mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
36 lines
1.4 KiB
C++
36 lines
1.4 KiB
C++
/*
|
|
* Copyright (c) 2019-2025, NVIDIA CORPORATION. All rights reserved.
|
|
* Copyright (c) 2021, NAVER Corp. Authored by CLOVA.
|
|
*
|
|
* 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 <cuda_bf16.h>
|
|
#include <cuda_fp16.h>
|
|
|
|
#include "tensorrt_llm/common/cudaUtils.h"
|
|
|
|
namespace tensorrt_llm::kernels
|
|
{
|
|
void invokeIndexerTopKDecode(float const* logits, int const* seqLens, int* indices, float* outLogitsAux,
|
|
int* outIndicesAux, int const splitWorkThreshold, int const numRows, int const numColumns, int const stride0,
|
|
int const stride1, int const next_n, int const topK = 2048, cudaStream_t const stream = 0);
|
|
|
|
void invokeIndexerTopKPrefill(float const* logits, int const* rowStarts, int const* rowEnds, int* indices,
|
|
int const numRows, int const numColumns, int const stride0, int const stride1, int const topK = 2048,
|
|
cudaStream_t const stream = 0);
|
|
|
|
} // namespace tensorrt_llm::kernels
|