mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
4bb65f216f
* Update TensorRT-LLM --------- Co-authored-by: meghagarwal <16129366+megha95@users.noreply.github.com> Co-authored-by: Shixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>
49 lines
1.6 KiB
C++
49 lines
1.6 KiB
C++
/*
|
|
* Copyright (c) 2020-2023, NVIDIA CORPORATION. All rights reserved.
|
|
*
|
|
* 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/kernels/decoderMaskedMultiheadAttention/decoderXQAImpl.h"
|
|
|
|
namespace tensorrt_llm
|
|
{
|
|
namespace kernels
|
|
{
|
|
|
|
class DecoderXQAImplPrecompiled : public DecoderXQAImpl
|
|
{
|
|
public:
|
|
DecoderXQAImplPrecompiled(DecoderXQARunner* runner)
|
|
: DecoderXQAImpl(runner)
|
|
{
|
|
}
|
|
|
|
bool shouldUse(XQAParams const& xqaParams) override;
|
|
void prepare(XQAParams const& xqa_params) override;
|
|
|
|
protected:
|
|
void runWithKVLinearBuffer(XQAParams const& xqa_params, KVLinearBuffer& kv_linear_buffer,
|
|
int2& rotary_kernel_launch_cache, cudaStream_t const& stream) override;
|
|
void runWithKVBlockArray(XQAParams const& xqa_params, KVBlockArray& kv_block_array,
|
|
int2& rotary_kernel_launch_cache, cudaStream_t const& stream) override;
|
|
|
|
private:
|
|
template <typename KVCacheBuffer>
|
|
void runDispatchBuffer(XQAParams const& xqa_params, KVCacheBuffer& kv_cache_buffer,
|
|
int2& rotary_kernel_launch_cache, cudaStream_t const& stream);
|
|
};
|
|
|
|
} // namespace kernels
|
|
} // namespace tensorrt_llm
|