mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
51 lines
1.6 KiB
C++
51 lines
1.6 KiB
C++
/*
|
|
* SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
|
* SPDX-License-Identifier: Apache-2.0
|
|
*
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
* you may not use this file except in compliance with the License.
|
|
* You may obtain a copy of the License at
|
|
*
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
*
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
* See the License for the specific language governing permissions and
|
|
* limitations under the License.
|
|
*/
|
|
#pragma once
|
|
|
|
#include "tensorrt_llm/common/config.h"
|
|
#include <stdint.h>
|
|
|
|
#include <cuda_fp16.h>
|
|
#include <cuda_fp8.h>
|
|
#include <cuda_runtime.h>
|
|
#include <cuda_runtime_api.h>
|
|
|
|
#if defined(ENABLE_BF16)
|
|
#include <cuda_bf16.h>
|
|
#endif
|
|
|
|
#include <type_traits>
|
|
#include <vector>
|
|
|
|
TRTLLM_NAMESPACE_BEGIN
|
|
|
|
namespace kernels
|
|
{
|
|
|
|
template <typename T_in, typename T_out = T_in>
|
|
void apply_per_channel_scale_kernel_launcher(T_out* smoothed_act, T_in const* act, T_in const* per_channel_scale,
|
|
int rows, int cols, int64_t const* num_valid_tokens_ptr = nullptr, cudaStream_t stream = 0);
|
|
|
|
template <typename T_in, typename T_out = T_in>
|
|
void apply_per_channel_scale_per_expert_kernel_launcher(T_out* smoothed_act, T_in const* act,
|
|
T_in const* per_channel_scale, int rows, int cols, int64_t* expert_first_token_offset,
|
|
int const num_experts_per_node, int64_t const* num_valid_tokens_ptr, cudaStream_t stream);
|
|
|
|
} // namespace kernels
|
|
|
|
TRTLLM_NAMESPACE_END
|