M0;
static constexpr index_t VectorSizeA = kPadA ? 1 : _VectorSize / sizeof(ADataType);
static constexpr index_t VectorSizeB = kPadB ? 1 : _VectorSize / sizeof(BDataType);
static constexpr index_t VectorSizeC = kPadC ? 1 : _VectorSize / sizeof(CDataType);
return std::min(M1, static_cast<index_t>(VectorLoadSize / sizeof(CDataType)));
}
else
{
constexpr index_t M1 = kBlockSize / get_warp_size();
constexpr index_t M2 = std::min(BlockGemmShape::kM / M1, get_warp_size());
constexpr index_t N0 = get_warp_size() / M2;
constexpr index_t N1 = BlockGemmShape::kN / N0;
return std::min(N1, static_cast<index_t>(VectorLoadSize / sizeof(CDataType)));
}
}
static constexpr index_t VectorSizeA = []() {
if constexpr(std::is_same_v<ALayout, tensor_layout::gemm::RowMajor>)
{
return kPadK ? 1 : GetAlignmentA();
}
else
{
return kPadM ? 1 : GetAlignmentA();
}
}();
static constexpr index_t VectorSizeB = []() {
if constexpr(std::is_same_v<BLayout, tensor_layout::gemm::ColumnMajor>)
{
return kPadN ? 1 : GetAlignmentB();
}
else
{
return kPadK ? 1 : GetAlignmentB();
}
}();
static constexpr index_t VectorSizeC = []() {
if constexpr(std::is_same_v<CLayout, tensor_layout::gemm::RowMajor>)
{
return kPadN ? 1 : GetAlignmentC();
}
else
{
return kPadM ? 1 : GetAlignmentC();
}
}();
};
// Alias for GemmPipelineProblem
template <typename ADataType_,
typename BDataType_,
typename CDataType_,
typename BlockGemmShape_,
typename TileGemmTraits_>
using GemmPipelineProblem =
GemmPipelineProblemBase<ADataType_, BDataType_, CDataType_, BlockGemmShape_, TileGemmTraits_>;
template <typename ADataType_,
typename BDataType_,
typename CDataType_,
......@@ -45,30 +138,15 @@ template <typename ADataType_,
GemmPipelineScheduler Scheduler_ = GemmPipelineScheduler::Intrawave,
bool HasHotLoop_ = true,
TailNumber TailNum_ = TailNumber::Full>
struct UniversalGemmPipelineProblem
struct UniversalGemmPipelineProblem : public GemmPipelineProblemBase<ADataType_,
BDataType_,
CDataType_,
BlockGemmShape_,
TileGemmTraits_>
{
using ADataType = remove_cvref_t<ADataType_>;
using BDataType = remove_cvref_t<BDataType_>;
using CDataType = remove_cvref_t<CDataType_>;
using BlockGemmShape = remove_cvref_t<BlockGemmShape_>;
using GemmTraits = remove_cvref_t<TileGemmTraits_>;
using ALayout = remove_cvref_t<typename GemmTraits::ALayout>;
using BLayout = remove_cvref_t<typename GemmTraits::BLayout>;
using CLayout = remove_cvref_t<typename GemmTraits::CLayout>;
static constexpr auto Scheduler = Scheduler_;
static constexpr auto HasHotLoop = HasHotLoop_;
static constexpr auto TailNum = TailNum_;
static constexpr index_t kBlockSize = BlockGemmShape::NumWarps * get_warp_size();
static constexpr bool kPadA = GemmTraits::kPadA;
static constexpr bool kPadB = GemmTraits::kPadB;
static constexpr bool kPadC = GemmTraits::kPadC;
static constexpr index_t VectorSizeA = kPadA ? _VectorSize / sizeof(ADataType) : 1;
static constexpr index_t VectorSizeB = kPadB ? _VectorSize / sizeof(BDataType) : 1;
static constexpr index_t VectorSizeC = kPadC ? _VectorSize / sizeof(CDataType) : 1;
static constexpr auto Scheduler = Scheduler_;
static constexpr auto HasHotLoop = HasHotLoop_;
static constexpr auto TailNum = TailNum_;
};
} // namespace ck_tile
include/ck_tile/ops/gemm/pipeline/gemm_universal_pipeline_ag_bg_cr_policy.hpp
View file @ fb9f0757
......@@ -9,12 +9,8 @@
namespace ck_tile {
// UniversalGemm Policy
template <typename LayoutA_, typename LayoutB_, typename LayoutC_>
struct UniversalGemmPipelineAgBgCrPolicy
{
using LayoutA = remove_cvref_t<LayoutA_>;
using LayoutB = remove_cvref_t<LayoutB_>;
using LayoutC = remove_cvref_t<LayoutC_>;
static constexpr auto I0 = number<0>{};
static constexpr auto I1 = number<1>{};
......@@ -34,13 +30,14 @@ struct UniversalGemmPipelineAgBgCrPolicy
TransposeC>;
using ADataType = remove_cvref_t<typename Problem::ADataType>;
using ALayout = remove_cvref_t<typename Problem::ALayout>;
constexpr index_t MPerBlock = Problem::BlockGemmShape::kM;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = WarpGemm::kK;
constexpr index_t K0 = KPerBlock / K1;
if constexpr(std::is_same<tensor_layout::gemm::RowMajor, LayoutA>::value)
if constexpr(std::is_same<tensor_layout::gemm::RowMajor, ALayout>::value)
{
constexpr auto MLdsLayer = 32 * 4 / KPerBlock / sizeof(ADataType) < 1
? 1
......@@ -176,13 +173,15 @@ struct UniversalGemmPipelineAgBgCrPolicy
using BDataType = remove_cvref_t<typename Problem::BDataType>;
using BLayout = remove_cvref_t<typename Problem::BLayout>;
constexpr index_t NPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = WarpGemm::kK;
constexpr index_t K0 = KPerBlock / K1;
if constexpr(std::is_same<tensor_layout::gemm::ColumnMajor, LayoutB>::value)
if constexpr(std::is_same<tensor_layout::gemm::ColumnMajor, BLayout>::value)
{
// NLdsLayer * K0 as logical Bank
constexpr auto NLdsLayer = 32 * 4 / KPerBlock / sizeof(BDataType) < 1
......@@ -331,72 +330,285 @@ struct UniversalGemmPipelineAgBgCrPolicy
return smem_size;
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetSmemPackA()
{
using ADataType = remove_cvref_t<typename Problem::ADataType>;
return Problem::VectorLoadSize / sizeof(ADataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetSmemPackB()
{
using BDataType = remove_cvref_t<typename Problem::BDataType>;
return Problem::VectorLoadSize / sizeof(BDataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeADramTileDistribution()
{
using WarpGemm = WarpGemmMfmaDispatcher<typename Problem::ADataType,
typename Problem::BDataType,
typename Problem::CDataType,
Problem::BlockGemmShape::WarpTile::at(I0),
Problem::BlockGemmShape::WarpTile::at(I1),
Problem::BlockGemmShape::WarpTile::at(I2),
TransposeC>;
using ADataType = remove_cvref_t<typename Problem::ADataType>;
using ALayout = remove_cvref_t<typename Problem::ALayout>;
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t MPerBlock = Problem::BlockGemmShape::kM;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = WarpGemm::kK;
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t M2 = get_warp_size() / K0;
constexpr index_t M1 = BlockSize / get_warp_size();
static_assert(M2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(M1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t M0 = MPerBlock / (M2 * M1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
if constexpr(std::is_same_v<ALayout, ck_tile::tensor_layout::gemm::ColumnMajor>)
{
constexpr index_t M1 = Problem::VectorLoadSize / sizeof(ADataType);
constexpr index_t M0 = MPerBlock / M1;
constexpr index_t total_pixels = MPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % M1 == 0);
constexpr index_t K3 = total_pixels / M1;
constexpr index_t KPack = GetSmemPackA<Problem>();
static_assert(KPack % K3 == 0);
constexpr index_t K2 = KPack / K3;
if constexpr(get_warp_size() % (K2 * M0) == 0)
{
constexpr index_t K1 = get_warp_size() / (K2 * M0);
constexpr index_t K0 = BlockSize / get_warp_size();
static_assert(KPerBlock == K0 * K1 * K2 * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
else
{
constexpr index_t K1 = (K2 * M0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
}
else
{
constexpr index_t K1 = Problem::VectorLoadSize / sizeof(ADataType);
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t M2 = get_warp_size() / K0;
if constexpr(get_warp_size() % (M2 * K0) == 0)
{
constexpr index_t M1 = BlockSize / get_warp_size();
static_assert(M2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(M1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t M0 = MPerBlock / (M2 * M1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
}
else
{
constexpr index_t M0 = BlockSize / get_warp_size();
constexpr index_t M1 = MPerBlock / (M2 * M0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1, M2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
}
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeBDramTileDistribution()
{
using WarpGemm = WarpGemmMfmaDispatcher<typename Problem::ADataType,
typename Problem::BDataType,
typename Problem::CDataType,
Problem::BlockGemmShape::WarpTile::at(I0),
Problem::BlockGemmShape::WarpTile::at(I1),
Problem::BlockGemmShape::WarpTile::at(I2),
TransposeC>;
using BDataType = remove_cvref_t<typename Problem::BDataType>;
using BLayout = remove_cvref_t<typename Problem::BLayout>;
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t NPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t K1 = WarpGemm::kK;
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t N2 = get_warp_size() / K0;
constexpr index_t N1 = BlockSize / get_warp_size();
static_assert(N2 != 0, "M2 is zero, which will lead to a division by zero error.");
static_assert(N1 != 0, "M1 is zero, which will lead to a division by zero error.");
constexpr index_t N0 = NPerBlock / (N2 * N1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
if constexpr(std::is_same_v<BLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
constexpr index_t N1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t N0 = NPerBlock / N1;
constexpr index_t total_pixels = NPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % N1 == 0);
constexpr index_t K3 = total_pixels / N1;
constexpr index_t KPack = GetSmemPackB<Problem>();
static_assert(KPack % K3 == 0);
constexpr index_t K2 = KPack / K3;
if constexpr(get_warp_size() % (K2 * N0) == 0)
{
constexpr index_t K1 = get_warp_size() / (K2 * N0);
constexpr index_t K0 = BlockSize / get_warp_size();
static_assert(KPerBlock == K0 * K1 * K2 * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
else
{
constexpr index_t K1 = (K2 * N0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<2, 1>,
sequence<3, 1>>{});
}
}
else
{
constexpr index_t K1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t K0 = KPerBlock / K1;
constexpr index_t N2 = get_warp_size() / K0;
// coalesce reading for each blocks
if constexpr(get_warp_size() % (N2 * K0) == 0)
{
constexpr index_t N1 = BlockSize / get_warp_size();
static_assert(N2 != 0, "N2 is zero, which will lead to a division by zero error.");
static_assert(N1 != 0, "N1 is zero, which will lead to a division by zero error.");
constexpr index_t N0 = NPerBlock / (N2 * N1);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<1>, sequence<2, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
}
// coalesce reading for each warps
else
{
constexpr index_t N0 = BlockSize / get_warp_size();
constexpr index_t N1 = NPerBlock / (N2 * N0);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1, N2>, sequence<K0, K1>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<0>, sequence<2, 0>>,
sequence<1, 2>,
sequence<1, 1>>{});
}
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeShuffledARegBlockDescriptor()
{
using ALayout = remove_cvref_t<typename Problem::ALayout>;
using ADataType = remove_cvref_t<typename Problem::ADataType>;
static_assert(std::is_same_v<ALayout, ck_tile::tensor_layout::gemm::ColumnMajor>);
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t MPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t M1 = Problem::VectorLoadSize / sizeof(ADataType);
constexpr index_t M0 = MPerBlock / M1;
constexpr index_t total_pixels = MPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % M1 == 0);
constexpr index_t K3 = total_pixels / M1;
constexpr index_t kKPack = GetSmemPackB<Problem>();
static_assert(kKPack % K3 == 0);
constexpr index_t K2 = kKPack / K3; // TODO: this dimention could be outside single wave
constexpr index_t warp_size = get_warp_size();
if constexpr(warp_size % (K2 * M0) == 0)
{
constexpr index_t K1 = warp_size / (K2 * M0);
constexpr index_t K0 = BlockSize / warp_size;
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
else
{
constexpr index_t K1 = (K2 * M0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<M0, M1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeShuffledBRegBlockDescriptor()
{
using BLayout = remove_cvref_t<typename Problem::BLayout>;
using BDataType = remove_cvref_t<typename Problem::BDataType>;
static_assert(std::is_same_v<BLayout, ck_tile::tensor_layout::gemm::RowMajor>);
constexpr index_t BlockSize = Problem::kBlockSize;
constexpr index_t NPerBlock = Problem::BlockGemmShape::kN;
constexpr index_t KPerBlock = Problem::BlockGemmShape::kK;
constexpr index_t N1 = Problem::VectorLoadSize / sizeof(BDataType);
constexpr index_t N0 = NPerBlock / N1;
constexpr index_t total_pixels = NPerBlock * KPerBlock / BlockSize;
static_assert(total_pixels % N1 == 0);
constexpr index_t K3 = total_pixels / N1;
constexpr index_t kKPack = GetSmemPackB<Problem>();
static_assert(kKPack % K3 == 0);
constexpr index_t K2 = kKPack / K3; // TODO: this dimention could be outside single wave
constexpr index_t warp_size = get_warp_size();
if constexpr(warp_size % (K2 * N0) == 0)
{
constexpr index_t K1 = warp_size / (K2 * N0);
constexpr index_t K0 = BlockSize / warp_size;
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2, K3>>,
tuple<sequence<2>, sequence<2, 1, 2>>,
tuple<sequence<0>, sequence<1, 0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
else
{
constexpr index_t K1 = (K2 * N0) / get_warp_size();
constexpr index_t K2_m = K2 / K1;
constexpr index_t K0 = BlockSize / get_warp_size() / K1;
static_assert(KPerBlock == K0 * K1 * K2_m * K3);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<N0, N1>, sequence<K0, K1, K2_m, K3>>,
tuple<sequence<2, 2>, sequence<1, 2>>,
tuple<sequence<0, 1>, sequence<0, 2>>,
sequence<1, 2>,
sequence<1, 3>>{});
}
}
template <typename Problem>
......
include/ck_tile/ops/gemm/pipeline/tile_gemm_traits.hpp
View file @ fb9f0757
......@@ -3,19 +3,23 @@
#pragma once
#include "ck_tile/core.hpp"
namespace ck_tile {
template <bool kPadA_,
bool kPadB_,
bool kPadC_,
template <bool kPadM_,
bool kPadN_,
bool kPadK_,
typename ALayout_,
typename BLayout_,
typename CLayout_>
struct TileGemmTraits
{
static constexpr bool kPadA = kPadA_;
static constexpr bool kPadB = kPadB_;
static constexpr bool kPadC = kPadC_;
static constexpr bool kPadM = kPadM_;
static constexpr bool kPadN = kPadN_;
static constexpr bool kPadK = kPadK_;
static constexpr int _VectorSize = 16;
using ALayout = ALayout_;
using BLayout = BLayout_;
......
include/ck_tile/ops/layernorm2d/kernel/layernorm2d_fwd_kernel.hpp
View file @ fb9f0757
......@@ -28,7 +28,10 @@ struct Layernorm2dFwdHostArgs
index_t m;
index_t n;
index_t stride; // row_stride
index_t x_stride; // x row_stride
index_t xr_stride; // x residule row stride
index_t y_stride; // y row stride
index_t yr_stride; // y residule row stride
};
// TODO: Extract some type to wrapper class
......@@ -93,7 +96,10 @@ struct Layernorm2dFwd
index_t m;
index_t n;
index_t stride; // row_stride
index_t x_stride; // x row_stride
index_t xr_stride; // x residule row stride
index_t y_stride; // y row stride
index_t yr_stride; // y residule row stride
};
using Hargs = Layernorm2dFwdHostArgs;
......@@ -112,12 +118,15 @@ struct Layernorm2dFwd
hargs.epsilon,
hargs.m,
hargs.n,
hargs.stride};
hargs.x_stride,
hargs.xr_stride,
hargs.y_stride,
hargs.yr_stride};
}
CK_TILE_HOST static constexpr auto GridSize(const Hargs& hargs)
{
return (hargs.m + Block_M - 1) / Block_M;
return dim3(integer_divide_ceil(hargs.m, Block_M));
}
CK_TILE_HOST static constexpr auto BlockSize() { return Problem::BlockShape::BlockSize; }
......@@ -165,7 +174,7 @@ struct Layernorm2dFwd
return base_str;
}();
return _SS_("layernorm2d_fwd_") + _SS_(prec_str) + "_" +
return _SS_("layernorm2d_fwd_") + _SS_(prec_str) + "_" +
_TS_(S_::Block_M) + "x" + _TS_(S_::Block_N) + "_" + _TS_(S_::WarpPerBlock_M) + "x" + _TS_(S_::WarpPerBlock_N) + "_" +
_TS_(S_::Warp_M) + "x" + _TS_(S_::Warp_N) + "_" + _TS_(S_::Vector_M) + "x" + _TS_(S_::Vector_N) + "_" +
_SS_(Pipeline::name) + surfix;
......@@ -182,7 +191,7 @@ struct Layernorm2dFwd
const auto tmp_ = make_naive_tensor_view<address_space_enum::global>(
static_cast<const XDataType*>(kargs.p_x),
make_tuple(kargs.m, kargs.n),
make_tuple(kargs.stride, 1),
make_tuple(kargs.x_stride, 1),
number<Vector_N>{},
number<1>{});
......@@ -201,7 +210,7 @@ struct Layernorm2dFwd
const auto tmp_ = make_naive_tensor_view<address_space_enum::global>(
static_cast<const XResidualDataType*>(kargs.p_x_residual),
make_tuple(kargs.m, kargs.n),
make_tuple(kargs.stride, 1),
make_tuple(kargs.xr_stride, 1),
number<Vector_N>{},
number<1>{});
......@@ -250,7 +259,7 @@ struct Layernorm2dFwd
auto tmp_ = make_naive_tensor_view<address_space_enum::global>(
static_cast<YDataType*>(kargs.p_y),
make_tuple(kargs.m, kargs.n),
make_tuple(kargs.stride, 1),
make_tuple(kargs.y_stride, 1),
number<Vector_N>{},
number<1>{});
......@@ -266,7 +275,7 @@ struct Layernorm2dFwd
auto tmp_ = make_naive_tensor_view<address_space_enum::global>(
static_cast<YResidualDataType*>(kargs.p_y_residual),
make_tuple(kargs.m, kargs.n),
make_tuple(kargs.stride, 1),
make_tuple(kargs.yr_stride, 1),
number<Vector_N>{},
number<1>{});
......
include/ck_tile/ops/layernorm2d/pipeline/layernorm2d_fwd_pipeline_default_policy.hpp
View file @ fb9f0757
......@@ -26,6 +26,7 @@ struct Layernorm2dFwdPipelineDefaultPolicy
sequence<1, 1, 2, 2>,
sequence<0, 3, 0, 3>>{});
}
template <typename Problem>
CK_TILE_DEVICE static constexpr auto MakeGammaBetaBlockTileDistribution()
{
......@@ -44,9 +45,10 @@ struct Layernorm2dFwdPipelineDefaultPolicy
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetBlockWelford()
{
using P_ = BlockWelfordProblem<typename Problem::XDataType,
using P_ = BlockWelfordProblem<typename Problem::ComputeDataType,
typename Problem::ComputeDataType,
typename Problem::BlockShape>;
typename Problem::BlockShape,
Problem::Traits::kFastFDiv>;
return BlockWelford<P_>{};
}
......@@ -54,9 +56,10 @@ struct Layernorm2dFwdPipelineDefaultPolicy
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetBlockWelfordSync()
{
using P_ = BlockWelfordProblem<typename Problem::XDataType,
using P_ = BlockWelfordProblem<typename Problem::ComputeDataType,
typename Problem::ComputeDataType,
typename Problem::BlockShape>;
typename Problem::BlockShape,
Problem::Traits::kFastFDiv>;
return BlockWelfordSync<P_>{};
}
......@@ -64,9 +67,10 @@ struct Layernorm2dFwdPipelineDefaultPolicy
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetBlockWelfordCrossWarpSync()
{
using P_ = BlockWelfordProblem<typename Problem::XDataType,
using P_ = BlockWelfordProblem<typename Problem::ComputeDataType,
typename Problem::ComputeDataType,
typename Problem::BlockShape>;
typename Problem::BlockShape,
Problem::Traits::kFastFDiv>;
return BlockWelfordCrossWarpSync<P_>{};
}
......@@ -76,13 +80,14 @@ struct Layernorm2dFwdPipelineDefaultPolicy
{
if constexpr(Problem::kNeedCrossWarpSync)
{
using P_ = BlockWelfordProblem<typename Problem::XDataType,
using P_ = BlockWelfordProblem<typename Problem::ComputeDataType,
typename Problem::ComputeDataType,
typename Problem::BlockShape>;
typename Problem::BlockShape,
Problem::Traits::kFastFDiv>;
using block_welford = BlockWelford<P_>;
using x_block_tile =
decltype(make_static_distributed_tensor<typename Problem::XDataType>(
decltype(make_static_distributed_tensor<typename Problem::ComputeDataType>(
MakeXBlockTileDistribution<Problem>()));
using mean_var_block_tile =
decltype(block_welford::template MakeMeanVarBlockTile<x_block_tile>());
......
include/ck_tile/ops/layernorm2d/pipeline/layernorm2d_fwd_pipeline_one_pass.hpp
View file @ fb9f0757
......@@ -36,6 +36,7 @@ struct Layernorm2dFwdPipelineOnePass
static constexpr bool kNeedCrossWarpSync = Problem::kNeedCrossWarpSync;
static constexpr bool kPadM = false; // TODO - BlockLayernorm2dFwdProblem::kPadM
static constexpr bool kPadN = Problem::Traits::kPadN;
static constexpr bool kFastFDiv = Problem::Traits::kFastFDiv;
static constexpr auto kFusedAdd = Problem::Traits::kFusedAdd;
static constexpr auto kFusedQuant = Problem::Traits::kFusedQuant;
......@@ -87,12 +88,9 @@ struct Layernorm2dFwdPipelineOnePass
x_residual_window_, Policy::template MakeXBlockTileDistribution<Problem>());
auto y_residual_window = make_tile_window(
y_residual_window_, Policy::template MakeXBlockTileDistribution<Problem>());
const auto x_scale_window = make_tile_window(
x_scale_window_, Policy::template MakeGammaBetaBlockTileDistribution<Problem>());
auto x = load_tile(x_window);
auto x_resi = load_tile(x_residual_window);
auto x_scale = load_tile(x_scale_window);
auto x = load_tile(x_window);
auto x_resi = load_tile(x_residual_window);
int cur_count = 0;
int max_count =
......@@ -106,20 +104,21 @@ struct Layernorm2dFwdPipelineOnePass
const auto gamma = load_tile(gamma_window);
const auto beta = load_tile(beta_window);
auto acc = cast_tile<ComputeDataType>(x);
if constexpr(kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD_STORE ||
kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD)
{
sweep_tile(x_resi, [&](auto idx) {
// compute x = x_resi + x
x(idx) = type_convert<YResidualDataType>(x_resi(idx)) +
type_convert<YResidualDataType>(x(idx));
acc(idx) = type_convert<ComputeDataType>(x_resi(idx)) + acc(idx);
});
if constexpr(kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD_STORE)
store_tile(y_residual_window, x);
store_tile(y_residual_window, cast_tile<YResidualDataType>(acc));
}
// compute welford each-thread->cross-lane->cross-warp
auto [mean, var] = block_welford(x, cur_count, max_count);
auto [mean, var] = block_welford(acc, cur_count, max_count);
block_welford_sync(mean, var, cur_count);
block_welford_cross_warp_sync(mean, var, cur_count, smem);
block_tile_welford_post_scale_var(var, cur_count);
......@@ -127,7 +126,15 @@ struct Layernorm2dFwdPipelineOnePass
// compute inv-std
auto inv_std = tile_elementwise_in(
[&](const auto& v_) {
return type_convert<ComputeDataType>(1.0f) / (sqrt(v_ + epsilon));
if(kFastFDiv && std::is_same_v<ComputeDataType, float>)
{
return type_convert<ComputeDataType>(1.0f) *
__builtin_amdgcn_rcpf(sqrt(v_ + epsilon));
}
else
{
return type_convert<ComputeDataType>(1.0f) / sqrt(v_ + epsilon);
}
},
var);
......@@ -137,7 +144,7 @@ struct Layernorm2dFwdPipelineOnePass
store_tile(inv_std_window, cast_tile<InvStdDataType>(inv_std));
// layernorm computation
auto ln = make_static_distributed_tensor<ComputeDataType>(x.get_tile_distribution());
auto ln = make_static_distributed_tensor<ComputeDataType>(acc.get_tile_distribution());
sweep_tile(ln, [&, mean_ = mean](auto idx) {
constexpr auto i_idx = make_tuple(idx[number<0>{}]);
constexpr auto j_idx = make_tuple(idx[number<1>{}]);
......@@ -145,26 +152,15 @@ struct Layernorm2dFwdPipelineOnePass
const auto gamma_ = type_convert<ComputeDataType>(gamma[j_idx]);
const auto beta_ = type_convert<ComputeDataType>(beta[j_idx]);
const auto x_ = type_convert<ComputeDataType>(x[idx]);
auto ln_ = (x_ - mean_[i_idx]) * inv_std[i_idx] * gamma_ + beta_;
auto ln_ = (acc[idx] - mean_[i_idx]) * inv_std[i_idx] * gamma_ + beta_;
ln(idx) = ln_;
});
if constexpr(kFusedQuant == Layernorm2dFusedQuantEnum::SMOOTH_DYNAMIC_QUANT)
{
// smooth-quant pre-scale, then run rowwise-quant
sweep_tile(ln, [&](auto idx) {
constexpr auto j_idx = make_tuple(idx[number<1>{}]);
const auto xs_ = type_convert<ComputeDataType>(x_scale[j_idx]);
ln(idx) = ln(idx) * xs_;
});
}
if constexpr(kFusedQuant == Layernorm2dFusedQuantEnum::DYNAMIC_QUANT ||
kFusedQuant == Layernorm2dFusedQuantEnum::SMOOTH_DYNAMIC_QUANT)
{
Epilogue{}(y_window_, y_scale_window, ln, smem);
Epilogue{}(y_window_, x_scale_window_, y_scale_window, ln, smem);
}
else
Epilogue{}(y_window_, ln);
......
include/ck_tile/ops/layernorm2d/pipeline/layernorm2d_fwd_pipeline_problem.hpp
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
......
include/ck_tile/ops/layernorm2d/pipeline/layernorm2d_fwd_pipeline_two_pass.hpp
View file @ fb9f0757
......@@ -106,7 +106,7 @@ struct Layernorm2dFwdPipelineTwoPass
auto block_welford_cross_warp_sync =
Policy::template GetBlockWelfordCrossWarpSync<Problem>();
using XTensorType = decltype(load_tile(x_window));
using XTensorType = decltype(cast_tile<ComputeDataType>(load_tile(x_window)));
auto mean = block_welford.template MakeMeanVarBlockTile<XTensorType>();
auto var = block_welford.template MakeMeanVarBlockTile<XTensorType>();
......@@ -117,21 +117,22 @@ struct Layernorm2dFwdPipelineTwoPass
move_tile_window(x_window, {0, Block_N});
move_tile_window(x_residual_window, {0, Block_N});
auto acc = cast_tile<ComputeDataType>(x);
if constexpr(kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD_STORE ||
kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD)
{
sweep_tile(x_resi, [&](auto idx) {
// compute x = x_resi + x
x(idx) = type_convert<YResidualDataType>(x_resi(idx)) +
type_convert<YResidualDataType>(x(idx));
acc(idx) = type_convert<ComputeDataType>(x_resi(idx)) + acc(idx);
});
if constexpr(kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD_STORE)
{
store_tile(y_residual_window, x);
store_tile(y_residual_window, cast_tile<YResidualDataType>(acc));
move_tile_window(y_residual_window, {0, Block_N});
}
}
block_welford(x, mean, var, cur_count, max_count);
block_welford(acc, mean, var, cur_count, max_count);
}
block_welford_sync(mean, var, cur_count);
......@@ -165,20 +166,21 @@ struct Layernorm2dFwdPipelineTwoPass
{
auto x = load_tile(x_window);
auto x_resi = load_tile(x_residual_window);
auto acc = cast_tile<ComputeDataType>(x);
if constexpr(kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD_STORE ||
kFusedAdd == Layernorm2dFusedAddEnum::PRE_ADD)
{
sweep_tile(x_resi, [&](auto idx) {
// compute x = x_resi + x
x(idx) = type_convert<YResidualDataType>(x_resi(idx)) +
type_convert<YResidualDataType>(x(idx));
acc(idx) = type_convert<ComputeDataType>(x_resi(idx)) + acc(idx);
});
}
// load gamma/beta (TODO: support no gamma/beta?)
const auto gamma = load_tile(gamma_window);
const auto beta = load_tile(beta_window);
auto ln = make_static_distributed_tensor<ComputeDataType>(x.get_tile_distribution());
auto ln = make_static_distributed_tensor<ComputeDataType>(acc.get_tile_distribution());
sweep_tile(ln, [&, mean_ = mean](auto idx) {
constexpr auto i_idx = make_tuple(idx[number<0>{}]);
......@@ -187,8 +189,7 @@ struct Layernorm2dFwdPipelineTwoPass
const auto gamma_ = type_convert<ComputeDataType>(gamma[j_idx]);
const auto beta_ = type_convert<ComputeDataType>(beta[j_idx]);
const auto x_ = type_convert<ComputeDataType>(x[idx]);
auto ln_ = (x_ - mean_[i_idx]) * inv_std[i_idx] * gamma_ + beta_;
auto ln_ = (acc(idx) - mean_[i_idx]) * inv_std[i_idx] * gamma_ + beta_;
ln(idx) = ln_;
});
......
include/ck_tile/ops/layernorm2d/pipeline/layernorm2d_fwd_traits.hpp
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
......@@ -39,6 +39,7 @@ template<> struct Layernorm2dFusedQuantEnumName<Layernorm2dFusedQuantEnum::SMOOT
template <bool kPadN_,
bool kSaveMeanInvStd_,
bool kFastFDiv_,
bool kTwoPass_,
Layernorm2dFusedAddEnum kFusedAdd_,
Layernorm2dFusedQuantEnum kFusedQuant_>
......@@ -46,6 +47,7 @@ struct Layernorm2dFwdTraits
{
static constexpr bool kPadN = kPadN_;
static constexpr bool kSaveMeanInvStd = kSaveMeanInvStd_;
static constexpr bool kFastFDiv = kFastFDiv_;
static constexpr bool kTwoPass = kTwoPass_;
static constexpr Layernorm2dFusedAddEnum kFusedAdd = kFusedAdd_;
static constexpr Layernorm2dFusedQuantEnum kFusedQuant = kFusedQuant_;
......
include/ck_tile/ops/moe_sorting.hpp 0 → 100644
View file @ fb9f0757
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/ops/fused_moe/kernel/moe_sorting_kernel.hpp"
#include "ck_tile/ops/fused_moe/pipeline/moe_sorting_pipeline.hpp"
#include "ck_tile/ops/fused_moe/pipeline/moe_sorting_policy.hpp"
#include "ck_tile/ops/fused_moe/pipeline/moe_sorting_problem.hpp"
#include "ck_tile/ops/common/generic_2d_block_shape.hpp"
#include "ck_tile/ops/common/tensor_layout.hpp"
include/ck_tile/ops/reduce/block/block_reduce2d.hpp
View file @ fb9f0757
......@@ -29,7 +29,8 @@ struct BlockReduce2d
sweep_tile<XDistributedTensor_>(
[&](auto... idx_) {
constexpr auto idx_0 = make_tuple(make_tuple(idx_[number<0>{}]...)[number<0>{}]);
y_tensor(idx_0) = reduce_func(y_tensor(idx_0), x_tensor[idx_]...);
y_tensor(idx_0) = reduce_func(
y_tensor(idx_0), ck_tile::type_convert<ComputeDataType>(x_tensor[idx_])...);
},
ReducePacksPerXDim{});
#if 0
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment