mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
728cc0044b
* Update TensorRT-LLM --------- Co-authored-by: Morgan Funtowicz <funtowiczmo@gmail.com> Co-authored-by: Shixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>
465 lines
17 KiB
Python
465 lines
17 KiB
Python
# 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.
|
|
from collections import defaultdict
|
|
from typing import List
|
|
|
|
import torch
|
|
|
|
|
|
class Block(object):
|
|
|
|
def __init__(self, block_idx, k_ptrs, v_ptrs):
|
|
self.idx = block_idx
|
|
self.k_ptrs = k_ptrs
|
|
self.v_ptrs = v_ptrs
|
|
self.ref_count = 0
|
|
|
|
def add_link(self):
|
|
self.ref_count += 1
|
|
|
|
def remove_link(self):
|
|
self.ref_count -= 1
|
|
|
|
def has_link(self) -> bool:
|
|
return self.ref_count > 0
|
|
|
|
def is_shared(self) -> bool:
|
|
return self.ref_count > 1
|
|
|
|
def get_k_ptr(self, idx) -> int:
|
|
return self.k_ptrs[idx]
|
|
|
|
def get_v_ptr(self, idx) -> int:
|
|
return self.v_ptrs[idx]
|
|
|
|
|
|
class GenerationSequence(object):
|
|
|
|
def __init__(self, seq_idx, batch_idx):
|
|
self.seq_idx = seq_idx
|
|
self.batch_idx = batch_idx
|
|
|
|
def get_batch_idx(self) -> int:
|
|
"""
|
|
Returns idx of sequence in batch
|
|
"""
|
|
return self.batch_idx
|
|
|
|
def get_seq_idx(self) -> int:
|
|
"""
|
|
Returns sequence idx
|
|
"""
|
|
return self.seq_idx
|
|
|
|
def __eq__(self, another):
|
|
return hasattr(another, 'seq_idx') and self.seq_idx == another.seq_idx and \
|
|
hasattr(another, 'batch_idx') and self.batch_idx == another.batch_idx
|
|
|
|
def __hash__(self):
|
|
return self.seq_idx
|
|
|
|
|
|
class BlocksManager(object):
|
|
_sizeof = {
|
|
torch.float32: 4,
|
|
torch.float16: 2,
|
|
torch.bfloat16: 2,
|
|
torch.int8: 1
|
|
}
|
|
|
|
def __init__(self,
|
|
memory_pools: List[torch.Tensor],
|
|
blocks: int,
|
|
max_blocks_per_seq: int = 128,
|
|
beam_width: int = 1):
|
|
self.max_blocks_per_seq = max_blocks_per_seq
|
|
|
|
self.pointer_array = None
|
|
self.memory_pools = memory_pools
|
|
self.blocks = blocks
|
|
self.beam_width = beam_width
|
|
|
|
self.elts_per_blocks = []
|
|
for pool in memory_pools:
|
|
# Pool consists of memory for K and V caches
|
|
self.elts_per_blocks.append(pool.nelement() // (2 * blocks))
|
|
|
|
self.free_blocks = []
|
|
for bi in range(blocks):
|
|
k_ptrs = []
|
|
v_ptrs = []
|
|
for pool, elts_per_block in zip(memory_pools, self.elts_per_blocks):
|
|
k_ptrs.append(self.get_mempool_pointer(bi, pool,
|
|
elts_per_block))
|
|
v_ptrs.append(
|
|
self.get_mempool_pointer(bi, pool, elts_per_block) +
|
|
self.blocks * elts_per_block * self._sizeof[pool.dtype])
|
|
self.free_blocks.append(Block(bi, k_ptrs, v_ptrs))
|
|
|
|
beam_width = self.beam_width
|
|
# Here use beam_width instead of self.beam_width to remove cyclic reference between self and
|
|
# self.allocated_blocks by preventing capture self, which may cause memory leak.
|
|
self.allocated_blocks = defaultdict(
|
|
lambda: [[] for _ in range(beam_width)])
|
|
|
|
def has_free_block(self) -> bool:
|
|
"""
|
|
Returns True if we have at least 1 free block
|
|
"""
|
|
return len(self.free_blocks) > 0
|
|
|
|
def allocate(self,
|
|
owner: GenerationSequence,
|
|
share_across_beam: bool = False):
|
|
"""
|
|
Add block to owner and increase ref count
|
|
"""
|
|
# Add blocks for whole beam width
|
|
block = None
|
|
for bi in range(self.beam_width):
|
|
if not self.has_free_block():
|
|
raise RuntimeError("Can't allocate new block for KV cache")
|
|
|
|
# Use the same block for all seqs in beam if share_across_beam
|
|
if block is None or share_across_beam == False:
|
|
block = self.free_blocks.pop(0)
|
|
# Add one reference to the block
|
|
block.add_link()
|
|
self.allocated_blocks[owner][bi].append(block)
|
|
|
|
def replace_shared_block(self, owner: GenerationSequence, block_idx: int):
|
|
"""
|
|
Replace the shared block.
|
|
Free the shared block, and allocate blocks with share_across_beam=False
|
|
"""
|
|
if not self.allocated_blocks[owner][0][block_idx].is_shared():
|
|
return
|
|
|
|
# Free shared block
|
|
for bi in range(self.beam_width):
|
|
block = self.allocated_blocks[owner][bi][block_idx]
|
|
block.remove_link()
|
|
if not block.has_link():
|
|
self.free_blocks.append(block)
|
|
|
|
# Allocate new block
|
|
for bi in range(self.beam_width):
|
|
if not self.has_free_block():
|
|
raise RuntimeError("Can't allocate new block for KV cache")
|
|
block = self.free_blocks.pop(0)
|
|
block.add_link()
|
|
self.allocated_blocks[owner][bi][block_idx] = block
|
|
return
|
|
|
|
def free(self, owner: GenerationSequence):
|
|
"""
|
|
Unlink all blocks of given owner.
|
|
Moves blocks with ref_count == 0 to free.
|
|
Removes owner from allocated blocks.
|
|
"""
|
|
for bi in range(self.beam_width):
|
|
for block in self.allocated_blocks[owner][bi]:
|
|
# Move block to free if no one refers to it
|
|
block.remove_link()
|
|
|
|
# Move block to free if no one refers to it
|
|
if not block.has_link():
|
|
self.free_blocks.append(block)
|
|
# Remove owner from allocated blocks
|
|
self.allocated_blocks.pop(owner)
|
|
|
|
def get_number_blocks(self, owner: GenerationSequence) -> int:
|
|
"""
|
|
Returns number of blocks allocated to the sequence owner
|
|
"""
|
|
return len(self.allocated_blocks[owner][0])
|
|
|
|
def get_mempool_pointer(self, block_idx: int, pool: torch.Tensor,
|
|
elts_per_block: int) -> int:
|
|
"""
|
|
Computes linear pointer
|
|
"""
|
|
return pool.data_ptr(
|
|
) + block_idx * elts_per_block * self._sizeof[pool.dtype]
|
|
|
|
def get_pointer_array(self, pool_idx: int, beam_width: int) -> torch.Tensor:
|
|
"""
|
|
Returns array of [batch size, beam_width, 2, max_blocks_per_seq] of pointers
|
|
to the allocated blocks in memory pool
|
|
"""
|
|
assert (beam_width <= self.beam_width)
|
|
|
|
def create_nested_list(dims):
|
|
"""Recursive function to generate nested list."""
|
|
if len(dims) == 1:
|
|
return [0 for _ in range(dims[0])]
|
|
return [create_nested_list(dims[1:]) for _ in range(dims[0])]
|
|
|
|
pointer_array = create_nested_list(
|
|
(len(self.allocated_blocks), beam_width, 2,
|
|
self.max_blocks_per_seq))
|
|
|
|
for owner, beams_blocks in self.allocated_blocks.items():
|
|
for bi in range(beam_width):
|
|
for block_linear_idx, block in enumerate(beams_blocks[bi]):
|
|
# K cache pointers
|
|
pointer_array[owner.get_batch_idx(
|
|
)][bi][0][block_linear_idx] = block.get_k_ptr(pool_idx)
|
|
# V cache pointers
|
|
pointer_array[owner.get_batch_idx(
|
|
)][bi][1][block_linear_idx] = block.get_v_ptr(pool_idx)
|
|
|
|
self.pointer_array = torch.tensor(pointer_array, dtype=torch.int64)
|
|
return self.pointer_array
|
|
|
|
def get_continuous_caches(self, pool_idx: int) -> torch.Tensor:
|
|
"""
|
|
Returns continuous KV caches.
|
|
Used only for debug purposes.
|
|
"""
|
|
assert self.beam_width == 1
|
|
|
|
elts_per_block = self.elts_per_blocks[pool_idx]
|
|
pool = self.memory_pools[pool_idx].flatten()
|
|
continuous_kv_cache = torch.zeros(len(self.allocated_blocks),
|
|
2,
|
|
self.max_blocks_per_seq *
|
|
elts_per_block,
|
|
dtype=pool.dtype,
|
|
device="cuda")
|
|
for owner, beam_blocks in self.allocated_blocks.items():
|
|
for bi in range(self.beam_width):
|
|
for block_linear_idx, block in enumerate(beam_blocks[bi]):
|
|
# The batch index.
|
|
batch_idx = owner.get_batch_idx()
|
|
# The first index in the sequence.
|
|
block_offset = block_linear_idx * elts_per_block
|
|
# The first index in the pool for K.
|
|
k_start = block.idx * elts_per_block
|
|
# The first index in the pool for V.
|
|
v_start = k_start + self.blocks * elts_per_block
|
|
|
|
continuous_kv_cache[batch_idx][0][
|
|
block_offset:block_offset +
|
|
elts_per_block] = pool[k_start:k_start + elts_per_block]
|
|
continuous_kv_cache[batch_idx][1][
|
|
block_offset:block_offset +
|
|
elts_per_block] = pool[v_start:v_start + elts_per_block]
|
|
|
|
return continuous_kv_cache
|
|
|
|
|
|
class KVCacheManager(object):
|
|
|
|
def __init__(self,
|
|
memory_pools: List[torch.Tensor],
|
|
blocks: int,
|
|
tokens_per_block: int,
|
|
max_blocks_per_seq: int,
|
|
max_attention_window_size: int,
|
|
sink_token_len: int,
|
|
beam_width: int = 1,
|
|
use_one_more_block: bool = False):
|
|
|
|
self.blocks_manager = BlocksManager(
|
|
memory_pools=memory_pools,
|
|
blocks=blocks,
|
|
max_blocks_per_seq=max_blocks_per_seq,
|
|
beam_width=beam_width)
|
|
self.num_pools = len(memory_pools)
|
|
self.tokens_per_block = tokens_per_block
|
|
self.max_attention_window_size = max_attention_window_size
|
|
self.sink_token_len = sink_token_len
|
|
self.beam_width = beam_width
|
|
|
|
# The sink tokens are not stored into the same block with other tokens.
|
|
# Need to add the bubble after the sink tokens.
|
|
if sink_token_len % tokens_per_block == 0:
|
|
self.bubble_len = 0
|
|
else:
|
|
self.bubble_len = tokens_per_block - sink_token_len % tokens_per_block
|
|
|
|
# Token num in the sink blocks
|
|
self.sink_block_token_num = self.sink_token_len + self.bubble_len
|
|
|
|
# Max token num in the cache
|
|
self.max_token_num = self.max_attention_window_size + self.bubble_len
|
|
if use_one_more_block:
|
|
self.max_token_num += self.tokens_per_block
|
|
|
|
self.lens = []
|
|
self.sequences = []
|
|
|
|
def step(self, finished: List[bool]):
|
|
"""
|
|
Iterate to the next generation step.
|
|
Add new blocks where needed and clear finished sequences.
|
|
"""
|
|
for seq in self.sequences:
|
|
batch_idx = seq.get_batch_idx()
|
|
# Enable cyclic kv cache when it exceeds the max_token_num
|
|
cyclic_token_num = self.max_token_num - self.sink_block_token_num
|
|
next_token_idx_in_cache = self.sink_block_token_num + \
|
|
(self.lens[batch_idx] - self.sink_block_token_num) % cyclic_token_num
|
|
if not finished[batch_idx] and (
|
|
next_token_idx_in_cache % self.tokens_per_block == 0 or
|
|
(next_token_idx_in_cache - self.sink_block_token_num) %
|
|
cyclic_token_num == 0):
|
|
if self.lens[batch_idx] < self.max_token_num:
|
|
self.blocks_manager.allocate(seq)
|
|
elif self.beam_width > 1:
|
|
# Get next block index
|
|
next_block_idx = next_token_idx_in_cache // self.tokens_per_block
|
|
# Replace the shared block with the unshared ones
|
|
self.blocks_manager.replace_shared_block(
|
|
seq, next_block_idx)
|
|
|
|
self.lens[batch_idx] += 1
|
|
|
|
# Remove finished sequences
|
|
for fi in range(len(finished)):
|
|
if finished[fi]:
|
|
self.blocks_manager.free(self.sequences[fi])
|
|
self.lens = [l for l, f in zip(self.lens, finished) if not f]
|
|
|
|
# Remap sequence ids
|
|
new_sequences = []
|
|
batch_idx = 0
|
|
for seq, finish in zip(self.sequences, finished):
|
|
if not finish:
|
|
seq.batch_idx = batch_idx
|
|
new_sequences.append(seq)
|
|
batch_idx += 1
|
|
self.sequences = new_sequences
|
|
|
|
def add_sequence(self, sequence: GenerationSequence, context_len: int):
|
|
"""
|
|
Add sequence to the manager and allocate minimum amount of blocks for context
|
|
"""
|
|
seq_len = context_len + self.bubble_len
|
|
self.lens.append(seq_len)
|
|
self.sequences.append(sequence)
|
|
|
|
# Get the final token index in kv cache
|
|
final_token_kv_index = self.sink_block_token_num + (
|
|
(seq_len - 1 - self.sink_block_token_num) %
|
|
(self.max_token_num - self.sink_block_token_num))
|
|
|
|
# Get block index that with shareAmongBeams=False.
|
|
unshared_block_idx = -1
|
|
if final_token_kv_index % self.tokens_per_block > 0:
|
|
unshared_block_idx = final_token_kv_index // self.tokens_per_block
|
|
|
|
# Get context block num.
|
|
# Allocate one more block if there are tokens that can't be shared across beams.
|
|
seq_len = min(seq_len, self.max_token_num)
|
|
context_blocks = seq_len // self.tokens_per_block
|
|
if seq_len % self.tokens_per_block > 0:
|
|
context_blocks += 1
|
|
|
|
# Allocate blocks
|
|
for i in range(context_blocks):
|
|
self.blocks_manager.allocate(
|
|
sequence, share_across_beam=i != unshared_block_idx)
|
|
|
|
def get_block_pointers(self, beam_width: int) -> torch.Tensor:
|
|
"""
|
|
Returns arrays of pointers for all memory pools
|
|
"""
|
|
pointer_arrays = []
|
|
for pool in range(self.num_pools):
|
|
pointer_arrays.append(
|
|
self.blocks_manager.get_pointer_array(
|
|
pool, beam_width).view(dtype=torch.int64))
|
|
|
|
return torch.stack(pointer_arrays, dim=0)
|
|
|
|
|
|
class KVCacheUpdater:
|
|
|
|
def __init__(self):
|
|
self.use_paged_kv_cache = None
|
|
self.num_kv_heads = None
|
|
self.head_dim = None
|
|
self.elt_size = None
|
|
self.past_key_value_list = None
|
|
self.max_kv_cache_length = None
|
|
self.kv_cache_manager = None
|
|
|
|
def init_linear_kv_cache(self, num_kv_heads, head_dim, kv_cache_type,
|
|
past_key_value_list):
|
|
self.use_paged_kv_cache = False
|
|
self.num_kv_heads = num_kv_heads
|
|
self.head_dim = head_dim
|
|
self.past_key_value_list = past_key_value_list
|
|
self.elt_size = torch.zeros(1, dtype=kv_cache_type).element_size()
|
|
self.max_kv_cache_length = past_key_value_list[0].shape[3]
|
|
|
|
def init_paged_kv_cache(self, num_kv_heads, head_dim, kv_cache_type,
|
|
kv_cache_manager):
|
|
self.use_paged_kv_cache = True
|
|
self.num_kv_heads = num_kv_heads
|
|
self.head_dim = head_dim
|
|
self.kv_cache_manager = kv_cache_manager
|
|
self.elt_size = torch.zeros(1, dtype=kv_cache_type).element_size()
|
|
|
|
def update(self, accepted_draft_token_offsets,
|
|
packed_accepted_draft_tokens_indices, sequence_length_buffer,
|
|
rewind_tokens):
|
|
assert isinstance(rewind_tokens, torch.Tensor) or isinstance(
|
|
rewind_tokens, int)
|
|
rewind_tokens_tensor = rewind_tokens if isinstance(
|
|
rewind_tokens, torch.Tensor) else None
|
|
rewind_tokens_count = rewind_tokens if isinstance(rewind_tokens,
|
|
int) else 0
|
|
assert self.use_paged_kv_cache is not None
|
|
if self.use_paged_kv_cache:
|
|
host_kv_cache_block_pointers = self.kv_cache_manager.get_block_pointers(
|
|
1)
|
|
kv_cache_block_pointers = host_kv_cache_block_pointers.to('cuda')
|
|
torch.ops.tensorrt_llm.update_kv_cache_draft_token_location(
|
|
accepted_draft_token_offsets,
|
|
packed_accepted_draft_tokens_indices,
|
|
sequence_length_buffer,
|
|
True,
|
|
self.num_kv_heads,
|
|
self.head_dim * self.elt_size,
|
|
rewind_tokens_count,
|
|
self.kv_cache_manager.max_attention_window_size,
|
|
rewind_tokens_tensor,
|
|
None,
|
|
kv_cache_block_pointers,
|
|
self.kv_cache_manager.blocks_manager.max_blocks_per_seq,
|
|
self.kv_cache_manager.tokens_per_block,
|
|
None,
|
|
)
|
|
else:
|
|
torch.ops.tensorrt_llm.update_kv_cache_draft_token_location(
|
|
accepted_draft_token_offsets,
|
|
packed_accepted_draft_tokens_indices,
|
|
sequence_length_buffer,
|
|
False,
|
|
self.num_kv_heads,
|
|
self.head_dim * self.elt_size,
|
|
rewind_tokens_count,
|
|
self.max_kv_cache_length,
|
|
rewind_tokens_tensor,
|
|
self.past_key_value_list,
|
|
None,
|
|
None,
|
|
None,
|
|
None,
|
|
)
|