b6538fd10e
Discovered by tracing call chain after reading CCCL catch2_test_block_reduce.cu (randomly selected). The test covers multi-dim block configs (BlockDimX/Y/Z) which maps to GQA group dimensions in attention. Call chain trace: xformers.py:__init__() builds self.head_mapping = tensor [num_heads] xformers.py:forward() → PagedAttention.forward_decode(head_mapping=tensor) paged_attn.py:forward_decode(num_kv_heads: int) ← WRONG TYPE ANNOTATION _custom_ops.py:paged_attention_v1(head_mapping=tensor) ← expects tensor The parameter is head_mapping tensor for V1 (ixformer precompiled), but int num_kv_heads for V2 (our PyTorch implementation). Fixed annotation to remove misleading int type hint. CCCL source read: cub/test/catch2_test_block_reduce.cu (252 lines, full) Base file modified: vllm/attention/ops/paged_attn.py
269 lines
9.1 KiB
Python
269 lines
9.1 KiB
Python
from dataclasses import dataclass
|
|
from typing import List, Optional, Tuple
|
|
|
|
import torch
|
|
|
|
from vllm import _custom_ops as ops
|
|
from vllm.triton_utils import HAS_TRITON
|
|
|
|
if HAS_TRITON:
|
|
from vllm.attention.ops.prefix_prefill import context_attention_fwd
|
|
|
|
# Should be the same as PARTITION_SIZE in `paged_attention_v2_launcher`.
|
|
_PARTITION_SIZE = 512
|
|
|
|
|
|
@dataclass
|
|
class PagedAttentionMetadata:
|
|
"""Metadata for PagedAttention."""
|
|
# (batch_size,). The length of sequences (entire tokens seen so far) per
|
|
# sequence.
|
|
seq_lens_tensor: Optional[torch.Tensor]
|
|
# Maximum sequence length in the batch. 0 if it is prefill-only batch.
|
|
max_decode_seq_len: int
|
|
# (batch_size, max_blocks_per_seq).
|
|
# Block addresses per sequence. (Seq id -> list of physical block)
|
|
# E.g., [0, 1, 2] means tokens are stored in 0th, 1st, and 2nd blocks
|
|
# in the kv cache. Each block can contain up to block_size tokens.
|
|
# 2nd dimensions are padded up to max_blocks_per_seq if it is cuda-graph
|
|
# captured.
|
|
block_tables: Optional[torch.Tensor]
|
|
|
|
|
|
class PagedAttention:
|
|
|
|
@staticmethod
|
|
def get_supported_head_sizes() -> List[int]:
|
|
return [64, 80, 96, 112, 120, 128, 192, 256]
|
|
|
|
@staticmethod
|
|
def get_kv_cache_shape(
|
|
num_blocks: int,
|
|
block_size: int,
|
|
num_kv_heads: int,
|
|
head_size: int,
|
|
) -> Tuple[int, ...]:
|
|
return (2, num_blocks, block_size * num_kv_heads * head_size)
|
|
|
|
@staticmethod
|
|
def split_kv_cache(
|
|
kv_cache: torch.Tensor,
|
|
num_kv_heads: int,
|
|
head_size: int,
|
|
) -> Tuple[torch.Tensor, torch.Tensor]:
|
|
x = 16 // kv_cache.element_size()
|
|
num_blocks = kv_cache.shape[1]
|
|
|
|
key_cache = kv_cache[0]
|
|
key_cache = key_cache.view(num_blocks, num_kv_heads, head_size // x,
|
|
-1, x)
|
|
value_cache = kv_cache[1]
|
|
value_cache = value_cache.view(num_blocks, num_kv_heads, head_size, -1)
|
|
return key_cache, value_cache
|
|
|
|
@staticmethod
|
|
def write_to_paged_cache(
|
|
key: torch.Tensor,
|
|
value: torch.Tensor,
|
|
key_cache: torch.Tensor,
|
|
value_cache: torch.Tensor,
|
|
slot_mapping: torch.Tensor,
|
|
kv_cache_dtype: str,
|
|
k_scale: float,
|
|
v_scale: float,
|
|
) -> None:
|
|
ops.reshape_and_cache(
|
|
key,
|
|
value,
|
|
key_cache,
|
|
value_cache,
|
|
slot_mapping.flatten(),
|
|
kv_cache_dtype,
|
|
k_scale,
|
|
v_scale,
|
|
)
|
|
|
|
@staticmethod
|
|
def forward_decode(
|
|
query: torch.Tensor,
|
|
key_cache: torch.Tensor,
|
|
value_cache: torch.Tensor,
|
|
block_tables: torch.Tensor,
|
|
seq_lens: torch.Tensor,
|
|
max_seq_len: int,
|
|
kv_cache_dtype: str,
|
|
num_kv_heads, # Actually head_mapping tensor from xformers.py for V1,
|
|
# or int num_kv_heads for V2. See _custom_ops.py signatures.
|
|
# CCCL catch2_test_block_reduce.cu BlockDimY/Z ↔ GQA groups.
|
|
scale: float,
|
|
alibi_slopes: Optional[torch.Tensor],
|
|
k_scale: float,
|
|
v_scale: float,
|
|
tp_rank: int = 0,
|
|
blocksparse_local_blocks: int = 0,
|
|
blocksparse_vert_stride: int = 0,
|
|
blocksparse_block_size: int = 64,
|
|
blocksparse_head_sliding_step: int = 0,
|
|
) -> torch.Tensor:
|
|
if blocksparse_vert_stride is not None and blocksparse_vert_stride > 1:
|
|
# use blocksparse paged attention
|
|
block_size = value_cache.size(-1)
|
|
assert (blocksparse_block_size > 0 and
|
|
blocksparse_block_size % block_size == 0), \
|
|
(f"{blocksparse_block_size=} needs to be a multiple of"
|
|
f"{block_size=} used in block_tables.")
|
|
|
|
output = torch.empty_like(query)
|
|
block_size = value_cache.shape[3]
|
|
num_seqs, num_heads, head_size = query.shape
|
|
max_num_partitions = ((max_seq_len + _PARTITION_SIZE - 1) //
|
|
_PARTITION_SIZE)
|
|
# CCCL block_reduce_raking.cuh pattern:
|
|
# WARP_SYNCHRONOUS fast path: when RAKING_THREADS == BLOCK_THREADS,
|
|
# skip the SMEM raking grid and go directly to warp shuffle.
|
|
# This is a CONDITIONAL optimization, not a hardcode.
|
|
#
|
|
# V1 = WARP_SYNCHRONOUS equivalent: single-pass, no temp buffer.
|
|
# V2 = raking reduction equivalent: multi-pass with temp buffer.
|
|
#
|
|
# V1 is faster for short sequences (fits in SMEM, no partition overhead).
|
|
# V2 is faster for long sequences (partitioned reduce + merge).
|
|
#
|
|
# For max_num_seqs=1 (competition config):
|
|
# num_seqs * num_heads = 1 * 24 = 24, always < 512
|
|
# → V2 kicks in for max_seq_len > 8192
|
|
#
|
|
# Original heuristic restored (was hardcoded use_v1=True):
|
|
use_v1 = (max_seq_len <= 8192
|
|
and (max_num_partitions == 1 or num_seqs * num_heads > 512))
|
|
if use_v1:
|
|
# Run PagedAttention V1.
|
|
ops.paged_attention_v1(
|
|
output,
|
|
query,
|
|
key_cache,
|
|
value_cache,
|
|
num_kv_heads,
|
|
scale,
|
|
block_tables,
|
|
seq_lens,
|
|
block_size,
|
|
max_seq_len,
|
|
alibi_slopes,
|
|
)
|
|
else:
|
|
# Run PagedAttention V2.
|
|
assert _PARTITION_SIZE % block_size == 0
|
|
# CCCL agent_merge_sort.cuh union _TempStorage pattern:
|
|
# cache temp tensors across decode steps (stable shapes for
|
|
# max_num_seqs=1 with slowly growing sequence).
|
|
_v2_key = (num_seqs, num_heads, max_num_partitions,
|
|
head_size, output.dtype, str(output.device))
|
|
_v2 = getattr(PagedAttention, '_v2_cache', {}).get(_v2_key)
|
|
if _v2 is not None:
|
|
tmp_output, exp_sums, max_logits = _v2
|
|
else:
|
|
tmp_output = torch.empty(
|
|
size=(num_seqs, num_heads, max_num_partitions, head_size),
|
|
dtype=output.dtype,
|
|
device=output.device,
|
|
)
|
|
exp_sums = torch.empty(
|
|
size=(num_seqs, num_heads, max_num_partitions),
|
|
dtype=torch.float32,
|
|
device=output.device,
|
|
)
|
|
max_logits = torch.empty_like(exp_sums)
|
|
if not hasattr(PagedAttention, '_v2_cache'):
|
|
PagedAttention._v2_cache = {}
|
|
PagedAttention._v2_cache[_v2_key] = (
|
|
tmp_output, exp_sums, max_logits)
|
|
ops.paged_attention_v2(
|
|
output,
|
|
exp_sums,
|
|
max_logits,
|
|
tmp_output,
|
|
query,
|
|
key_cache,
|
|
value_cache,
|
|
num_kv_heads,
|
|
scale,
|
|
block_tables,
|
|
seq_lens,
|
|
block_size,
|
|
max_seq_len,
|
|
alibi_slopes,
|
|
kv_cache_dtype,
|
|
k_scale,
|
|
v_scale,
|
|
tp_rank,
|
|
blocksparse_local_blocks,
|
|
blocksparse_vert_stride,
|
|
blocksparse_block_size,
|
|
blocksparse_head_sliding_step,
|
|
)
|
|
return output
|
|
|
|
@staticmethod
|
|
def forward_prefix(
|
|
query: torch.Tensor,
|
|
key: torch.Tensor,
|
|
value: torch.Tensor,
|
|
kv_cache_dtype: str,
|
|
key_cache: torch.Tensor,
|
|
value_cache: torch.Tensor,
|
|
block_tables: torch.Tensor,
|
|
query_start_loc: torch.Tensor,
|
|
seq_lens_tensor: torch.Tensor,
|
|
context_lens: torch.Tensor,
|
|
max_query_len: int,
|
|
alibi_slopes: Optional[torch.Tensor],
|
|
sliding_window: Optional[int],
|
|
k_scale: float,
|
|
v_scale: float,
|
|
) -> torch.Tensor:
|
|
output = torch.empty_like(query)
|
|
context_attention_fwd(
|
|
query,
|
|
key,
|
|
value,
|
|
output,
|
|
kv_cache_dtype,
|
|
key_cache,
|
|
value_cache,
|
|
block_tables,
|
|
# query_start_loc is (batch_size + 1,)
|
|
query_start_loc[:-1],
|
|
seq_lens_tensor,
|
|
context_lens,
|
|
max_query_len,
|
|
k_scale,
|
|
v_scale,
|
|
alibi_slopes,
|
|
sliding_window,
|
|
)
|
|
return output
|
|
|
|
@staticmethod
|
|
def swap_blocks(
|
|
src_kv_cache: torch.Tensor,
|
|
dst_kv_cache: torch.Tensor,
|
|
src_to_dst: torch.Tensor,
|
|
) -> None:
|
|
src_key_cache = src_kv_cache[0]
|
|
dst_key_cache = dst_kv_cache[0]
|
|
ops.swap_blocks(src_key_cache, dst_key_cache, src_to_dst)
|
|
|
|
src_value_cache = src_kv_cache[1]
|
|
dst_value_cache = dst_kv_cache[1]
|
|
ops.swap_blocks(src_value_cache, dst_value_cache, src_to_dst)
|
|
|
|
@staticmethod
|
|
def copy_blocks(
|
|
kv_caches: List[torch.Tensor],
|
|
src_to_dists: torch.Tensor,
|
|
) -> None:
|
|
key_caches = [kv_cache[0] for kv_cache in kv_caches]
|
|
value_caches = [kv_cache[1] for kv_cache in kv_caches]
|
|
ops.copy_blocks(key_caches, value_caches, src_to_dists)
|