TensorRT-LLMs/cpp/tensorrt_llm/kernels/mlaKernels.h
Yihan Wang 9df4dad3b6
[None][fix] Introduce inline namespace to avoid symbol collision (#9541)
Signed-off-by: Yihan Wang <yihwang@nvidia.com>
2025-12-12 23:32:15 +08:00

139 lines
5.0 KiB
C++

/*
* Copyright (c) 2019-2023, 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/common/config.h"
#include "tensorrt_llm/common/cudaUtils.h"
#include "tensorrt_llm/kernels/kvCacheUtils.h"
#include "tensorrt_llm/kernels/unfusedAttentionKernels.h"
#include <assert.h>
#include <cstdint>
#include <cuda_fp16.h>
#include <cuda_runtime.h>
TRTLLM_NAMESPACE_BEGIN
namespace kernels
{
enum class KvCacheDataType;
struct MlaMetaParams
{
int32_t q_lora_rank = 0;
int32_t kv_lora_rank = 0;
int32_t qk_nope_head_dim = 0;
int32_t qk_rope_head_dim = 0;
int32_t v_head_dim = 0;
int32_t predicted_tokens_per_seq = 1;
int32_t num_layers = 0;
auto data() const
{
return std::make_tuple(q_lora_rank, kv_lora_rank, qk_nope_head_dim, qk_rope_head_dim, v_head_dim,
predicted_tokens_per_seq, num_layers);
}
};
template <typename T>
struct MlaParams
{
T const* latent_cache; // cKV + k_pe
// Tensor Q for both context and generation MLA, contiguous. Pre-process kernel will apply RoPE and modify it
// in-place. For context MLA, shape: [total_q_len, h * (d_nope + d_rope)], stride: [h * (d_nope + d_rope), 1]
T* q_buf;
// Separate tensor K for context MLA, contiguous. Pre-process kernel will apply RoPE and modify it in-place.
// shape: [total_kv_len, h * (d_nope + d_rope)], stride: [h * (d_nope + d_rope), 1]
T* k_buf = nullptr;
// Separate tensor V for context MLA, NOT contiguous,
// shape: [total_kv_len, h * d_v], stride: [h * (d_nope + d_v), 1]
T const* v_buf = nullptr;
// Tensor quantized Q for both context and generation MLA.
// For context MLA, shape: [total_q_len, h * (d_nope + d_rope)], stride: [h * (d_nope + d_rope), 1]
void* quant_q_buf = nullptr;
// Tensor quantized K for context MLA, contiguous
// shape: [total_kv_len, h * (d_nope + d_rope)], stride: [h * (d_nope + d_rope), 1]
void* quant_k_buf = nullptr;
// Tensor quantized V for context MLA, contiguous
// shape: [total_kv_len, h * d_v], stride: [h * d_v, 1]
void* quant_v_buf = nullptr;
T* context_buf;
T* q_pe; // [b, h, d_r], strided
float2 const* cos_sin_cache; // [s, rope]
int32_t batch_size;
int32_t acc_q_len;
int32_t head_num; // h
void* workspace;
int32_t const* cache_seq_lens;
int* seqQOffset;
uint32_t* fmha_tile_counter;
int32_t max_input_seq_len;
int* cu_q_seqlens;
int* cu_kv_seqlens;
int32_t q_pe_ld;
int32_t q_pe_stride;
MlaMetaParams meta;
int const* block_ids_per_seq;
KvCacheDataType cache_type;
// Scales for mla quantization
float* bmm1_scale;
float* bmm2_scale;
float const* quant_scale_o;
float const* quant_scale_q;
float const* quant_scale_kv;
float const* dequant_scale_q;
float const* dequant_scale_kv;
float host_bmm1_scale;
// Is it absorption mode?
bool absorption_mode = false;
// For FP8 context qkv quantization
float const* quant_scale_qkv = nullptr;
// for Helix parallelism: the rotary position offsets [b]
int32_t const* helix_position_offsets{nullptr};
// for Helix parallelism: whether the current rank is inactive, shape [b]
// (the current query tokens are not appended to this rank's KV cache)
bool const* helix_is_inactive_rank{nullptr};
};
template <typename T, typename KVCacheBuffer>
void invokeMLARopeContext(MlaParams<T>& params, KVCacheBuffer kv_cache_buffer, cudaStream_t stream);
template <typename T>
void invokeMLAContextFp8Quantize(MlaParams<T>& params, int total_kv_len, cudaStream_t stream);
template <typename T, typename KVCacheBuffer>
void invokeMLARopeGeneration(MlaParams<T>& params, KVCacheBuffer kv_cache_buffer, cudaStream_t stream);
template <typename T, typename TCache>
void invokeMLALoadPagedKV(T* compressed_kv_ptr, T* k_pe_ptr, KVBlockArray& kv_cache, int const num_contexts,
int64_t const* cu_ctx_cached_kv_lens, int const max_input_seq_len, int const lora_size, int const rope_size,
float const* kv_scale_quant_orig_ptr, cudaStream_t stream);
template <typename T, typename TCache>
void invokeMLARopeAppendPagedKVAssignQ(KVBlockArray& kv_cache, T* q_ptr, T* latent_cache_ptr, int const num_requests,
int64_t const* cu_ctx_cached_kv_lens, int64_t const* cu_seq_lens, int const max_input_uncached_seq_len,
float2 const* cos_sin_cache, size_t head_num, int nope_size, int rope_size, int lora_size,
float const* kv_scale_orig_quant_ptr, cudaStream_t stream);
} // namespace kernels
TRTLLM_NAMESPACE_END