Add hstu attention kernel implementation, instances and interfaces (building succeeded)

This commit is contained in:
Qianfeng Zhang
2025-04-03 08:20:54 +00:00
parent 83f29243df
commit 121a950df5
122 changed files with 5904 additions and 126 deletions

View File

@@ -2,10 +2,11 @@ set(EXAMPLE_HSTU_ATTENTION "tile_example_hstu_attention")
# not using add_example_executable() to add this target, since we don't want this to have
# to be included in "make all/install/check"
message("adding example ${EXAMPLE_HSTU_ATTENTION}")
##file(GLOB INSTANCE_SRCS instances/*.cpp)
file(GLOB INSTANCE_SRCS instances/*.cpp)
set(INTERFACES_SRCS hstu_attention_jagged_forward_bf16.cpp hstu_attention_jagged_forward_fp16.cpp hstu_attention_batched_forward_bf16.cpp hstu_attention_batched_forward_fp16.cpp)
add_executable(${EXAMPLE_HSTU_ATTENTION} EXCLUDE_FROM_ALL example_hstu_attention.cpp)
target_include_directories(${EXAMPLE_HSTU_ATTENTION} PRIVATE ${CMAKE_CURRENT_LIST_DIR})
##target_sources(${EXAMPLE_HSTU_ATTENTION} PRIVATE hstu_attention_bf16.cpp hstu_attention_fp16.cpp ${INSTANCE_SRCS})
target_sources(${EXAMPLE_HSTU_ATTENTION} PRIVATE ${INTERFACES_SRCS} ${INSTANCE_SRCS})
set(EXAMPLE_HSTU_ATTENTION_COMPILE_OPTIONS)

View File

@@ -22,10 +22,16 @@
#include <ck_tile/host/check_err.hpp>
#include <ck_tile/host/timer.hpp>
#include "hstu_attention_setting.hpp"
#include "bool_switch.hpp"
#include "hstu_attention_fwd_type_config.hpp"
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_params.hpp"
#include "reference_hstu_attention.hpp"
extern void hstu_attention_batched_forward_fp16(HstuAttentionFwdParams& param, hipStream_t stream);
extern void hstu_attention_batched_forward_bf16(HstuAttentionFwdParams& param, hipStream_t stream);
extern void hstu_attention_jagged_forward_fp16(HstuAttentionFwdParams& param, hipStream_t stream);
extern void hstu_attention_jagged_forward_bf16(HstuAttentionFwdParams& param, hipStream_t stream);
template <typename T>
std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
{
@@ -120,25 +126,21 @@ bool run(const ck_tile::ArgParser& arg_parser)
bool do_validation = static_cast<bool>(arg_parser.get_int("v"));
bool is_jagged = static_cast<bool>(arg_parser.get_int("jagged"));
int num_batch = arg_parser.get_int("b");
int nhead = arg_parser.get_int("nhead");
int num_head = arg_parser.get_int("nhead");
int hdim_qk = arg_parser.get_int("hdim_qk");
int hdim_v = arg_parser.get_int("hdim_v");
bool use_causal = static_cast<bool>(arg_parser.get_int("causal"));
int max_attn_len = arg_parser.get_int("local_len");
int window_size = arg_parser.get_int("local_len");
bool use_local = (max_attn_len > 0);
bool use_local = (window_size > 0);
int contextual_seq_len = arg_parser.get_int("context_len");
int min_full_seq_len = arg_parser.get_int("minfull_len");
int seed = arg_parser.get_int("seed");
int contextual_seqlen = arg_parser.get_int("context_len");
int min_full_attn_seqlen = arg_parser.get_int("minfull_len");
int seed = arg_parser.get_int("seed");
bool measure_perf = static_cast<bool>(arg_parser.get_int("perf"));
(void)do_validation;
(void)measure_perf;
std::string str_of_targets = arg_parser.get_str("targets");
std::vector<int> num_targets = get_integers_from_string(str_of_targets);
@@ -147,7 +149,8 @@ bool run(const ck_tile::ArgParser& arg_parser)
std::vector<int> seq_offsets;
int seqlen = 0; // means total seq lengths for jagged
int seqlen = 0; // means total seq lengths for jagged
int max_seqlen = 0;
if(is_jagged)
{
@@ -156,6 +159,7 @@ bool run(const ck_tile::ArgParser& arg_parser)
seq_offsets.push_back(0);
for(size_t i = 0; i < seq_lengths.size(); i++)
{
max_seqlen = max(max_seqlen, seq_lengths[i]);
seqlen += seq_lengths[i];
seq_offsets.push_back(seqlen);
};
@@ -166,16 +170,16 @@ bool run(const ck_tile::ArgParser& arg_parser)
for(size_t i = 0; i < seq_lengths.size(); i++)
{
assert(seq_lengths[i] - num_targets[i] >= min_full_seq_len);
assert(seq_lengths[i] - num_targets[i] >= contextual_seq_len);
assert(seq_lengths[i] - num_targets[i] >= min_full_attn_seqlen);
assert(seq_lengths[i] - num_targets[i] >= contextual_seqlen);
};
}
else
{
for(size_t i = 0; i < seq_lengths.size(); i++)
{
assert(seq_lengths[i] >= min_full_seq_len);
assert(seq_lengths[i] >= contextual_seq_len);
assert(seq_lengths[i] >= min_full_attn_seqlen);
assert(seq_lengths[i] >= contextual_seqlen);
};
};
}
@@ -188,53 +192,212 @@ bool run(const ck_tile::ArgParser& arg_parser)
{
assert(1 == num_targets.size());
assert(seqlen - num_targets[0] >= min_full_seq_len);
assert(seqlen - num_targets[0] >= contextual_seq_len);
assert(seqlen - num_targets[0] >= min_full_attn_seqlen);
assert(seqlen - num_targets[0] >= contextual_seqlen);
}
else
{
assert(seqlen >= min_full_seq_len);
assert(seqlen >= contextual_seq_len);
assert(seqlen >= min_full_attn_seqlen);
assert(seqlen >= contextual_seqlen);
};
};
int batches_for_alloc = is_jagged ? 1 : num_batch;
ck_tile::HostTensor<InOutDataType> q_host(
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, nhead, hdim_qk});
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, num_head, hdim_qk});
ck_tile::HostTensor<InOutDataType> k_host(
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, nhead, hdim_qk});
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, num_head, hdim_qk});
ck_tile::HostTensor<InOutDataType> v_host(
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, nhead, hdim_v});
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, num_head, hdim_v});
ck_tile::HostTensor<InOutDataType> o_host_ref(
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, nhead, hdim_v});
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, num_head, hdim_v});
ck_tile::FillNormalDistributionIntegerValue<InOutDataType>{-2.f, 2.f, seed}(q_host);
ck_tile::FillNormalDistributionIntegerValue<InOutDataType>{-2.f, 2.f, seed}(k_host);
ck_tile::FillNormalDistributionIntegerValue<InOutDataType>{-2.f, 2.f, seed}(v_host);
using GemmAccDataType = typename HSTUAttentionTypeConfig<InOutDataType>::GemmAccDataType;
using SMComputeDataType = typename HSTUAttentionTypeConfig<InOutDataType>::SMComputeDataType;
ck_tile::DeviceMem q_dev(q_host.get_element_space_size_in_bytes());
ck_tile::DeviceMem k_dev(k_host.get_element_space_size_in_bytes());
ck_tile::DeviceMem v_dev(v_host.get_element_space_size_in_bytes());
ck_tile::DeviceMem o_dev(o_host_ref.get_element_space_size_in_bytes());
BOOL_SWITCH_3(is_jagged, kIsJagged, use_causal, kUseCausal, use_local, kUseLocal, [&] {
ck_tile::reference_hstu_attention<InOutDataType,
GemmAccDataType,
SMComputeDataType,
kIsJagged,
kUseCausal,
kUseLocal>::Run(q_host,
k_host,
v_host,
o_host_ref,
num_batch,
1.0f,
seq_offsets,
num_targets,
max_attn_len,
contextual_seq_len,
min_full_seq_len);
});
return 0;
ck_tile::DeviceMem seq_offsets_dev(seq_offsets.size() * sizeof(int));
ck_tile::DeviceMem num_targets_dev(num_targets.size() * sizeof(int));
q_dev.ToDevice(q_host.data());
k_dev.ToDevice(k_host.data());
v_dev.ToDevice(v_host.data());
if(is_jagged)
seq_offsets_dev.ToDevice(seq_offsets.data());
if(!num_targets.empty())
num_targets_dev.ToDevice(num_targets.data());
HstuAttentionFwdParams params;
if(is_jagged)
{
params.is_jagged = true;
params.num_batch = num_batch;
params.seq_offsets_ptr = seq_offsets_dev.GetDeviceBuffer();
params.max_seqlen = max_seqlen;
params.q_ptr = q_dev.GetDeviceBuffer();
params.k_ptr = k_dev.GetDeviceBuffer();
params.v_ptr = v_dev.GetDeviceBuffer();
params.bias_ptr = nullptr;
params.o_ptr = o_dev.GetDeviceBuffer();
params.hdim_qk = hdim_qk;
params.hdim_v = hdim_v;
params.num_head = num_head;
params.scale_s = 1.0f / std::sqrt(params.hdim_qk);
params.seq_stride_q = q_host.get_strides()[1];
params.seq_stride_k = k_host.get_strides()[1];
params.seq_stride_v = v_host.get_strides()[1];
params.seq_stride_bias = 0;
params.seq_stride_o = o_host_ref.get_strides()[1];
params.nhead_stride_q = q_host.get_strides()[2];
params.nhead_stride_k = k_host.get_strides()[2];
params.nhead_stride_v = v_host.get_strides()[2];
params.nhead_stride_bias = 0;
params.nhead_stride_o = o_host_ref.get_strides()[2];
params.num_targets_ptr = num_targets.empty() ? nullptr : num_targets_dev.GetDeviceBuffer();
params.use_causal = use_causal;
params.window_size = window_size;
params.contextual_seqlen = contextual_seqlen;
params.min_full_attn_seqlen = min_full_attn_seqlen;
params.p_drop = 0.0f; // dropout is not supported at present
params.philox_seed = 0UL;
params.philox_offset = 0UL;
}
else
{
params.is_jagged = false;
params.num_batch = num_batch;
params.seqlen = seqlen;
params.q_ptr = q_dev.GetDeviceBuffer();
params.k_ptr = k_dev.GetDeviceBuffer();
params.v_ptr = v_dev.GetDeviceBuffer();
params.bias_ptr = nullptr;
params.o_ptr = o_dev.GetDeviceBuffer();
params.hdim_qk = hdim_qk;
params.hdim_v = hdim_v;
params.num_head = num_head;
params.scale_s = 1.0f / std::sqrt(params.hdim_qk);
params.seq_stride_q = q_host.get_strides()[1];
params.seq_stride_k = k_host.get_strides()[1];
params.seq_stride_v = v_host.get_strides()[1];
params.seq_stride_bias = 0;
params.seq_stride_o = o_host_ref.get_strides()[1];
params.nhead_stride_q = q_host.get_strides()[2];
params.nhead_stride_k = k_host.get_strides()[2];
params.nhead_stride_v = v_host.get_strides()[2];
params.nhead_stride_bias = 0;
params.nhead_stride_o = o_host_ref.get_strides()[2];
params.batch_stride_q = q_host.get_strides()[0];
params.batch_stride_k = k_host.get_strides()[0];
params.batch_stride_v = v_host.get_strides()[0];
params.batch_stride_bias = 0;
params.batch_stride_o = o_host_ref.get_strides()[0];
params.num_targets_ptr = num_targets.empty() ? nullptr : num_targets_dev.GetDeviceBuffer();
params.use_causal = use_causal;
params.window_size = window_size;
params.contextual_seqlen = contextual_seqlen;
params.min_full_attn_seqlen = min_full_attn_seqlen;
params.p_drop = 0.0f; // dropout is not supported at present
params.philox_seed = 0UL;
params.philox_offset = 0UL;
};
hipStream_t stream;
HIP_CHECK_ERROR(hipStreamCreate(&stream));
if constexpr(std::is_same<InOutDataType, ck_tile::fp16_t>::value)
{
if(is_jagged)
hstu_attention_jagged_forward_fp16(params, stream);
else
hstu_attention_batched_forward_fp16(params, stream);
}
else if constexpr(std::is_same<InOutDataType, ck_tile::bf16_t>::value)
{
if(is_jagged)
hstu_attention_jagged_forward_bf16(params, stream);
else
hstu_attention_batched_forward_bf16(params, stream);
}
else
throw std::runtime_error("Other data type is not supported at present!");
bool res = true;
if(do_validation)
{
using GemmAccDataType = typename HstuAttentionFwdTypeConfig<InOutDataType>::GemmAccDataType;
using CompDataType = typename HstuAttentionFwdTypeConfig<InOutDataType>::CompDataType;
BOOL_SWITCH_3(is_jagged, kIsJagged, use_causal, kUseCausal, use_local, kUseLocal, [&] {
ck_tile::reference_hstu_attention<InOutDataType,
GemmAccDataType,
CompDataType,
kIsJagged,
kUseCausal,
kUseLocal>::Run(q_host,
k_host,
v_host,
o_host_ref,
num_batch,
1.0f,
seq_offsets,
num_targets,
window_size,
contextual_seqlen,
min_full_attn_seqlen);
});
ck_tile::HostTensor<InOutDataType> o_host(
std::array<ck_tile::index_t, 4>{batches_for_alloc, seqlen, num_head, hdim_v});
o_dev.FromDevice(o_host.data());
auto [rtol, atol] = get_elimit<InOutDataType>();
res = ck_tile::check_err(
o_host, o_host_ref, std::string("hstu_attention output error"), atol, rtol);
};
if(measure_perf)
{
ck_tile::gpu_timer timer{};
timer.start(stream);
for(int i = 0; i < 20; i++)
{
if constexpr(std::is_same<InOutDataType, ck_tile::fp16_t>::value)
{
if(is_jagged)
hstu_attention_jagged_forward_fp16(params, stream);
else
hstu_attention_batched_forward_fp16(params, stream);
}
else if constexpr(std::is_same<InOutDataType, ck_tile::bf16_t>::value)
{
if(is_jagged)
hstu_attention_jagged_forward_bf16(params, stream);
else
hstu_attention_batched_forward_bf16(params, stream);
}
}
timer.stop(stream);
auto ms = timer.duration() / 20.f;
std::cout << "Average execution time of the gather_attention operator is " << ms
<< " milli-seconds" << std::endl;
}
return res;
}
int main(int argc, char* argv[])

View File

@@ -0,0 +1,177 @@
# noqa: C801
# Copyright (c) 2023-2024, Advanced Micro Devices, Inc. All rights reserved.
#
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.
#
import os
from pathlib import Path
from typing import List
HSTU_COPYRIGHT_HEADER = """
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `{file}`
*/
""".format(
file=os.path.relpath(os.path.realpath(__file__), start=Path(__file__).parents[4])
)
HSTU_FORWARD_INSTANCE_TEMPLATE_INC = """
#include <ck_tile/core/numeric/{dtype_file}.hpp>
#include \"hstu_attention_{mode}_forward_dispatch.hpp\"
"""
HSTU_FORWARD_INSTANCE_TEMPLATE = """
{extern}template void run_{mode}_forward_causal_local_bias_dropout_dispatch<
{dtype},
{has_causal},
{has_local},
{has_bias},
{has_dropout},
{max_k}>(HstuAttentionFwdParams& param, hipStream_t stream);
"""
HSTU_FORWARD_INSTANCE_FNAME = (
"hstu_attention_{mode}_forward_{dtype_str}_{has_or_no_causal_str}_{has_or_no_local_str}_"
"{has_or_no_bias_str}_{has_or_no_dropout_str}_{max_k_str}.cpp"
)
HSTU_INSTANCE_REF_FNAME = "hstu_attention_{mode}_{function}_{dtype}_instances_ref.hpp"
BOOL_MAP = {True: "true", False: "false"}
BOOL_MAP_CAUSAL = {
True: "has_causal",
False: "no_causal",
}
BOOL_MAP_LOCAL = {
True: "has_local",
False: "no_local",
}
BOOL_MAP_BIAS = {
True: "has_bias",
False: "no_bias",
}
BOOL_MAP_DROPOUT = {
True: "has_dropout",
False: "no_dropout",
}
INT_MAP_MAX_K = {hd: f"maxk_{hd}" for hd in [64, 128, 256]}
TYPE_CTYPE_MAP = {
"fp16": "ck_tile::fp16_t",
"bf16": "ck_tile::bf16_t",
}
TYPE_FNAME_MAP = {
"fp16": "bfloat16",
"bf16": "half",
}
MODE_NAME_MAP = {
"batched": "Batched",
"jagged": "Jagged",
}
def create_forward_instances(instance_dir: Path, headdims: List) -> None:
for mode in ["batched", "jagged"]:
for dtype in ["fp16", "bf16"]:
for has_causal, has_local in zip([True, False],[True, False]):
for has_bias in [True, False]:
for has_dropout in [True, False]:
for max_k in headdims:
fname = HSTU_FORWARD_INSTANCE_FNAME.format(
mode=mode,
dtype_str=dtype,
has_or_no_causal_str=BOOL_MAP_CAUSAL[has_causal],
has_or_no_local_str=BOOL_MAP_CAUSAL[has_local],
has_or_no_bias_str=BOOL_MAP_BIAS[has_bias],
has_or_no_dropout_str=BOOL_MAP_DROPOUT[has_dropout],
max_k_str=INT_MAP_MAX_K[max_k],
)
forward_instance_inc = (
HSTU_FORWARD_INSTANCE_TEMPLATE_INC.format(
mode=mode,
dtype_file=TYPE_FNAME_MAP[dtype],
)
)
forward_instance = HSTU_FORWARD_INSTANCE_TEMPLATE.format(
extern="",
mode=mode,
dtype=TYPE_CTYPE_MAP[dtype],
has_causal=BOOL_MAP[has_causal],
has_local=BOOL_MAP[has_causal],
has_bias=BOOL_MAP[has_bias],
has_dropout=BOOL_MAP[has_dropout],
max_k=max_k,
cap_mode=MODE_NAME_MAP[mode],
)
(instance_dir / fname).write_text(
HSTU_COPYRIGHT_HEADER
+ forward_instance_inc
+ forward_instance
)
def create_forward_instances_ref(instance_dir: Path, headdims: List) -> None:
for mode in ["batched", "jagged"]:
for dtype in ["fp16", "bf16"]:
ref_fname = HSTU_INSTANCE_REF_FNAME.format(
mode=mode,
function="forward",
dtype=dtype,
)
ref_fname_path = instance_dir / ref_fname
forward_instance_inc = HSTU_FORWARD_INSTANCE_TEMPLATE_INC.format(
mode=mode,
dtype_file=TYPE_FNAME_MAP[dtype],
)
with open(ref_fname_path, "a") as file:
file.write(HSTU_COPYRIGHT_HEADER)
file.write(forward_instance_inc)
for max_k in headdims:
for has_bias in [True, False]:
for has_dropout in [True, False]:
for has_causal, has_local in zip([True, False],[True, False]):
forward_instance = (
HSTU_FORWARD_INSTANCE_TEMPLATE.format(
extern="extern ",
mode=mode,
dtype=TYPE_CTYPE_MAP[dtype],
has_causal=BOOL_MAP[has_causal],
has_local=BOOL_MAP[has_local],
has_bias=BOOL_MAP[has_bias],
has_dropout=BOOL_MAP[has_dropout],
max_k=max_k,
cap_mode=MODE_NAME_MAP[mode],
)
)
file.write(forward_instance)
if __name__ == "__main__":
headdims_fwd = [64, 128, 256]
this_dir = os.path.dirname(__file__)
output_dir = Path(this_dir) / "instances"
output_dir.mkdir(parents=True, exist_ok=True)
# remove existing files in the directory
files = os.listdir(output_dir)
for ff in files:
file_path = os.path.join(output_dir, ff)
os.remove(file_path)
create_forward_instances(output_dir, headdims_fwd)
create_forward_instances_ref(output_dir, headdims_fwd)

View File

@@ -0,0 +1,43 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <ck_tile/core.hpp>
#include <stdexcept>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_attention_batched_forward_dispatch.hpp"
#include "instances/hstu_attention_batched_forward_bf16_instances_ref.hpp"
void hstu_attention_batched_forward_bf16(HstuAttentionFwdParams& param, hipStream_t stream)
{
const bool has_dropout = (param.p_drop > 0.0f);
const bool has_bias = (param.bias_ptr != nullptr);
const bool use_causal = param.use_causal;
BOOL_SWITCH_3(has_bias, kHasBias, has_dropout, kHasDropout, use_causal, kUseCausal, [&] {
HDIM_SWITCH(param.hdim_qk, param.hdim_v, MaxK, [&] {
if(param.window_size > 0)
{
run_batched_forward_causal_local_bias_dropout_dispatch<ck_tile::bf16_t,
kUseCausal,
true,
kHasBias,
kHasDropout,
MaxK>(param, stream);
}
else
{
run_batched_forward_causal_local_bias_dropout_dispatch<ck_tile::bf16_t,
kUseCausal,
false,
kHasBias,
kHasDropout,
MaxK>(param, stream);
};
});
});
};

View File

@@ -0,0 +1,155 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#pragma once
#include <ck_tile/core/numeric/integer.hpp>
#include <ck_tile/host/kernel_launch.hpp>
#include <ck_tile/host/stream_config.hpp>
#include <ck_tile/ops/epilogue.hpp>
#include <ck_tile/ops/fmha.hpp>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_fwd_type_config.hpp"
#include "hstu_attention_fwd_setting.hpp"
#include "hstu_attention_params.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_block_masking.hpp"
#include "hstu_attention_pipeline_problem.hpp"
#include "hstu_attention_traits.hpp"
#include "hstu_attention_fwd_pipeline.hpp"
#include "hstu_attention_fwd_kernel.hpp"
template <typename InOutDataType,
bool kUseCausal,
bool kUseLocal,
bool kHasBias,
bool kHasDropout,
ck_tile::index_t MaxK>
struct batched_forward_causal_local_bias_dropout_dispatch
{
using HstuAttentionShape = typename HstuAttentionFwdShape<MaxK>::Type;
using HstuMask = ck_tile::HstuBlockMasking<kUseCausal, kUseLocal>;
template <typename HstuTraits>
using HstuPipelineProblemTemp = ck_tile::HstuAttentionFwdPipelineProblem<
InOutDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::GemmAccDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::CompDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::BiasDataType,
false, // kIsJagged
kHasBias,
kHasDropout,
HstuMask,
HstuAttentionShape,
HstuTraits>;
static void Run(HstuAttentionFwdParams& param, hipStream_t stream)
{
constexpr ck_tile::index_t occupancy = -1;
const bool pad_seqlen_k = !(param.seqlen % HstuAttentionShape::kN0 == 0);
const bool pad_headdim_qk = !(param.hdim_qk % HstuAttentionShape::kSubQKHeaddim == 0);
const bool pad_headdim_v = !(param.hdim_v % HstuAttentionShape::kN1 == 0);
// no need to check seqlen_q since it is not used as fastest dim,
// buffer_load_dwordxx/buffer_store_dwordxx can handle oob access
constexpr bool kPadSeqLenQ = false;
BOOL_SWITCH_3(
pad_seqlen_k,
kPadSeqLenK,
pad_headdim_qk,
kPadHeadDimQK,
pad_headdim_v,
kPadHeadDimV,
[&] {
using HstuTraits = ck_tile::HstuAttentionFwdTraits<kPadSeqLenQ,
kPadSeqLenK,
kPadHeadDimQK,
kPadHeadDimV,
occupancy>;
using HstuPipelineProblem = HstuPipelineProblemTemp<HstuTraits>;
using HstuEpilogue = ck_tile::Default2DEpilogue<ck_tile::Default2DEpilogueProblem<
typename HstuAttentionFwdTypeConfig<InOutDataType>::OaccDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::ODataType,
kPadSeqLenQ,
kPadHeadDimV>>;
using HstuPipeline = ck_tile::HstuAttentionFwdPipelineQRKSVS<HstuPipelineProblem>;
using HstuKernel = ck_tile::HstuAttentionFwdKernel<HstuPipeline, HstuEpilogue>;
RunWithKernel<HstuKernel>(param, stream);
});
};
template <typename HstuKernel>
static void RunWithKernel(HstuAttentionFwdParams& param, hipStream_t stream)
{
const auto kargs = [&] {
return HstuKernel::MakeKargs(param.q_ptr,
param.k_ptr,
param.v_ptr,
param.bias_ptr,
param.o_ptr,
param.seqlen,
param.hdim_qk,
param.hdim_v,
param.num_head,
param.scale_s,
param.seq_stride_q,
param.seq_stride_k,
param.seq_stride_v,
param.seq_stride_bias,
param.seq_stride_o,
param.nhead_stride_q,
param.nhead_stride_k,
param.nhead_stride_v,
param.nhead_stride_bias,
param.nhead_stride_o,
param.batch_stride_q,
param.batch_stride_k,
param.batch_stride_v,
param.batch_stride_bias,
param.batch_stride_o,
param.num_targets_ptr,
param.window_size,
param.contextual_seqlen,
param.min_full_attn_seqlen,
param.p_drop,
param.philox_seed,
param.philox_offset);
}();
dim3 kGridSize =
HstuKernel::GridSize(param.num_batch, param.num_head, param.seqlen, param.hdim_v);
constexpr dim3 kBlockSize = HstuKernel::BlockSize();
constexpr ck_tile::index_t kBlockPerCu = HstuKernel::kBlockPerCu;
(void)ck_tile::launch_kernel(ck_tile::stream_config{stream, false},
ck_tile::make_kernel<kBlockSize.x, kBlockPerCu>(
HstuKernel{}, kGridSize, kBlockSize, 0, kargs));
};
};
template <typename InOutDataType,
bool kUseCausal,
bool kUseLocal,
bool kHasBias,
bool kHasDropout,
ck_tile::index_t MaxK>
void run_batched_forward_causal_local_bias_dropout_dispatch(HstuAttentionFwdParams& param,
hipStream_t stream)
{
batched_forward_causal_local_bias_dropout_dispatch<InOutDataType,
kUseCausal,
kUseLocal,
kHasBias,
kHasDropout,
MaxK>::Run(param, stream);
};

View File

@@ -0,0 +1,43 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <ck_tile/core.hpp>
#include <stdexcept>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_attention_batched_forward_dispatch.hpp"
#include "instances/hstu_attention_batched_forward_fp16_instances_ref.hpp"
void hstu_attention_batched_forward_fp16(HstuAttentionFwdParams& param, hipStream_t stream)
{
const bool has_dropout = (param.p_drop > 0.0f);
const bool has_bias = (param.bias_ptr != nullptr);
const bool use_causal = param.use_causal;
BOOL_SWITCH_3(has_bias, kHasBias, has_dropout, kHasDropout, use_causal, kUseCausal, [&] {
HDIM_SWITCH(param.hdim_qk, param.hdim_v, MaxK, [&] {
if(param.window_size > 0)
{
run_batched_forward_causal_local_bias_dropout_dispatch<ck_tile::fp16_t,
kUseCausal,
true,
kHasBias,
kHasDropout,
MaxK>(param, stream);
}
else
{
run_batched_forward_causal_local_bias_dropout_dispatch<ck_tile::fp16_t,
kUseCausal,
false,
kHasBias,
kHasDropout,
MaxK>(param, stream);
};
});
});
};

View File

@@ -0,0 +1,763 @@
// 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 <string>
#include <type_traits>
#include <utility>
#include <variant>
// S[seqlen_q, seqlen_k] = Q[seqlen_q, hdim_q] @ K[seqlen_k, hdim_q]
// S'[seqlen_q, seqlen_k] = S[seqlen_q, seqlen_k] * Scale[1]
// S''[seqlen_q, seqlen_k] = S'[seqlen_q, seqlen_k] + Bias[seqlen_q, seqlen_k]
// P[seqlen_q, seqlen_k] = Softmax(S''[seqlen_q, seqlen_k])
// O[seqlen_q, hdim_v] = P[seqlen_q, seqlen_k] @ V^T[hdim_v, seqlen_k]
namespace ck_tile {
template <typename HstuAttentionPipeline_, typename EpiloguePipeline_>
struct HstuAttentionFwdKernel
{
using HstuAttentionPipeline = ck_tile::remove_cvref_t<HstuAttentionPipeline_>;
using EpiloguePipeline = ck_tile::remove_cvref_t<EpiloguePipeline_>;
static constexpr ck_tile::index_t kBlockSize = HstuAttentionPipeline::kBlockSize;
static constexpr ck_tile::index_t kBlockPerCu = HstuAttentionPipeline::kBlockPerCu;
static_assert(kBlockPerCu > 0);
static constexpr ck_tile::index_t kBlockPerCuInput =
HstuAttentionPipeline::Problem::kBlockPerCu;
using QKVDataType = ck_tile::remove_cvref_t<typename HstuAttentionPipeline::QKVDataType>;
using BiasDataType = ck_tile::remove_cvref_t<typename HstuAttentionPipeline::BiasDataType>;
using ODataType = ck_tile::remove_cvref_t<typename HstuAttentionPipeline::ODataType>;
using VLayout = ck_tile::remove_cvref_t<typename HstuAttentionPipeline::VLayout>;
static constexpr bool kIsJagged = HstuAttentionPipeline::kIsJagged;
static constexpr bool kPadSeqLenQ = HstuAttentionPipeline::kPadSeqLenQ;
static constexpr bool kPadSeqLenK = HstuAttentionPipeline::kPadSeqLenK;
static constexpr bool kPadHeadDimQK = HstuAttentionPipeline::kPadHeadDimQK;
static constexpr bool kPadHeadDimV = HstuAttentionPipeline::kPadHeadDimV;
static constexpr auto kHasBias = HstuAttentionPipeline::kHasBias;
static constexpr bool kHasDropout = HstuAttentionPipeline::kHasDropout;
using HstuMask = ck_tile::remove_cvref_t<typename HstuAttentionPipeline::HstuMask>;
static constexpr bool kHasMask = HstuMask::IsMasking;
template <ck_tile::index_t I> // to avoid duplicated base class problem, introduce an template
// arg
struct HstuAttentionFwdEmptyKargs
{
};
// kargs use aggregate initializer, so no constructor will provided
// use inheritance to minimize karg size
// user need to use MakeKargs() function to create kargs.
struct HstuAttentionFwdCommonKargs
{
const void* q_ptr;
const void* k_ptr;
const void* v_ptr;
void* o_ptr;
ck_tile::index_t seqlen;
ck_tile::index_t hdim_qk;
ck_tile::index_t hdim_v;
ck_tile::index_t num_head;
float scale_s;
ck_tile::index_t seq_stride_q;
ck_tile::index_t seq_stride_k;
ck_tile::index_t seq_stride_v;
ck_tile::index_t seq_stride_o;
ck_tile::index_t nhead_stride_q;
ck_tile::index_t nhead_stride_k;
ck_tile::index_t nhead_stride_v;
ck_tile::index_t nhead_stride_o;
const int32_t* num_targets_ptr;
};
struct HstuAttentionFwdCommonBiasKargs
{
const void* bias_ptr = nullptr;
ck_tile::index_t seq_stride_bias = 0;
ck_tile::index_t nhead_stride_bias = 0;
};
struct HstuAttentionFwdBatchModeBiasKargs : HstuAttentionFwdCommonBiasKargs
{
ck_tile::index_t batch_stride_bias = 0;
};
struct HstuAttentionFwdMaskKargs
{
ck_tile::index_t window_size;
ck_tile::index_t contextual_seqlen;
ck_tile::index_t min_full_attn_seqlen;
};
struct HstuAttentionFwdDropoutSeedOffset
{
uint64_t drop_seed;
uint64_t drop_offset;
};
struct HstuAttentionFwdCommonDropoutKargs : HstuAttentionFwdDropoutSeedOffset
{
void init_dropout(float p_drop, uint64_t seed, uint64_t offset)
{
float p_undrop = 1.0 - p_drop;
p_undrop_in_uint8_t =
uint8_t(std::floor(p_undrop * std::numeric_limits<uint8_t>::max()));
rp_undrop = 1.0 / p_undrop;
this->drop_seed = seed;
this->drop_offset = offset;
}
float rp_undrop = 1;
uint8_t p_undrop_in_uint8_t = std::numeric_limits<uint8_t>::max();
};
struct HstuAttentionFwdBatchModeKargs
: HstuAttentionFwdCommonKargs,
std::conditional_t<kHasBias,
HstuAttentionFwdBatchModeBiasKargs,
HstuAttentionFwdEmptyKargs<0>>,
std::conditional_t<kHasMask, HstuAttentionFwdMaskKargs, HstuAttentionFwdEmptyKargs<1>>,
std::conditional_t<kHasDropout,
HstuAttentionFwdCommonDropoutKargs,
HstuAttentionFwdEmptyKargs<2>>
{
ck_tile::index_t batch_stride_q;
ck_tile::index_t batch_stride_k;
ck_tile::index_t batch_stride_v;
ck_tile::index_t batch_stride_o;
};
struct HstuAttentionFwdJaggModeKargs
: HstuAttentionFwdCommonKargs,
std::conditional_t<kHasBias,
HstuAttentionFwdCommonBiasKargs,
HstuAttentionFwdEmptyKargs<0>>,
std::conditional_t<kHasMask, HstuAttentionFwdMaskKargs, HstuAttentionFwdEmptyKargs<1>>,
std::conditional_t<kHasDropout,
HstuAttentionFwdCommonDropoutKargs,
HstuAttentionFwdEmptyKargs<2>>
{
const int32_t* seq_offsets_ptr;
};
using Kargs = std::
conditional_t<kIsJagged, HstuAttentionFwdJaggModeKargs, HstuAttentionFwdBatchModeKargs>;
template <bool Cond = !kIsJagged>
CK_TILE_HOST static constexpr std::enable_if_t<Cond, Kargs>
MakeKargsImpl(const void* q_ptr,
const void* k_ptr,
const void* v_ptr,
const void* bias_ptr,
void* o_ptr,
ck_tile::index_t seqlen,
ck_tile::index_t hdim_qk,
ck_tile::index_t hdim_v,
ck_tile::index_t num_head,
float scale_s,
ck_tile::index_t seq_stride_q,
ck_tile::index_t seq_stride_k,
ck_tile::index_t seq_stride_v,
ck_tile::index_t seq_stride_bias,
ck_tile::index_t seq_stride_o,
ck_tile::index_t nhead_stride_q,
ck_tile::index_t nhead_stride_k,
ck_tile::index_t nhead_stride_v,
ck_tile::index_t nhead_stride_bias,
ck_tile::index_t nhead_stride_o,
ck_tile::index_t batch_stride_q,
ck_tile::index_t batch_stride_k,
ck_tile::index_t batch_stride_v,
ck_tile::index_t batch_stride_bias,
ck_tile::index_t batch_stride_o,
const void* num_targets_ptr,
ck_tile::index_t window_size,
ck_tile::index_t contextual_seqlen,
ck_tile::index_t min_full_attn_seqlen,
float p_drop,
const std::pair<uint64_t, uint64_t>& drop_seed_offset)
{
Kargs kargs{{q_ptr,
k_ptr,
v_ptr,
o_ptr,
seqlen,
hdim_qk,
hdim_v,
num_head,
scale_s,
seq_stride_q,
seq_stride_k,
seq_stride_v,
seq_stride_o,
nhead_stride_q,
nhead_stride_k,
nhead_stride_v,
nhead_stride_o,
reinterpret_cast<const int32_t*>(num_targets_ptr)}, // args for common karg
{}, // placeholder for bias
{}, // placeholder for mask
{}, // placeholder for dropout
batch_stride_q,
batch_stride_k,
batch_stride_v,
batch_stride_o};
if constexpr(kHasBias)
{
kargs.bias_ptr = bias_ptr;
kargs.seq_stride_bias = seq_stride_bias;
kargs.nhead_stride_bias = nhead_stride_bias;
kargs.batch_stride_bias = batch_stride_bias;
}
if constexpr(kHasMask)
{
kargs.window_size = window_size;
kargs.contextual_seqlen = contextual_seqlen;
kargs.min_full_attn_seqlen = min_full_attn_seqlen;
}
if constexpr(kHasDropout)
{
auto seed = std::get<0>(drop_seed_offset);
auto offset = std::get<1>(drop_seed_offset);
kargs.init_dropout(p_drop, seed, offset);
}
return kargs;
}
template <bool Cond = !kIsJagged>
CK_TILE_HOST static constexpr std::enable_if_t<Cond, Kargs>
MakeKargs(const void* q_ptr,
const void* k_ptr,
const void* v_ptr,
const void* bias_ptr,
void* o_ptr,
ck_tile::index_t seqlen,
ck_tile::index_t hdim_qk,
ck_tile::index_t hdim_v,
ck_tile::index_t num_head,
float scale_s,
ck_tile::index_t seq_stride_q,
ck_tile::index_t seq_stride_k,
ck_tile::index_t seq_stride_v,
ck_tile::index_t seq_stride_bias,
ck_tile::index_t seq_stride_o,
ck_tile::index_t nhead_stride_q,
ck_tile::index_t nhead_stride_k,
ck_tile::index_t nhead_stride_v,
ck_tile::index_t nhead_stride_bias,
ck_tile::index_t nhead_stride_o,
ck_tile::index_t batch_stride_q,
ck_tile::index_t batch_stride_k,
ck_tile::index_t batch_stride_v,
ck_tile::index_t batch_stride_bias,
ck_tile::index_t batch_stride_o,
const void* num_targets_ptr,
ck_tile::index_t window_size,
ck_tile::index_t contextual_seqlen,
ck_tile::index_t min_full_attn_seqlen,
float p_drop,
uint64_t philox_seed,
uint64_t philox_offset)
{
return MakeKargsImpl(q_ptr,
k_ptr,
v_ptr,
bias_ptr,
o_ptr,
seqlen,
hdim_qk,
hdim_v,
num_head,
scale_s,
seq_stride_q,
seq_stride_k,
seq_stride_v,
seq_stride_bias,
seq_stride_o,
nhead_stride_q,
nhead_stride_k,
nhead_stride_v,
nhead_stride_bias,
nhead_stride_o,
batch_stride_q,
batch_stride_k,
batch_stride_v,
batch_stride_bias,
batch_stride_o,
num_targets_ptr,
window_size,
contextual_seqlen,
min_full_attn_seqlen,
p_drop,
std::make_pair(philox_seed, philox_offset));
}
template <bool Cond = kIsJagged>
CK_TILE_HOST static constexpr std::enable_if_t<Cond, Kargs>
MakeKargsImpl(const void* q_ptr,
const void* k_ptr,
const void* v_ptr,
const void* bias_ptr,
void* o_ptr,
const void* seq_offsets_ptr,
ck_tile::index_t hdim_qk,
ck_tile::index_t hdim_v,
ck_tile::index_t num_head,
float scale_s,
ck_tile::index_t seq_stride_q,
ck_tile::index_t seq_stride_k,
ck_tile::index_t seq_stride_v,
ck_tile::index_t seq_stride_bias,
ck_tile::index_t seq_stride_o,
ck_tile::index_t nhead_stride_q,
ck_tile::index_t nhead_stride_k,
ck_tile::index_t nhead_stride_v,
ck_tile::index_t nhead_stride_bias,
ck_tile::index_t nhead_stride_o,
const void* num_targets_ptr,
ck_tile::index_t window_size,
ck_tile::index_t contextual_seqlen,
ck_tile::index_t min_full_attn_seqlen,
float p_drop,
const std::pair<uint64_t, uint64_t>& drop_seed_offset)
{
Kargs kargs{{q_ptr,
k_ptr,
v_ptr,
o_ptr,
-1, // seqlen will be updated by another pointer
hdim_qk,
hdim_v,
num_head,
scale_s,
seq_stride_q,
seq_stride_k,
seq_stride_v,
seq_stride_o,
nhead_stride_q,
nhead_stride_k,
nhead_stride_v,
nhead_stride_o,
reinterpret_cast<const int32_t*>(num_targets_ptr)}, // args for common karg
{}, // placeholder for bias
{}, // placeholder for mask
{}, // placeholder for dropout
reinterpret_cast<const int32_t*>(seq_offsets_ptr)};
if constexpr(kHasBias)
{
kargs.bias_ptr = bias_ptr;
kargs.seq_stride_bias = seq_stride_bias;
kargs.nhead_stride_bias = nhead_stride_bias;
}
if constexpr(kHasMask)
{
kargs.window_size = window_size;
kargs.contextual_seqlen = contextual_seqlen;
kargs.min_full_attn_seqlen = min_full_attn_seqlen;
}
if constexpr(kHasDropout)
{
auto seed = std::get<0>(drop_seed_offset);
auto offset = std::get<1>(drop_seed_offset);
kargs.init_dropout(p_drop, seed, offset);
}
return kargs;
}
template <bool Cond = kIsJagged>
CK_TILE_HOST static constexpr std::enable_if_t<Cond, Kargs>
MakeKargs(const void* q_ptr,
const void* k_ptr,
const void* v_ptr,
const void* bias_ptr,
void* o_ptr,
const void* seq_offsets_ptr,
ck_tile::index_t hdim_qk,
ck_tile::index_t hdim_v,
ck_tile::index_t num_head,
float scale_s,
ck_tile::index_t seq_stride_q,
ck_tile::index_t seq_stride_k,
ck_tile::index_t seq_stride_v,
ck_tile::index_t seq_stride_bias,
ck_tile::index_t seq_stride_o,
ck_tile::index_t nhead_stride_q,
ck_tile::index_t nhead_stride_k,
ck_tile::index_t nhead_stride_v,
ck_tile::index_t nhead_stride_bias,
ck_tile::index_t nhead_stride_o,
const void* num_targets_ptr,
ck_tile::index_t window_size,
ck_tile::index_t contextual_seqlen,
ck_tile::index_t min_full_attn_seqlen,
float p_drop,
uint64_t philox_seed,
uint64_t philox_offset)
{
return MakeKargsImpl(q_ptr,
k_ptr,
v_ptr,
bias_ptr,
o_ptr,
seq_offsets_ptr,
hdim_qk,
hdim_v,
num_head,
scale_s,
seq_stride_q,
seq_stride_k,
seq_stride_v,
seq_stride_bias,
seq_stride_o,
nhead_stride_q,
nhead_stride_k,
nhead_stride_v,
nhead_stride_bias,
nhead_stride_o,
num_targets_ptr,
window_size,
contextual_seqlen,
min_full_attn_seqlen,
p_drop,
std::make_pair(philox_seed, philox_offset));
}
CK_TILE_HOST static constexpr auto GridSize(ck_tile::index_t batch_size_,
ck_tile::index_t nhead_,
ck_tile::index_t seqlen_,
ck_tile::index_t hdim_v_)
{
// TODO: this may need tuning
return dim3(ck_tile::integer_divide_ceil(seqlen_, HstuAttentionPipeline::kM0) *
ck_tile::integer_divide_ceil(hdim_v_, HstuAttentionPipeline::kN1),
nhead_,
batch_size_);
}
CK_TILE_DEVICE static constexpr auto GetTileIndex(const Kargs& kargs)
{
// const index_t num_tile_m0 = seqlen_q / kM0;
const index_t num_tile_n1 =
ck_tile::integer_divide_ceil(kargs.hdim_v, HstuAttentionPipeline::kN1);
const index_t i_block = blockIdx.x;
const index_t i_nhead = blockIdx.y;
const index_t i_batch = blockIdx.z;
const auto f = [](index_t dividend, index_t divisor) {
index_t quotient = dividend / divisor;
index_t modulus = dividend - quotient * divisor;
return ck_tile::make_tuple(quotient, modulus);
};
const auto [i_tile_m, i_tile_n] = f(i_block, num_tile_n1);
return ck_tile::make_tuple(i_tile_m, i_tile_n, i_nhead, i_batch);
}
CK_TILE_HOST static constexpr auto BlockSize() { return dim3(kBlockSize); }
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetSmemSize()
{
return ck_tile::max(HstuAttentionPipeline::GetSmemSize(), EpiloguePipeline::GetSmemSize());
}
CK_TILE_DEVICE void operator()(Kargs kargs) const
{
// allocate LDS
__shared__ char smem_ptr[GetSmemSize()];
// divide problem
const auto [i_tile_m, i_tile_n, i_nhead, i_batch] = GetTileIndex(kargs);
const index_t i_m0 = __builtin_amdgcn_readfirstlane(i_tile_m * HstuAttentionPipeline::kM0);
const index_t i_n1 = __builtin_amdgcn_readfirstlane(i_tile_n * HstuAttentionPipeline::kN1);
long_index_t batch_offset_q = 0;
long_index_t batch_offset_k = 0;
long_index_t batch_offset_v = 0;
long_index_t batch_offset_bias = 0;
long_index_t batch_offset_o = 0;
if constexpr(kIsJagged)
{
// get starting offset for each batch
const long_index_t query_start = kargs.seq_offsets_ptr[i_batch];
const long_index_t key_start = query_start;
batch_offset_q = query_start * kargs.seq_stride_q;
batch_offset_k = key_start * kargs.seq_stride_k;
if constexpr(std::is_same_v<VLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
batch_offset_v = key_start * kargs.seq_stride_v;
}
else
{
batch_offset_v = key_start;
}
if constexpr(kHasBias)
{
batch_offset_bias = query_start * kargs.seq_stride_bias;
}
batch_offset_o = query_start * kargs.seq_stride_o;
kargs.seqlen = kargs.seq_offsets_ptr[1] - kargs.seq_offsets_ptr[0];
// # of required blocks is different in each groups, terminate unnecessary blocks
// earlier
if(kargs.seqlen <= i_m0)
{
return;
}
}
else
{
batch_offset_q = static_cast<long_index_t>(i_batch) * kargs.batch_stride_q;
batch_offset_k = static_cast<long_index_t>(i_batch) * kargs.batch_stride_k;
batch_offset_v = static_cast<long_index_t>(i_batch) * kargs.batch_stride_v;
if constexpr(kHasBias)
{
batch_offset_bias = static_cast<long_index_t>(i_batch) * kargs.batch_stride_bias;
}
batch_offset_o = static_cast<long_index_t>(i_batch) * kargs.batch_stride_o;
}
int max_uih_len = kargs.seqlen;
if constexpr(kHasMask)
{
if(kargs.contextual_seqlen > 0)
max_uih_len -= kargs.contextual_seqlen - 1;
};
if(kargs.num_targets_ptr != nullptr)
{
if constexpr(kIsJagged)
max_uih_len -= kargs.num_targets_ptr[i_batch];
else
max_uih_len -= kargs.num_targets_ptr[0];
};
HstuMask mask = [&]() {
if constexpr(kHasMask)
return HstuMask{kargs.window_size,
kargs.contextual_seqlen,
kargs.min_full_attn_seqlen,
max_uih_len};
else
return HstuMask{0, 0, 0, 0};
}();
// for simplicity, batch stride we just modify the pointer
const QKVDataType* q_ptr = reinterpret_cast<const QKVDataType*>(kargs.q_ptr) +
static_cast<long_index_t>(i_nhead) * kargs.nhead_stride_q +
batch_offset_q;
const QKVDataType* k_ptr = reinterpret_cast<const QKVDataType*>(kargs.k_ptr) +
static_cast<long_index_t>(i_nhead) * kargs.nhead_stride_k +
batch_offset_k;
const QKVDataType* v_ptr = reinterpret_cast<const QKVDataType*>(kargs.v_ptr) +
static_cast<long_index_t>(i_nhead) * kargs.nhead_stride_v +
batch_offset_v;
ODataType* o_ptr = reinterpret_cast<ODataType*>(kargs.o_ptr) +
static_cast<long_index_t>(i_nhead) * kargs.nhead_stride_o +
batch_offset_o;
// Q/K/V DRAM and DRAM window
const auto q_dram = [&]() {
const auto q_dram_naive = make_naive_tensor_view<address_space_enum::global>(
q_ptr,
make_tuple(kargs.seqlen, kargs.hdim_qk),
make_tuple(kargs.seq_stride_q, 1),
number<HstuAttentionPipeline::kAlignmentQ>{},
number<1>{});
if constexpr(HstuAttentionPipeline::kQLoadOnce)
{
return pad_tensor_view(q_dram_naive,
make_tuple(number<HstuAttentionPipeline::kM0>{},
number<HstuAttentionPipeline::kSubQKHeaddim>{}),
sequence<kPadSeqLenQ, kPadHeadDimQK>{});
}
else
{
return pad_tensor_view(q_dram_naive,
make_tuple(number<HstuAttentionPipeline::kM0>{},
number<HstuAttentionPipeline::kK0>{}),
sequence<kPadSeqLenQ, kPadHeadDimQK>{});
}
}();
const auto k_dram = [&]() {
const auto k_dram_naive = make_naive_tensor_view<address_space_enum::global>(
k_ptr,
make_tuple(kargs.seqlen, kargs.hdim_qk),
make_tuple(kargs.seq_stride_k, 1),
number<HstuAttentionPipeline::kAlignmentK>{},
number<1>{});
return pad_tensor_view(k_dram_naive,
make_tuple(number<HstuAttentionPipeline::kN0>{},
number<HstuAttentionPipeline::kK0>{}),
sequence<false, kPadHeadDimQK>{});
}();
const auto v_dram = [&]() {
if constexpr(std::is_same_v<VLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
const auto v_dram_naive = make_naive_tensor_view<address_space_enum::global>(
v_ptr,
make_tuple(kargs.seqlen, kargs.hdim_v),
make_tuple(kargs.seq_stride_v, 1),
number<HstuAttentionPipeline::kAlignmentV>{},
number<1>{});
const auto v_dram_transposed =
transform_tensor_view(v_dram_naive,
make_tuple(make_pass_through_transform(kargs.hdim_v),
make_pass_through_transform(kargs.seqlen)),
make_tuple(sequence<1>{}, sequence<0>{}),
make_tuple(sequence<0>{}, sequence<1>{}));
return pad_tensor_view(v_dram_transposed,
make_tuple(number<HstuAttentionPipeline::kN1>{},
number<HstuAttentionPipeline::kK1>{}),
sequence<kPadHeadDimV, false>{});
}
else
{
const auto v_dram_naive = make_naive_tensor_view<address_space_enum::global>(
v_ptr,
make_tuple(kargs.hdim_v, kargs.seqlen),
make_tuple(kargs.seq_stride_v, 1),
number<HstuAttentionPipeline::kAlignmentV>{},
number<1>{});
return pad_tensor_view(v_dram_naive,
make_tuple(number<HstuAttentionPipeline::kN1>{},
number<HstuAttentionPipeline::kK1>{}),
sequence<kPadHeadDimV, false>{});
}
}();
auto q_dram_window = make_tile_window(
q_dram,
[&]() {
if constexpr(HstuAttentionPipeline::kQLoadOnce)
return make_tuple(number<HstuAttentionPipeline::kM0>{},
number<HstuAttentionPipeline::kSubQKHeaddim>{});
else
return make_tuple(number<HstuAttentionPipeline::kM0>{},
number<HstuAttentionPipeline::kK0>{});
}(),
{i_m0, 0});
auto k_dram_window = make_tile_window(
k_dram,
make_tuple(number<HstuAttentionPipeline::kN0>{}, number<HstuAttentionPipeline::kK0>{}),
{0, 0});
auto v_dram_window = make_tile_window(
v_dram,
make_tuple(number<HstuAttentionPipeline::kN1>{}, number<HstuAttentionPipeline::kK1>{}),
{i_n1, 0});
/// FIXME: Before C++20, capturing structured binding variables are not supported. Remove
/// following copy capture of the 'i_nhead' if in C++20
const auto bias_dram_window = [&, i_nhead_ = i_nhead]() {
constexpr auto bias_dram_window_lengths = make_tuple(
number<HstuAttentionPipeline::kM0>{}, number<HstuAttentionPipeline::kN0>{});
if constexpr(kHasBias)
{
const BiasDataType* bias_ptr =
reinterpret_cast<const BiasDataType*>(kargs.bias_ptr) +
static_cast<long_index_t>(i_nhead_) * kargs.nhead_stride_bias +
batch_offset_bias;
const auto bias_dram = [&]() {
const auto bias_dram_naive = make_naive_tensor_view<address_space_enum::global>(
bias_ptr,
make_tuple(kargs.seqlen, kargs.seqlen),
make_tuple(kargs.seq_stride_bias, 1),
number<HstuAttentionPipeline::kAlignmentBias>{},
number<1>{});
return pad_tensor_view(bias_dram_naive,
bias_dram_window_lengths,
sequence<kPadSeqLenQ, kPadSeqLenK>{});
}();
return make_tile_window(bias_dram, bias_dram_window_lengths, {i_m0, 0});
}
else
{
return make_null_tile_window(bias_dram_window_lengths);
}
}();
auto dropout = [&, i_nhead_ = i_nhead, i_batch_ = i_batch]() {
if constexpr(kHasDropout)
{
return BlockDropout{i_batch_,
i_nhead_,
kargs.num_head,
kargs.drop_seed,
kargs.drop_offset,
kargs.rp_undrop,
kargs.p_undrop_in_uint8_t,
false};
}
else
{
return NullBlockDropout{};
};
}();
auto o_acc_tile = [&]() {
return HstuAttentionPipeline{}(q_dram_window,
k_dram_window,
v_dram_window,
bias_dram_window,
mask,
kargs.scale_s,
smem_ptr,
dropout);
}();
// O DRAM and O DRAM window
auto o_dram = [&]() {
const auto o_dram_naive = make_naive_tensor_view<address_space_enum::global>(
o_ptr,
make_tuple(kargs.seqlen, kargs.hdim_v),
make_tuple(kargs.seq_stride_o, 1),
number<HstuAttentionPipeline::kAlignmentO>{},
number<1>{});
return pad_tensor_view(o_dram_naive,
make_tuple(number<HstuAttentionPipeline::kM0>{},
number<HstuAttentionPipeline::kN1>{}),
sequence<kPadSeqLenQ, kPadHeadDimV>{});
}();
auto o_dram_window = make_tile_window(
o_dram,
make_tuple(number<HstuAttentionPipeline::kM0>{}, number<HstuAttentionPipeline::kN1>{}),
{i_m0, i_n1});
EpiloguePipeline{}(o_dram_window, o_acc_tile);
}
};
} // namespace ck_tile

View File

@@ -0,0 +1,548 @@
// 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/fmha/block/block_dropout.hpp"
#include "hstu_attention_fwd_pipeline_default_policy.hpp"
namespace ck_tile {
template <typename Problem_, typename Policy_ = HstuAttentionFwdPipelineQRKSVSDefaultPolicy>
struct HstuAttentionFwdPipelineQRKSVS
{
using Problem = remove_cvref_t<Problem_>;
using Policy = remove_cvref_t<Policy_>;
using QKVDataType = remove_cvref_t<typename Problem::InOutDataType>;
using GemmAccDataType = remove_cvref_t<typename Problem::GemmAccDataType>;
using CompDataType = remove_cvref_t<typename Problem::CompDataType>;
using BiasDataType = remove_cvref_t<typename Problem::BiasDataType>;
using PDataType = remove_cvref_t<typename Problem::InOutDataType>;
using ODataType = remove_cvref_t<typename Problem::InOutDataType>;
using HstuMask = remove_cvref_t<typename Problem::HstuMask>;
using HstuAttentionTileShape = remove_cvref_t<typename Problem::HstuAttentionTileShape>;
using VLayout = remove_cvref_t<typename HstuAttentionTileShape::VLayout>;
static constexpr bool kQLoadOnce = true;
static_assert(kQLoadOnce == Policy::QLoadOnce);
static constexpr index_t kBlockSize = Problem::kBlockSize;
static constexpr index_t kM0 = HstuAttentionTileShape::kM0;
static constexpr index_t kN0 = HstuAttentionTileShape::kN0;
static constexpr index_t kK0 = HstuAttentionTileShape::kK0;
static constexpr index_t kN1 = HstuAttentionTileShape::kN1;
static constexpr index_t kK1 = HstuAttentionTileShape::kK1;
static constexpr index_t kQKHeaddim = HstuAttentionTileShape::kQKHeaddim;
static constexpr index_t kSubQKHeaddim = HstuAttentionTileShape::kSubQKHeaddim;
static_assert(kSubQKHeaddim <= 256, "hdim bigger than 256 is not suitable for this pipeline!");
static constexpr bool kIsJagged = Problem::kIsJagged;
static constexpr auto kHasBias = Problem::kHasBias;
static constexpr bool kHasDropout = Problem::kHasDropout;
static constexpr bool kPadSeqLenQ = Problem::Traits::kPadSeqLenQ;
static constexpr bool kPadSeqLenK = Problem::Traits::kPadSeqLenK;
static constexpr bool kPadHeadDimQK = Problem::Traits::kPadHeadDimQK;
static constexpr bool kPadHeadDimV =
(kQKHeaddim < kSubQKHeaddim) ? 1 : Problem::Traits::kPadHeadDimV;
// last dimension vector length used to create tensor view(and decide buffer_load vector length)
// ... together with tensor distribution. tensor dist should able to overwrite this
static constexpr index_t kAlignmentQ =
kPadHeadDimQK ? 1 : Policy::template GetAlignmentQ<Problem>();
static constexpr index_t kAlignmentK =
kPadHeadDimQK ? 1 : Policy::template GetAlignmentK<Problem>();
static constexpr index_t kAlignmentV = []() {
if constexpr(std::is_same_v<VLayout, ck_tile::tensor_layout::gemm::RowMajor>)
return Problem::Traits::kPadHeadDimV ? 1 : Policy::template GetAlignmentV<Problem>();
else
return kPadSeqLenK ? 1 : Policy::template GetAlignmentV<Problem>();
}();
static constexpr index_t kAlignmentO =
kPadHeadDimV ? 1 : Policy::template GetAlignmentO<Problem>();
static constexpr index_t kAlignmentBias =
kPadSeqLenK ? 1 : Policy::template GetAlignmentBias<Problem>();
static constexpr index_t kBlockPerCu = []() {
if constexpr(Problem::Traits::kBlockPerCu != -1)
return Problem::Traits::kBlockPerCu;
else
{
if constexpr(kQKHeaddim == 32)
{
return 2;
}
else if constexpr(kQKHeaddim == 64)
{
return 2;
}
else if constexpr(kQKHeaddim == 96 || kQKHeaddim == 128)
{
if constexpr(kHasBias)
return 1;
else
return 2;
}
else if constexpr(kQKHeaddim == 256)
{
return 1;
}
else
{
return 1;
};
}
}();
static constexpr const char* name = "qr_hstu";
using DropoutType = std::conditional_t<kHasDropout, BlockDropout, NullBlockDropout>;
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetSmemSize()
{
return Policy::template GetSmemSize<Problem>();
}
template <typename QDramBlockWindowTmp,
typename KDramBlockWindowTmp,
typename VDramBlockWindowTmp,
typename BiasDramBlockWindowTmp,
typename QElementFunction,
typename KElementFunction,
typename VElementFunction,
typename BiasElementFunction,
typename SAccElementFunction,
typename PComputeElementFunction,
typename OAccElementFunction>
CK_TILE_HOST_DEVICE auto
operator()(const QDramBlockWindowTmp& q_dram_block_window_tmp, // M0*kSubQKHeaddim tile
const QElementFunction& q_element_func,
const KDramBlockWindowTmp& k_dram_block_window_tmp, // N0*kSubQKHeaddim tile
const KElementFunction& k_element_func,
const VDramBlockWindowTmp& v_dram_block_window_tmp, // N1*K1 tile
const VElementFunction& v_element_func,
const BiasDramBlockWindowTmp& bias_dram_block_window_tmp, // M0*N0 tile
const BiasElementFunction& bias_element_func,
const SAccElementFunction& s_acc_element_func,
const PComputeElementFunction& p_compute_element_func,
const OAccElementFunction& o_acc_element_func,
HstuMask mask,
float scale_s,
void* smem_ptr,
DropoutType& dropout) const
{
ignore = q_element_func;
ignore = k_element_func;
static_assert(
std::is_same_v<QKVDataType, remove_cvref_t<typename QDramBlockWindowTmp::DataType>> &&
std::is_same_v<QKVDataType,
remove_cvref_t<typename KDramBlockWindowTmp::DataType>> &&
std::is_same_v<QKVDataType, remove_cvref_t<typename VDramBlockWindowTmp::DataType>>,
"wrong!");
static_assert(kM0 == QDramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
kN0 == KDramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
kK0 == KDramBlockWindowTmp{}.get_window_lengths()[number<1>{}] &&
kN1 == VDramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
kK1 == VDramBlockWindowTmp{}.get_window_lengths()[number<1>{}] &&
kM0 == BiasDramBlockWindowTmp{}.get_window_lengths()[number<0>{}] &&
kN0 == BiasDramBlockWindowTmp{}.get_window_lengths()[number<1>{}],
"wrong!");
constexpr auto I0 = number<0>{};
constexpr index_t k0_loops = kQKHeaddim / kK0;
constexpr index_t k1_loops = kN0 / kK1;
static_assert(2 <= k0_loops);
static_assert(2 <= k1_loops);
constexpr auto NumKLdsBuffers = Policy::template GetNumKLdsBuffers<Problem>();
constexpr auto NumVLdsBuffers = Policy::template GetNumVLdsBuffers<Problem>();
constexpr auto NumPrefetchV = Policy::template GetNumPrefetchV<Problem>();
static_assert(NumKLdsBuffers >= 2);
auto q_dram_window = make_tile_window(q_dram_block_window_tmp.get_bottom_tensor_view(),
q_dram_block_window_tmp.get_window_lengths(),
q_dram_block_window_tmp.get_window_origin(),
Policy::template MakeQRegTileDistribution<Problem>());
const auto q_origin = q_dram_window.get_window_origin();
const auto [seqlen_k_start, seqlen_k_end] =
mask.GetTileRangeAlongX(q_origin.at(number<0>{}), number<kM0>{}, number<kN0>{});
auto k_dram_block_window =
make_tile_window(k_dram_block_window_tmp.get_bottom_tensor_view(),
k_dram_block_window_tmp.get_window_lengths(),
{seqlen_k_start, 0});
auto k_dram_window =
make_tile_window(k_dram_block_window.get_bottom_tensor_view(),
k_dram_block_window.get_window_lengths(),
k_dram_block_window.get_window_origin(),
Policy::template MakeKDramTileDistribution<Problem>());
auto k_tile = load_tile(k_dram_window);
move_tile_window(k_dram_window, {0, kK0});
auto q_tile = load_tile(q_dram_window);
__builtin_amdgcn_sched_barrier(0);
// K tile in LDS
QKVDataType* k_lds_ptr = static_cast<QKVDataType*>(smem_ptr);
auto k_lds = make_tensor_view<address_space_enum::lds>(
k_lds_ptr, Policy::template MakeKLdsBlockDescriptor<Problem>());
auto k_lds_window = make_tile_window(
k_lds, Policy::template MakeKLdsBlockDescriptor<Problem>().get_lengths(), {0, 0});
using k_lds_window_type =
decltype(get_slice_tile(k_lds_window, sequence<0, 0>{}, sequence<kN0, kK0>{}));
statically_indexed_array<k_lds_window_type, NumKLdsBuffers> k_lds_windows;
static_for<0, NumKLdsBuffers, 1>{}([&](auto i_buf) {
k_lds_windows[i_buf] = get_slice_tile(
k_lds_window, sequence<i_buf * kN0, 0>{}, sequence<(i_buf + 1) * kN0, kK0>{});
});
auto v_dram_window =
make_tile_window(v_dram_block_window_tmp.get_bottom_tensor_view(),
v_dram_block_window_tmp.get_window_lengths(),
{0, seqlen_k_start}, // TODO: hdim split?
Policy::template MakeVDramTileDistribution<Problem>());
// V tile in LDS
auto v_lds = make_tensor_view<address_space_enum::lds>(
reinterpret_cast<QKVDataType*>(static_cast<char*>(smem_ptr) +
Policy::template GetExclusiveKLdsBytes<Problem>()),
Policy::template MakeVLdsBlockDescriptor<Problem>());
auto v_lds_window = make_tile_window(
v_lds, Policy::template MakeVLdsBlockDescriptor<Problem>().get_lengths(), {0, 0});
using v_tile_type = decltype(load_tile(v_dram_window));
statically_indexed_array<v_tile_type, NumPrefetchV> v_tiles;
using v_lds_window_type =
decltype(get_slice_tile(v_lds_window, sequence<0, 0>{}, sequence<kN1, kK1>{}));
statically_indexed_array<v_lds_window_type, NumVLdsBuffers> v_lds_windows;
static_for<0, NumVLdsBuffers, 1>{}([&](auto i_buf) {
v_lds_windows[i_buf] = get_slice_tile(
v_lds_window, sequence<i_buf * kN1, 0>{}, sequence<(i_buf + 1) * kN1, kK1>{});
});
// Block GEMM
constexpr auto gemm_0 = Policy::template GetQKBlockGemm<Problem>();
constexpr auto gemm_1 = Policy::template GetKVBlockGemm<Problem>();
using SaccBlockTileType = decltype(gemm_0.MakeCBlockTile());
auto s_acc = SaccBlockTileType{};
// reduction function for softmax
const auto f_silu = [](CompDataType x) {
auto one = ck_tile::type_convert<CompDataType>(1.0f);
auto sigmod_val = one / (one + exp(-x));
return sigmod_val * x;
};
using OaccBlockTileType = decltype(gemm_1.MakeCBlockTile());
// init Oacc, M, L
auto o_acc = OaccBlockTileType{};
clear_tile(o_acc);
const auto num_loops = integer_divide_ceil(seqlen_k_end - seqlen_k_start, kN0);
// check early exit if no work to do
if constexpr(HstuMask::IsMasking || kPadSeqLenK)
{
if(num_loops <= 0)
{
return o_acc;
}
}
const auto bias_origin = bias_dram_block_window_tmp.get_window_origin();
auto bias_dram_window =
make_tile_window(bias_dram_block_window_tmp.get_bottom_tensor_view(),
bias_dram_block_window_tmp.get_window_lengths(),
{bias_origin.at(number<0>{}), seqlen_k_start}, // M/N
Policy::template MakeBiasDramTileDistribution<decltype(gemm_0)>());
auto null_randval_window = [&]() {
if constexpr(kHasDropout)
{
const auto null_randval_dram = [&]() {
const auto null_dram_naive = make_naive_tensor_view<address_space_enum::global>(
static_cast<uint8_t*>(nullptr),
make_tuple(1, 1),
make_tuple(1, 1),
number<1>{},
number<1>{});
return pad_tensor_view(null_dram_naive,
make_tuple(number<1>{}, number<1>{}),
sequence<true, true>{});
}();
return make_tile_window(
null_randval_dram, make_tuple(number<1>{}, number<1>{}), {0, 0});
}
else
return make_null_tile_window(make_tuple(number<1>{}, number<1>{}));
}();
q_tile = tile_elementwise_in(q_element_func, q_tile);
index_t i_loop = 0;
do
{
static_for<0, k0_loops - 1, 1>{}([&](auto i_k0) {
store_tile(k_lds_windows[number<i_k0 % NumKLdsBuffers>{}],
tile_elementwise_in(k_element_func, k_tile));
if constexpr(i_k0 == 0)
clear_tile(s_acc);
if constexpr(i_k0 < k0_loops - 1)
k_tile = load_tile(k_dram_window);
if constexpr(i_k0 < k0_loops - 2)
move_tile_window(k_dram_window, {0, kK0});
block_sync_lds();
// execute current unroll of gemm_0
gemm_0(s_acc,
get_slice_tile(
q_tile, sequence<0, i_k0 * kK0>{}, sequence<kM0, (i_k0 + 1) * kK0>{}),
k_lds_windows[number<i_k0 % NumKLdsBuffers>{}]);
});
store_tile(k_lds_windows[number<(k0_loops - 1) % NumKLdsBuffers>{}],
tile_elementwise_in(k_element_func, k_tile));
// prefetch first v_tile
v_tiles[I0] = load_tile(v_dram_window);
move_tile_window(v_dram_window, {0, kK1});
block_sync_lds();
gemm_0(s_acc,
get_slice_tile(q_tile,
sequence<0, (k0_loops - 1) * kK0>{},
sequence<kM0, k0_loops * kK0>{}),
k_lds_windows[number<(k0_loops - 1) % NumKLdsBuffers>{}]);
__builtin_amdgcn_sched_barrier(0);
const auto bias_tile = load_tile(bias_dram_window); // load bias tile
static_for<1, NumPrefetchV, 1>{}([&](auto i_buf) {
v_tiles[i_buf] = load_tile(v_dram_window);
move_tile_window(v_dram_window, {0, kK1});
});
// STAGE 2, scale_s, add bias, mask, siLU
if constexpr(kHasBias)
{
s_acc = tile_elementwise_in(s_acc_element_func, s_acc);
tile_elementwise_inout([&scale_s](auto& x) { x = x * scale_s; }, s_acc);
tile_elementwise_inout(
[&](auto& x, const auto& y) {
x += type_convert<GemmAccDataType>(bias_element_func(y));
},
s_acc,
bias_tile);
}
else
{
s_acc = tile_elementwise_in(s_acc_element_func, s_acc);
tile_elementwise_inout([&scale_s](auto& x) { x = x * scale_s; }, s_acc);
}
move_tile_window(bias_dram_window, {0, kN0});
if constexpr(HstuMask::IsMasking)
{
const auto k_origin = k_dram_block_window.get_window_origin();
set_tile_if(s_acc, -numeric<CompDataType>::infinity(), [&](auto tile_idx) {
const auto row = q_origin.at(number<0>{}) + tile_idx.at(number<0>{});
const auto col = k_origin.at(number<0>{}) + tile_idx.at(number<1>{});
return mask.IsTokenPairInsideMask(row, col);
});
}
else if constexpr(kPadSeqLenK)
{
const auto k_origin = k_dram_block_window.get_window_origin();
set_tile_if(s_acc, -numeric<CompDataType>::infinity(), [&](auto tile_idx) {
if(i_loop < num_loops)
return false;
const auto row = q_origin.at(number<0>{}) + tile_idx.at(number<0>{});
const auto col = k_origin.at(number<0>{}) + tile_idx.at(number<1>{});
return mask.IsTokenPairInsideMask(row, col);
});
};
auto s = cast_tile<CompDataType>(s_acc); // S{j}
s = tile_elementwise_in(f_silu, s);
if constexpr(kHasDropout)
{
auto randval_lds_ptr =
reinterpret_cast<char*>(smem_ptr) + Policy::template GetSmemSizeK<Problem>();
dropout.template Run<decltype(gemm_0), CompDataType, uint8_t>(
randval_lds_ptr, seqlen_k_start + i_loop * kN0, s, null_randval_window);
}
__builtin_amdgcn_sched_barrier(0x7f);
if constexpr(std::is_same_v<VLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
auto v_shuffle_tmp = make_static_distributed_tensor<QKVDataType>(
Policy::template MakeShuffledVRegBlockDescriptor<Problem>());
shuffle_tile(v_shuffle_tmp, v_tiles[I0]);
store_tile(
v_lds_windows[I0],
tile_elementwise_in(v_element_func, v_shuffle_tmp)); // store the prefetch
}
else
{
store_tile(v_lds_windows[I0],
tile_elementwise_in(v_element_func, v_tiles[I0])); // store the prefetch
}
__builtin_amdgcn_sched_barrier(0);
const auto p = cast_tile<PDataType>(tile_elementwise_in(p_compute_element_func, s));
// STAGE 3, KV gemm
if constexpr(k1_loops > 1)
{
if constexpr(NumPrefetchV == 1) // NumVLdsBuffers == 2
{
static_for<0, k1_loops - 1, 1>{}([&](auto i_k1) {
v_tiles[I0] = load_tile(v_dram_window);
block_sync_lds();
gemm_1(o_acc,
get_slice_tile(
p, sequence<0, i_k1 * kK1>{}, sequence<kM0, (i_k1 + 1) * kK1>{}),
v_lds_windows[number<i_k1 % NumVLdsBuffers>{}]);
if constexpr(std::is_same_v<VLayout,
ck_tile::tensor_layout::gemm::RowMajor>)
{
auto v_shuffle_tmp = make_static_distributed_tensor<QKVDataType>(
Policy::template MakeShuffledVRegBlockDescriptor<Problem>());
shuffle_tile(v_shuffle_tmp, v_tiles[I0]);
store_tile(v_lds_windows[number<(i_k1 + 1) % NumVLdsBuffers>{}],
tile_elementwise_in(v_element_func, v_shuffle_tmp));
}
else
{
store_tile(v_lds_windows[number<(i_k1 + 1) % NumVLdsBuffers>{}],
tile_elementwise_in(v_element_func, v_tiles[I0]));
}
move_tile_window(v_dram_window, {0, kK1});
});
}
else // NumVLdsBuffers == 3 or 2
{
static_for<0, k1_loops - 1, 1>{}([&](auto i_k1) {
if constexpr(i_k1 < k1_loops - NumPrefetchV)
v_tiles[number<i_k1 % NumPrefetchV>{}] = load_tile(v_dram_window);
block_sync_lds();
gemm_1(o_acc,
get_slice_tile(
p, sequence<0, i_k1 * kK1>{}, sequence<kM0, (i_k1 + 1) * kK1>{}),
v_lds_windows[number<i_k1 % NumVLdsBuffers>{}]);
if constexpr(std::is_same_v<VLayout,
ck_tile::tensor_layout::gemm::RowMajor>)
{
auto v_shuffle_tmp = make_static_distributed_tensor<QKVDataType>(
Policy::template MakeShuffledVRegBlockDescriptor<Problem>());
shuffle_tile(v_shuffle_tmp,
v_tiles[number<(i_k1 + 1) % NumPrefetchV>{}]);
store_tile(v_lds_windows[number<(i_k1 + 1) % NumVLdsBuffers>{}],
tile_elementwise_in(v_element_func, v_shuffle_tmp));
}
else
{
store_tile(
v_lds_windows[number<(i_k1 + 1) % NumVLdsBuffers>{}],
tile_elementwise_in(v_element_func,
v_tiles[number<(i_k1 + 1) % NumPrefetchV>{}]));
}
if constexpr(i_k1 < k1_loops - NumPrefetchV)
move_tile_window(v_dram_window, {0, kK1});
});
}
}
// move K tile windows
move_tile_window(k_dram_block_window, {kN0, 0});
block_sync_lds();
gemm_1(o_acc,
get_slice_tile(p, sequence<0, (k1_loops - 1) * kK1>{}, sequence<kM0, kN0>{}),
v_lds_windows[number<(k1_loops - 1) % NumVLdsBuffers>{}]);
if constexpr(Policy::template IsFirstKLdsBufferOverlapLastVLdsBuffer<Problem>())
{
__builtin_amdgcn_sched_barrier(0);
__builtin_amdgcn_s_barrier();
};
} while(++i_loop < num_loops);
o_acc = tile_elementwise_in(o_acc_element_func, o_acc);
return o_acc;
}
template <typename QDramBlockWindowTmp,
typename KDramBlockWindowTmp,
typename VDramBlockWindowTmp,
typename BiasDramBlockWindowTmp>
CK_TILE_HOST_DEVICE auto
operator()(const QDramBlockWindowTmp& q_dram_block_window_tmp, // M0*K0 tile
const KDramBlockWindowTmp& k_dram_block_window_tmp, // N0*K0 tile
const VDramBlockWindowTmp& v_dram_block_window_tmp, // N1*K1 tile
const BiasDramBlockWindowTmp& bias_dram_block_window_tmp, // M0*N0 tile
HstuMask mask,
float scale_s,
void* smem_ptr,
DropoutType& dropout) const
{
return operator()(q_dram_block_window_tmp,
identity{},
k_dram_block_window_tmp,
identity{},
v_dram_block_window_tmp,
identity{},
bias_dram_block_window_tmp,
identity{},
identity{},
identity{},
identity{},
mask,
scale_s,
smem_ptr,
dropout);
}
};
} // namespace ck_tile

View File

@@ -0,0 +1,370 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
#include "ck_tile/ops/fmha/pipeline/block_fmha_pipeline_qx_ks_vs_custom_policy.hpp"
namespace ck_tile {
struct HstuAttentionFwdPipelineQRKSVSDefaultPolicy
: BlockFmhaPipelineQXKSVSCustomPolicy</* QLoadOnce = */ true,
/* AsyncCopy = */ false,
/* NumPrefetchK = */ -1,
/* NumPrefetchV = */ 2>
{
static constexpr index_t NumPrefetchV = 2;
template <typename Problem>
CK_TILE_DEVICE static constexpr auto GetNumKLdsBuffers()
{
return 2;
}
template <typename Problem>
CK_TILE_DEVICE static constexpr auto GetNumPrefetchV()
{
using BlockFmhaShape = remove_cvref_t<typename Problem::BlockFmhaShape>;
constexpr index_t kN0 = BlockFmhaShape::kN0;
constexpr index_t kK1 = BlockFmhaShape::kK1;
constexpr index_t k1_loops = kN0 / kK1;
return min(NumPrefetchV, k1_loops);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetNumVLdsBuffers()
{
return 2;
};
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeQRegTileDistribution()
{
using BlockGemm = remove_cvref_t<decltype(GetQKBlockGemm<Problem>())>;
return BlockGemm::template MakeABlockTileDistribution<
Problem::BlockFmhaShape::kM0,
Problem::BlockFmhaShape::kQKHeaddim>();
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetSmemKPackK()
{
using QKVDataType = remove_cvref_t<typename Problem::QKVDataType>;
return 8 / sizeof(QKVDataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeKLdsBlockDescriptor()
{
constexpr index_t NumKLdsBuffers = GetNumKLdsBuffers<Problem>();
constexpr index_t kNPerBlock = Problem::BlockFmhaShape::kN0;
constexpr index_t kKPerBlock = Problem::BlockFmhaShape::kK0;
constexpr index_t kKPack = GetSmemKPackK<Problem>();
constexpr index_t kKVector = GetAlignmentK<Problem>();
static_assert(kKVector % kKPack == 0);
constexpr auto k_lds_block_desc_0 = make_naive_tensor_descriptor(
make_tuple(number<NumKLdsBuffers>{},
number<kKPerBlock / kKVector>{},
number<kKVector / kKPack>{},
number<kNPerBlock>{},
number<kKPack>{}),
make_tuple(number<kKPerBlock * kNPerBlock + kKPerBlock * kKPack / kKVector>{},
number<kNPerBlock * kKVector + kKPack>{},
number<kNPerBlock * kKPack>{},
number<kKPack>{},
number<1>{}),
number<kKPack>{},
number<1>{});
constexpr auto k_lds_block_desc = transform_tensor_descriptor(
k_lds_block_desc_0,
make_tuple(
make_merge_transform(make_tuple(number<NumKLdsBuffers>{}, number<kNPerBlock>{})),
make_merge_transform(make_tuple(number<kKPerBlock / kKVector>{},
number<kKVector / kKPack>{},
number<kKPack>{}))),
make_tuple(sequence<0, 3>{}, sequence<1, 2, 4>{}),
make_tuple(sequence<0>{}, sequence<1>{}));
return k_lds_block_desc;
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeKDramTileDistribution()
{
using QKVDataType = remove_cvref_t<typename Problem::QKVDataType>;
constexpr index_t kBlockSize = Problem::kBlockSize;
constexpr index_t kNPerBlock = Problem::BlockFmhaShape::kN0;
constexpr index_t kKPerBlock = Problem::BlockFmhaShape::kK0;
constexpr index_t MaxVectorSize = 16 / sizeof(QKVDataType);
constexpr index_t ElemPerThread = (kNPerBlock * kKPerBlock) / kBlockSize;
static_assert(0 < ElemPerThread);
constexpr index_t kMaxVecLoad = min(ElemPerThread, MaxVectorSize);
constexpr index_t KPerThread = kMaxVecLoad;
constexpr index_t KThreads = kKPerBlock / KPerThread;
constexpr index_t NThreadPerWarp = get_warp_size() / KThreads;
constexpr index_t NumWarps = kBlockSize / get_warp_size();
constexpr index_t NPerThread = kNPerBlock / (NThreadPerWarp * NumWarps);
return make_static_tile_distribution(
tile_distribution_encoding<sequence<1>,
tuple<sequence<NPerThread, NThreadPerWarp, NumWarps>,
sequence<KThreads, KPerThread>>,
tuple<sequence<1>, sequence<1, 2>>,
tuple<sequence<2>, sequence<1, 0>>,
sequence<1, 2>,
sequence<0, 1>>{});
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto MakeVLdsBlockDescriptor()
{
using QKVDataType = remove_cvref_t<typename Problem::QKVDataType>;
constexpr index_t NumVLdsBuffers = GetNumVLdsBuffers<Problem>();
constexpr index_t Banks = 32; // TODO: need change based on arch
constexpr index_t PixelsPerRow = Banks * 4 / sizeof(QKVDataType);
constexpr index_t kKPack = GetSmemKPackV<Problem>();
static_assert(PixelsPerRow % kKPack == 0);
constexpr index_t NPerRow = PixelsPerRow / kKPack;
constexpr index_t kNPerBlock = Problem::BlockFmhaShape::kN1;
constexpr index_t kKPerBlock = Problem::BlockFmhaShape::kK1;
static_assert(kNPerBlock % NPerRow == 0);
static_assert(kKPerBlock % kKPack == 0);
constexpr index_t VSingleSmemElementSpaceSize =
(kKPerBlock / kKPack) * (kNPerBlock / NPerRow) * (PixelsPerRow + kKPack);
constexpr auto v_lds_block_desc_0 = make_naive_tensor_descriptor(
make_tuple(number<NumVLdsBuffers>{},
number<kKPerBlock / kKPack>{},
number<kNPerBlock / NPerRow>{},
number<NPerRow>{},
number<kKPack>{}),
make_tuple(number<VSingleSmemElementSpaceSize>{},
number<(kNPerBlock / NPerRow) * (PixelsPerRow + kKPack)>{},
number<PixelsPerRow + kKPack>{},
number<kKPack>{},
number<1>{}),
number<kKPack>{},
number<1>{});
constexpr auto v_lds_block_desc = transform_tensor_descriptor(
v_lds_block_desc_0,
make_tuple(
make_merge_transform(make_tuple(
number<NumVLdsBuffers>{}, number<kNPerBlock / NPerRow>{}, number<NPerRow>{})),
make_merge_transform(make_tuple(number<kKPerBlock / kKPack>{}, number<kKPack>{}))),
make_tuple(sequence<0, 2, 3>{}, sequence<1, 4>{}),
make_tuple(sequence<0>{}, sequence<1>{}));
return v_lds_block_desc;
}
template <typename Problem>
CK_TILE_DEVICE static constexpr auto MakeVDramTileDistribution()
{
using VLayout = remove_cvref_t<typename Problem::BlockFmhaShape::VLayout>;
constexpr index_t kBlockSize = Problem::kBlockSize;
constexpr index_t kNPerBlock = Problem::BlockFmhaShape::kN1;
constexpr index_t kKPerBlock = Problem::BlockFmhaShape::kK1;
if constexpr(std::is_same_v<VLayout, ck_tile::tensor_layout::gemm::RowMajor>)
{
constexpr index_t N1 = GetAlignmentV<Problem>();
constexpr index_t N0 = kNPerBlock / N1; // P
constexpr index_t ElemPerThread = kNPerBlock * kKPerBlock / kBlockSize;
static_assert(ElemPerThread % N1 == 0);
constexpr index_t K3 = ElemPerThread / N1;
constexpr index_t kKPack = GetSmemKPackV<Problem>();
static_assert(kKPack % K3 == 0);
constexpr index_t K2 = kKPack / K3;
if constexpr(get_warp_size() % (K2 * N0) == 0)
{
constexpr index_t K1 = get_warp_size() / (K2 * N0);
constexpr index_t K0 = kBlockSize / get_warp_size();
static_assert(kKPerBlock == 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 = 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<2, 1>,
sequence<3, 1>>{});
}
}
else
{
constexpr index_t K1 = GetAlignmentV<Problem>();
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);
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>>{});
}
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetQKBlockGemm()
{
using GemmProblem =
BlockGemmProblem<typename Problem::QKVDataType,
typename Problem::QKVDataType,
typename Problem::GemmAccDataType,
Problem::kNumGemm0Warps * get_warp_size(),
TileGemmShape<sequence<Problem::BlockFmhaShape::kM0,
Problem::BlockFmhaShape::kN0,
Problem::BlockFmhaShape::kK0>,
typename Problem::BlockFmhaShape::Gemm0BlockWarps,
typename Problem::BlockFmhaShape::Gemm0WarpTile>>;
constexpr auto warp_gemm = []() {
constexpr index_t WarpGemmM = Problem::BlockFmhaShape::Gemm0WarpTile::at(number<0>{});
static_assert(WarpGemmM == 4 || WarpGemmM == 16 || WarpGemmM == 32);
if constexpr(std::is_same_v<typename Problem::QKVDataType, half_t> &&
std::is_same_v<typename Problem::GemmAccDataType, float>)
{
if constexpr(WarpGemmM == 32)
return WarpGemmMfmaF16F16F32M32N32K16SwizzleBTransposedCDistribution{};
else if constexpr(WarpGemmM == 16)
return WarpGemmMfmaF16F16F32M16N16K16TransposedCDistribution{};
else // WarpGemmM == 4
return WarpGemmMfmaF16F16F32M4N64K16{};
}
else if constexpr(std::is_same_v<typename Problem::QKVDataType, bf16_t> &&
std::is_same_v<typename Problem::GemmAccDataType, float>)
{
if constexpr(WarpGemmM == 32)
return WarpGemmMfmaBf16Bf16F32M32N32K16SwizzleBTransposedCDistribution{};
else if constexpr(WarpGemmM == 16)
return WarpGemmMfmaBf16Bf16F32M16N16K16TransposedCDistribution{};
else // WarpGemmM == 4
return WarpGemmMfmaBf16Bf16F32M4N64K16{};
}
else if constexpr(std::is_same_v<typename Problem::QKVDataType, fp8_t> &&
std::is_same_v<typename Problem::GemmAccDataType, float>)
{
static_assert(WarpGemmM == 32);
// TODO: hard coded here. Otherwise, it may incorrect result
constexpr index_t swizzle_factor = 4;
return WarpGemmMfmaFp8Fp8F32M32N32K16SwizzleBTransposedCDistribution<
swizzle_factor>{};
} // TODO - bf8_t
}();
using BlockGemmPolicy =
BlockGemmARegBSmemCRegV2CustomPolicy<typename Problem::QKVDataType,
typename Problem::QKVDataType,
typename Problem::GemmAccDataType,
typename Problem::BlockFmhaShape::Gemm0BlockWarps,
decltype(warp_gemm)>;
if constexpr(1 < Problem::kNumGemm0Warps)
return BlockGemmARegBSmemCRegV2<GemmProblem, BlockGemmPolicy>{};
else
return BlockGemmARegBSmemCRegOneWarpV1<GemmProblem, BlockGemmPolicy>{};
}
// leave some exclusive space so that the second v_lds buffer will nenver overlap with the first
// k_lds bufffer
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetExclusiveKLdsBytes()
{
constexpr index_t single_k_lds_buffer_size =
GetSmemSizeK<Problem>() / GetNumKLdsBuffers<Problem>();
constexpr index_t single_v_lds_buffer_size =
GetSmemSizeV<Problem>() / GetNumVLdsBuffers<Problem>();
if constexpr(single_k_lds_buffer_size <= single_v_lds_buffer_size)
return 0;
else
return integer_least_multiple(single_k_lds_buffer_size - single_v_lds_buffer_size, 64);
};
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t IsFirstKLdsBufferOverlapLastVLdsBuffer()
{
using BlockFmhaShape = remove_cvref_t<typename Problem::BlockFmhaShape>;
constexpr index_t k1_loops = BlockFmhaShape::kN0 / BlockFmhaShape::kK1;
constexpr index_t num_k_lds_buffers = GetNumKLdsBuffers<Problem>();
constexpr index_t num_v_lds_buffers = GetNumVLdsBuffers<Problem>();
constexpr index_t last_v_lds_buffer_offset =
MakeVLdsBlockDescriptor<Problem>().get_element_space_size() / num_v_lds_buffers *
((k1_loops - 1) % num_v_lds_buffers) * sizeof(typename Problem::VDataType);
constexpr index_t first_k_lds_buffer_size =
MakeKLdsBlockDescriptor<Problem>().get_element_space_size() / num_k_lds_buffers *
sizeof(typename Problem::QKVDataType);
return GetExclusiveKLdsBytes<Problem>() + last_v_lds_buffer_offset <
first_k_lds_buffer_size;
};
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetSmemSizeK()
{
return MakeKLdsBlockDescriptor<Problem>().get_element_space_size() *
sizeof(typename Problem::QKVDataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetSmemSizeV()
{
return MakeVLdsBlockDescriptor<Problem>().get_element_space_size() *
sizeof(typename Problem::QKVDataType);
}
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr ck_tile::index_t GetSmemSize()
{
// assume V can reuse the other shared memory by K except the first
// assume Dropout can reuse the shared memory by V
return GetExclusiveKLdsBytes<Problem>() +
max(GetSmemSizeK<Problem>() - GetExclusiveKLdsBytes<Problem>(),
max(GetSmemSizeV<Problem>(), GetSmemSizeDropout<Problem>(0)));
}
};
} // namespace ck_tile

View File

@@ -0,0 +1,98 @@
/*
* Copyright (c) 2023-2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#pragma once
#include <ck_tile/core.hpp>
#include <ck_tile/ops/fmha.hpp>
#include "hstu_attention_fwd_type_config.hpp"
template <ck_tile::index_t MaxK>
struct HstuAttentionFwdBlockTile;
// Tile-sizes: M N0 K0 N1 K1 MaxK (MaxK % K0 == 0, MaxK % N1 == 0, N0 % K1 == 0)
//
template <>
struct HstuAttentionFwdBlockTile<32>
{
using type = ck_tile::sequence<64, 64, 16, 32, 32, 32>;
using gemm0_warps = ck_tile::sequence<2, 1, 1>;
using gemm1_warps = ck_tile::sequence<2, 1, 1>;
};
template <>
struct HstuAttentionFwdBlockTile<64>
{
using type = ck_tile::sequence<128, 64, 32, 64, 32, 64>;
using gemm0_warps = ck_tile::sequence<4, 1, 1>;
using gemm1_warps = ck_tile::sequence<4, 1, 1>;
};
template <>
struct HstuAttentionFwdBlockTile<128>
{
using type = ck_tile::sequence<128, 128, 32, 128, 32, 128>;
using gemm0_warps = ck_tile::sequence<4, 1, 1>;
using gemm1_warps = ck_tile::sequence<4, 1, 1>;
};
template <>
struct HstuAttentionFwdBlockTile<256>
{
using type = ck_tile::sequence<128, 128, 32, 256, 32, 256>;
using gemm0_warps = ck_tile::sequence<4, 1, 1>;
using gemm1_warps = ck_tile::sequence<4, 1, 1>;
};
using HstuAttentionFwdWarpTile1 = ck_tile::sequence<32, 32, 16>;
template <ck_tile::index_t MaxK>
struct HstuAttentionFwdShape;
template <>
struct HstuAttentionFwdShape<32>
{
using Type = ck_tile::TileFmhaShape<typename HstuAttentionFwdBlockTile<32>::type,
typename HstuAttentionFwdBlockTile<32>::gemm0_warps,
HstuAttentionFwdWarpTile1,
typename HstuAttentionFwdBlockTile<32>::gemm1_warps,
HstuAttentionFwdWarpTile1,
IsVLayoutRowMajor>;
};
template <>
struct HstuAttentionFwdShape<64>
{
using Type = ck_tile::TileFmhaShape<typename HstuAttentionFwdBlockTile<64>::type,
typename HstuAttentionFwdBlockTile<64>::gemm0_warps,
HstuAttentionFwdWarpTile1,
typename HstuAttentionFwdBlockTile<64>::gemm1_warps,
HstuAttentionFwdWarpTile1,
IsVLayoutRowMajor>;
};
template <>
struct HstuAttentionFwdShape<128>
{
using Type = ck_tile::TileFmhaShape<typename HstuAttentionFwdBlockTile<128>::type,
typename HstuAttentionFwdBlockTile<128>::gemm0_warps,
HstuAttentionFwdWarpTile1,
typename HstuAttentionFwdBlockTile<128>::gemm1_warps,
HstuAttentionFwdWarpTile1,
IsVLayoutRowMajor>;
};
template <>
struct HstuAttentionFwdShape<256>
{
using Type = ck_tile::TileFmhaShape<typename HstuAttentionFwdBlockTile<256>::type,
typename HstuAttentionFwdBlockTile<256>::gemm0_warps,
HstuAttentionFwdWarpTile1,
typename HstuAttentionFwdBlockTile<256>::gemm1_warps,
HstuAttentionFwdWarpTile1,
IsVLayoutRowMajor>;
};

View File

@@ -0,0 +1,34 @@
/*
* Copyright (c) 2023-2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#pragma once
#include <ck_tile/core.hpp>
template <typename InOutDataType>
struct HstuAttentionFwdTypeConfig;
template <>
struct HstuAttentionFwdTypeConfig<ck_tile::fp16_t>
{
using BiasDataType = ck_tile::fp16_t;
using GemmAccDataType = float;
using CompDataType = float; // data type for non-linear calculation
using OaccDataType = GemmAccDataType;
using ODataType = ck_tile::fp16_t;
};
template <>
struct HstuAttentionFwdTypeConfig<ck_tile::bf16_t>
{
using BiasDataType = ck_tile::bf16_t;
using GemmAccDataType = float;
using CompDataType = float; // data type for non-linear calculation
using OaccDataType = GemmAccDataType;
using ODataType = ck_tile::bf16_t;
};
static constexpr bool IsVLayoutRowMajor = true;

View File

@@ -0,0 +1,30 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include <ck_tile/core.hpp>
#include <stdexcept>
#define HDIM_SWITCH(HDIM_1, HDIM_2, CONST_NAME, ...) \
[&] { \
if(HDIM_1 <= 64 && HDIM_2 <= 64) \
{ \
constexpr ck_tile::index_t CONST_NAME = 64; \
__VA_ARGS__(); \
} \
else if(HDIM_1 <= 128 && HDIM_2 <= 128) \
{ \
constexpr ck_tile::index_t CONST_NAME = 128; \
__VA_ARGS__(); \
} \
else if(HDIM_1 <= 256 && HDIM_2 <= 256) \
{ \
constexpr ck_tile::index_t CONST_NAME = 256; \
__VA_ARGS__(); \
} \
else \
{ \
throw std::runtime_error("Head-dim sizes not supported!"); \
} \
}()

View File

@@ -0,0 +1,43 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <ck_tile/core.hpp>
#include <stdexcept>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_attention_jagged_forward_dispatch.hpp"
#include "instances/hstu_attention_jagged_forward_bf16_instances_ref.hpp"
void hstu_attention_jagged_forward_bf16(HstuAttentionFwdParams& param, hipStream_t stream)
{
const bool has_dropout = (param.p_drop > 0.0f);
const bool has_bias = (param.bias_ptr != nullptr);
const bool use_causal = param.use_causal;
BOOL_SWITCH_3(has_bias, kHasBias, has_dropout, kHasDropout, use_causal, kUseCausal, [&] {
HDIM_SWITCH(param.hdim_qk, param.hdim_v, MaxK, [&] {
if(param.window_size > 0)
{
run_jagged_forward_causal_local_bias_dropout_dispatch<ck_tile::bf16_t,
kUseCausal,
true,
kHasBias,
kHasDropout,
MaxK>(param, stream);
}
else
{
run_jagged_forward_causal_local_bias_dropout_dispatch<ck_tile::bf16_t,
kUseCausal,
false,
kHasBias,
kHasDropout,
MaxK>(param, stream);
};
});
});
};

View File

@@ -0,0 +1,144 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#pragma once
#include <ck_tile/core/numeric/integer.hpp>
#include <ck_tile/host/kernel_launch.hpp>
#include <ck_tile/host/stream_config.hpp>
#include <ck_tile/ops/epilogue.hpp>
#include <ck_tile/ops/fmha.hpp>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_fwd_type_config.hpp"
#include "hstu_attention_fwd_setting.hpp"
#include "hstu_attention_params.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_block_masking.hpp"
#include "hstu_attention_pipeline_problem.hpp"
#include "hstu_attention_traits.hpp"
#include "hstu_attention_fwd_pipeline.hpp"
#include "hstu_attention_fwd_kernel.hpp"
template <typename InOutDataType,
bool kUseCausal,
bool kUseLocal,
bool kHasBias,
bool kHasDropout,
ck_tile::index_t MaxK>
struct jagged_forward_causal_local_bias_dropout_dispatch
{
using HstuAttentionShape = typename HstuAttentionFwdShape<MaxK>::Type;
using HstuMask = ck_tile::HstuBlockMasking<kUseCausal, kUseLocal>;
template <typename HstuTraits>
using HstuPipelineProblemTemp = ck_tile::HstuAttentionFwdPipelineProblem<
InOutDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::GemmAccDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::CompDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::BiasDataType,
true, // kIsJagged
kHasBias,
kHasDropout,
HstuMask,
HstuAttentionShape,
HstuTraits>;
static void Run(HstuAttentionFwdParams& param, hipStream_t stream)
{
constexpr ck_tile::index_t occupancy = -1;
const bool pad_headdim_qk = !(param.hdim_qk % HstuAttentionShape::kSubQKHeaddim == 0);
const bool pad_headdim_v = !(param.hdim_v % HstuAttentionShape::kN1 == 0);
// no need to check seqlen_q since it is not used as fastest dim,
// buffer_load_dwordxx/buffer_store_dwordxx can handle oob access
constexpr bool kPadSeqLenQ = false;
constexpr bool kPadSeqLenK = true;
BOOL_SWITCH_2(pad_headdim_qk, kPadHeadDimQK, pad_headdim_v, kPadHeadDimV, [&] {
using HstuTraits = ck_tile::HstuAttentionFwdTraits<kPadSeqLenQ,
kPadSeqLenK,
kPadHeadDimQK,
kPadHeadDimV,
occupancy>;
using HstuPipelineProblem = HstuPipelineProblemTemp<HstuTraits>;
using HstuEpilogue = ck_tile::Default2DEpilogue<ck_tile::Default2DEpilogueProblem<
typename HstuAttentionFwdTypeConfig<InOutDataType>::OaccDataType,
typename HstuAttentionFwdTypeConfig<InOutDataType>::ODataType,
kPadSeqLenQ,
kPadHeadDimV>>;
using HstuPipeline = ck_tile::HstuAttentionFwdPipelineQRKSVS<HstuPipelineProblem>;
using HstuKernel = ck_tile::HstuAttentionFwdKernel<HstuPipeline, HstuEpilogue>;
RunWithKernel<HstuKernel>(param, stream);
});
};
template <typename HstuKernel>
static void RunWithKernel(HstuAttentionFwdParams& param, hipStream_t stream)
{
const auto kargs = [&] {
return HstuKernel::MakeKargs(param.q_ptr,
param.k_ptr,
param.v_ptr,
param.bias_ptr,
param.o_ptr,
param.seq_offsets_ptr,
param.hdim_qk,
param.hdim_v,
param.num_head,
param.scale_s,
param.seq_stride_q,
param.seq_stride_k,
param.seq_stride_v,
param.seq_stride_bias,
param.seq_stride_o,
param.nhead_stride_q,
param.nhead_stride_k,
param.nhead_stride_v,
param.nhead_stride_bias,
param.nhead_stride_o,
param.num_targets_ptr,
param.window_size,
param.contextual_seqlen,
param.min_full_attn_seqlen,
param.p_drop,
param.philox_seed,
param.philox_offset);
}();
dim3 kGridSize =
HstuKernel::GridSize(param.num_batch, param.num_head, param.seqlen, param.hdim_v);
constexpr dim3 kBlockSize = HstuKernel::BlockSize();
constexpr ck_tile::index_t kBlockPerCu = HstuKernel::kBlockPerCu;
(void)ck_tile::launch_kernel(ck_tile::stream_config{stream, false},
ck_tile::make_kernel<kBlockSize.x, kBlockPerCu>(
HstuKernel{}, kGridSize, kBlockSize, 0, kargs));
};
};
template <typename InOutDataType,
bool kUseCausal,
bool kUseLocal,
bool kHasBias,
bool kHasDropout,
ck_tile::index_t MaxK>
void run_jagged_forward_causal_local_bias_dropout_dispatch(HstuAttentionFwdParams& param,
hipStream_t stream)
{
jagged_forward_causal_local_bias_dropout_dispatch<InOutDataType,
kUseCausal,
kUseLocal,
kHasBias,
kHasDropout,
MaxK>::Run(param, stream);
};

View File

@@ -0,0 +1,43 @@
/*
* Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <ck_tile/core.hpp>
#include <stdexcept>
#include "hstu_attention_bool_switch.hpp"
#include "hstu_attention_hdim_switch.hpp"
#include "hstu_attention_jagged_forward_dispatch.hpp"
#include "instances/hstu_attention_jagged_forward_fp16_instances_ref.hpp"
void hstu_attention_jagged_forward_fp16(HstuAttentionFwdParams& param, hipStream_t stream)
{
const bool has_dropout = (param.p_drop > 0.0f);
const bool has_bias = (param.bias_ptr != nullptr);
const bool use_causal = param.use_causal;
BOOL_SWITCH_3(has_bias, kHasBias, has_dropout, kHasDropout, use_causal, kUseCausal, [&] {
HDIM_SWITCH(param.hdim_qk, param.hdim_v, MaxK, [&] {
if(param.window_size > 0)
{
run_jagged_forward_causal_local_bias_dropout_dispatch<ck_tile::fp16_t,
kUseCausal,
true,
kHasBias,
kHasDropout,
MaxK>(param, stream);
}
else
{
run_jagged_forward_causal_local_bias_dropout_dispatch<ck_tile::fp16_t,
kUseCausal,
false,
kHasBias,
kHasDropout,
MaxK>(param, stream);
};
});
});
};

View File

@@ -0,0 +1,57 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include <ck_tile/core.hpp>
struct HstuAttentionFwdParams
{
bool is_jagged;
ck_tile::index_t num_batch;
ck_tile::index_t seqlen; // batched mode only
const void* seq_offsets_ptr; // jagged mode only
ck_tile::index_t max_seqlen; // jagged mode only
const void* q_ptr;
const void* k_ptr;
const void* v_ptr;
const void* bias_ptr;
void* o_ptr;
ck_tile::index_t hdim_qk;
ck_tile::index_t hdim_v;
ck_tile::index_t num_head;
float scale_s;
ck_tile::index_t seq_stride_q;
ck_tile::index_t seq_stride_k;
ck_tile::index_t seq_stride_v;
ck_tile::index_t seq_stride_bias;
ck_tile::index_t seq_stride_o;
ck_tile::index_t nhead_stride_q;
ck_tile::index_t nhead_stride_k;
ck_tile::index_t nhead_stride_v;
ck_tile::index_t nhead_stride_bias;
ck_tile::index_t nhead_stride_o;
// batched mode only parameters
ck_tile::index_t batch_stride_q;
ck_tile::index_t batch_stride_k;
ck_tile::index_t batch_stride_v;
ck_tile::index_t batch_stride_bias;
ck_tile::index_t batch_stride_o;
const void* num_targets_ptr;
bool use_causal;
ck_tile::index_t window_size;
ck_tile::index_t contextual_seqlen;
ck_tile::index_t min_full_attn_seqlen;
float p_drop;
uint64_t philox_seed;
uint64_t philox_offset;
};

View File

@@ -0,0 +1,62 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include <ck_tile/core.hpp>
namespace ck_tile {
// PipelineProblem encodes information not only from the original user-problem,
// but it also contains other information needed by the pipeline, which includes
// TileShape -- which determines how block-layer calculation is done in tiles and
// how warps are allocated on dimensions
// Traits -- other information required for running the kernel and pipeline
template <typename InOutDataType_,
typename GemmAccDataType_,
typename CompDataType_, // data type for SiLU and other non-linear calculation
typename BiasDataType_,
bool kIsJagged_,
bool kHasBias_,
bool kHasDropout_,
typename HstuMask_, // encoding Causal and Local, contextual masking
typename AttentionTileShape_,
typename Traits_>
struct HstuAttentionFwdPipelineProblem
{
using InOutDataType = remove_cvref_t<InOutDataType_>;
using QKVDataType = InOutDataType;
using ODataType = InOutDataType;
using GemmAccDataType = remove_cvref_t<GemmAccDataType_>;
// DataType used when siLU calculation
using CompDataType = remove_cvref_t<CompDataType_>;
using BiasDataType = remove_cvref_t<BiasDataType_>;
// to be compatible with ck_tile existing policy codes
using QDataType = QKVDataType;
using KDataType = QKVDataType;
using VDataType = QKVDataType;
using SaccDataType = GemmAccDataType;
using OaccDataType = GemmAccDataType;
using PDataType = QKVDataType;
static constexpr bool kIsJagged = kIsJagged_;
static constexpr bool kHasBias = kHasBias_;
static constexpr bool kHasDropout = kHasDropout_;
using HstuMask = remove_cvref_t<HstuMask_>;
using HstuAttentionTileShape = remove_cvref_t<AttentionTileShape_>;
// Keep the name compatible with ck_tile existing policy codes, to be changed
using BlockFmhaShape = HstuAttentionTileShape;
using Traits = remove_cvref_t<Traits_>;
static constexpr index_t kNumGemm0Warps = AttentionTileShape_::NumGemm0Warps;
static constexpr index_t kNumGemm1Warps = AttentionTileShape_::NumGemm1Warps;
static constexpr index_t kBlockSize = AttentionTileShape_::NumWarps * get_warp_size();
};
} // namespace ck_tile

View File

@@ -1,24 +0,0 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include <ck_tile/core.hpp>
// Type configuration
template <typename DataType>
struct HSTUAttentionTypeConfig;
template <>
struct HSTUAttentionTypeConfig<ck_tile::fp16_t>
{
using GemmAccDataType = float;
using SMComputeDataType = float;
};
template <>
struct HSTUAttentionTypeConfig<ck_tile::bf16_t>
{
using GemmAccDataType = float;
using SMComputeDataType = float;
};

View File

@@ -0,0 +1,25 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
namespace ck_tile {
template <bool kPadSeqLenQ_,
bool kPadSeqLenK_,
bool kPadHeadDimQK_,
bool kPadHeadDimV_,
index_t kBlockPerCu_>
struct HstuAttentionFwdTraits
{
static constexpr bool kPadSeqLenQ = kPadSeqLenQ_;
static constexpr bool kPadSeqLenK = kPadSeqLenK_;
static constexpr bool kPadHeadDimQK = kPadHeadDimQK_;
static constexpr bool kPadHeadDimV = kPadHeadDimV_;
static constexpr index_t kBlockPerCu = kBlockPerCu_;
};
} // namespace ck_tile

View File

@@ -0,0 +1,109 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck_tile/core.hpp"
namespace ck_tile {
template <bool kUseCausal, bool kUseLocal>
struct HstuBlockMasking
{
static constexpr bool IsMasking = (kUseCausal || kUseLocal);
int max_attn_len;
int contextual_seqlen;
int min_full_attn_seqlen;
int max_uih_len;
CK_TILE_HOST_DEVICE HstuBlockMasking(int max_attn_len_,
int contextual_seqlen_,
int min_full_attn_seqlen_,
int max_uih_len_)
{
max_attn_len = max_attn_len_;
contextual_seqlen = contextual_seqlen_;
min_full_attn_seqlen = min_full_attn_seqlen_;
max_uih_len = max_uih_len_;
};
// to get the loop length along X axis, return index:[start, end), end-start=length
// use this if need loop over X axis tile by tile (eg. seqlen_k loop-over)
template <index_t YTile, index_t XTile>
CK_TILE_HOST_DEVICE constexpr auto
GetTileRangeAlongX(index_t i_y, number<YTile>, number<XTile>) const
{
if constexpr(!IsMasking)
{
return ck_tile::make_tuple(0, max_uih_len);
}
else
{
if(contextual_seqlen > 0 && (i_y < contextual_seqlen))
return ck_tile::make_tuple(0, max_uih_len);
if constexpr(kUseCausal && !kUseLocal)
{
index_t x_end =
min(i_y + YTile, max_uih_len); // for lower-triangular masking, x <= y
return ck_tile::make_tuple(0, x_end);
}
else if constexpr(!kUseCausal && kUseLocal)
{
if(min_full_attn_seqlen > 0 && i_y + YTile > max_uih_len - min_full_attn_seqlen)
{
return ck_tile::make_tuple(0, max_uih_len);
}
else
{
index_t x_start = max(0, i_y - max_attn_len);
index_t x_end = i_y + YTile + max_attn_len;
return ck_tile::make_tuple(x_start - x_start % XTile, x_end);
};
}
else // kUseCausal && kUseLocal
{
if(min_full_attn_seqlen > 0 && i_y + YTile > max_uih_len - min_full_attn_seqlen)
{
return ck_tile::make_tuple(0, max_uih_len);
}
else
{
index_t x_end = i_y + YTile + max_attn_len;
return ck_tile::make_tuple(0, x_end);
};
};
};
}
CK_TILE_HOST_DEVICE constexpr bool IsTokenPairInsideMask(int row, int col)
{
if(row < contextual_seqlen)
return true;
bool result = false;
if constexpr(kUseLocal)
{
if constexpr(kUseCausal)
result = (row >= col) && (row - col <= max_attn_len);
else
result = std::abs(row - col) <= max_attn_len;
if(min_full_attn_seqlen > 0)
result = result || (row >= max_uih_len - min_full_attn_seqlen);
}
else
{
if constexpr(kUseCausal)
result = (row >= col);
};
return result;
};
};
} // namespace ck_tile

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,206 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,206 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_batched_forward_dispatch.hpp"
template void run_batched_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,206 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
true,
true,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);
extern template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
true,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/half.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::bf16_t,
false,
false,
false,
false,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
256>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
true,
64>(HstuAttentionFwdParams& param, hipStream_t stream);

View File

@@ -0,0 +1,22 @@
/*
Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree.
*
* The file is automatically generated, don't modify!
* See the generator script
* `composable_kernel/example/ck_tile/18_hstu_attention/generate_instances.py`
*/
#include <ck_tile/core/numeric/bfloat16.hpp>
#include "hstu_attention_jagged_forward_dispatch.hpp"
template void run_jagged_forward_causal_local_bias_dropout_dispatch<
ck_tile::fp16_t,
true,
true,
true,
false,
128>(HstuAttentionFwdParams& param, hipStream_t stream);

Some files were not shown because too many files have changed in this diff Show More