"docs/features/quantization/quantized_kvcache.md" did not exist on "01a55941f5443c16a88889995f8149edcee2eec3"
Unverified Commit 073a4bd1 authored by Woosuk Kwon's avatar Woosuk Kwon Committed by GitHub
Browse files

[Kernel] Use `out` arg in flash_attn_varlen_func (#10811)


Signed-off-by: default avatarWoosuk Kwon <woosuk.kwon@berkeley.edu>
parent b7954776
...@@ -522,7 +522,7 @@ else() ...@@ -522,7 +522,7 @@ else()
FetchContent_Declare( FetchContent_Declare(
vllm-flash-attn vllm-flash-attn
GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git
GIT_TAG fdf6d72b48aea41f4ae6a89139a453dae554abc8 GIT_TAG 04325b6798bcc326c86fb35af62d05a9c8c8eceb
GIT_PROGRESS TRUE GIT_PROGRESS TRUE
# Don't share the vllm-flash-attn build between build types # Don't share the vllm-flash-attn build between build types
BINARY_DIR ${CMAKE_BINARY_DIR}/vllm-flash-attn BINARY_DIR ${CMAKE_BINARY_DIR}/vllm-flash-attn
......
...@@ -71,6 +71,7 @@ def ref_paged_attn( ...@@ -71,6 +71,7 @@ def ref_paged_attn(
return torch.cat(outputs, dim=0) return torch.cat(outputs, dim=0)
@pytest.mark.parametrize("use_out", [True, False])
@pytest.mark.parametrize("kv_lens", [[1328, 18, 463], [1, 54, 293, 70]]) @pytest.mark.parametrize("kv_lens", [[1328, 18, 463], [1, 54, 293, 70]])
@pytest.mark.parametrize("num_heads", NUM_HEADS) @pytest.mark.parametrize("num_heads", NUM_HEADS)
@pytest.mark.parametrize("head_size", HEAD_SIZES) @pytest.mark.parametrize("head_size", HEAD_SIZES)
...@@ -81,6 +82,7 @@ def ref_paged_attn( ...@@ -81,6 +82,7 @@ def ref_paged_attn(
@pytest.mark.parametrize("sliding_window", [None, 256]) @pytest.mark.parametrize("sliding_window", [None, 256])
@torch.inference_mode() @torch.inference_mode()
def test_flash_attn_with_paged_kv( def test_flash_attn_with_paged_kv(
use_out: bool,
kv_lens: List[int], kv_lens: List[int],
num_heads: Tuple[int, int], num_heads: Tuple[int, int],
head_size: int, head_size: int,
...@@ -116,17 +118,22 @@ def test_flash_attn_with_paged_kv( ...@@ -116,17 +118,22 @@ def test_flash_attn_with_paged_kv(
(num_seqs, max_num_blocks_per_seq), (num_seqs, max_num_blocks_per_seq),
dtype=torch.int32) dtype=torch.int32)
q = query.unsqueeze(1)
out = torch.empty_like(q) if use_out else None
output = flash_attn_with_kvcache( output = flash_attn_with_kvcache(
q=query.unsqueeze(1), q=q,
k_cache=key_cache, k_cache=key_cache,
v_cache=value_cache, v_cache=value_cache,
out=out,
softmax_scale=scale, softmax_scale=scale,
causal=True, causal=True,
block_table=block_tables, block_table=block_tables,
cache_seqlens=kv_lens_tensor, cache_seqlens=kv_lens_tensor,
softcap=soft_cap if soft_cap is not None else 0, softcap=soft_cap if soft_cap is not None else 0,
window_size=window_size, window_size=window_size,
).squeeze(1) )
output = output if not use_out else out
output = output.squeeze(1)
ref_output = ref_paged_attn(query=query, ref_output = ref_paged_attn(query=query,
key_cache=key_cache, key_cache=key_cache,
...@@ -141,7 +148,10 @@ def test_flash_attn_with_paged_kv( ...@@ -141,7 +148,10 @@ def test_flash_attn_with_paged_kv(
f"{torch.max(torch.abs(output - ref_output))}" f"{torch.max(torch.abs(output - ref_output))}"
@pytest.mark.parametrize("seq_lens", [[(1, 1328), (5, 18), (129, 463)]]) @pytest.mark.parametrize("use_out", [True, False])
@pytest.mark.parametrize("seq_lens",
[[(1, 1328), (5, 18),
(129, 463)], [(1, 523), (1, 37), (1, 2011)]])
@pytest.mark.parametrize("num_heads", NUM_HEADS) @pytest.mark.parametrize("num_heads", NUM_HEADS)
@pytest.mark.parametrize("head_size", HEAD_SIZES) @pytest.mark.parametrize("head_size", HEAD_SIZES)
@pytest.mark.parametrize("block_size", BLOCK_SIZES) @pytest.mark.parametrize("block_size", BLOCK_SIZES)
...@@ -151,6 +161,7 @@ def test_flash_attn_with_paged_kv( ...@@ -151,6 +161,7 @@ def test_flash_attn_with_paged_kv(
@pytest.mark.parametrize("num_blocks", NUM_BLOCKS) @pytest.mark.parametrize("num_blocks", NUM_BLOCKS)
@torch.inference_mode() @torch.inference_mode()
def test_varlen_with_paged_kv( def test_varlen_with_paged_kv(
use_out: bool,
seq_lens: List[Tuple[int, int]], seq_lens: List[Tuple[int, int]],
num_heads: Tuple[int, int], num_heads: Tuple[int, int],
head_size: int, head_size: int,
...@@ -197,10 +208,12 @@ def test_varlen_with_paged_kv( ...@@ -197,10 +208,12 @@ def test_varlen_with_paged_kv(
(num_seqs, max_num_blocks_per_seq), (num_seqs, max_num_blocks_per_seq),
dtype=torch.int32) dtype=torch.int32)
out = torch.empty_like(query) if use_out else None
output = flash_attn_varlen_func( output = flash_attn_varlen_func(
q=query, q=query,
k=key_cache, k=key_cache,
v=value_cache, v=value_cache,
out=out,
cu_seqlens_q=cu_query_lens, cu_seqlens_q=cu_query_lens,
cu_seqlens_k=cu_kv_lens, cu_seqlens_k=cu_kv_lens,
max_seqlen_q=max_query_len, max_seqlen_q=max_query_len,
...@@ -211,6 +224,7 @@ def test_varlen_with_paged_kv( ...@@ -211,6 +224,7 @@ def test_varlen_with_paged_kv(
block_table=block_tables, block_table=block_tables,
softcap=soft_cap if soft_cap is not None else 0, softcap=soft_cap if soft_cap is not None else 0,
) )
output = output if not use_out else out
ref_output = ref_paged_attn( ref_output = ref_paged_attn(
query=query, query=query,
......
...@@ -205,10 +205,12 @@ def unified_v1_flash_attention( ...@@ -205,10 +205,12 @@ def unified_v1_flash_attention(
v_scale, v_scale,
) )
attn_output = flash_attn_varlen_func( # Compute attention and update output up to `num_actual_tokens`.
flash_attn_varlen_func(
q=query[:num_actual_tokens], q=query[:num_actual_tokens],
k=key_cache, k=key_cache,
v=value_cache, v=value_cache,
out=output[:num_actual_tokens],
cu_seqlens_q=attn_metadata.query_start_loc, cu_seqlens_q=attn_metadata.query_start_loc,
max_seqlen_q=attn_metadata.max_query_len, max_seqlen_q=attn_metadata.max_query_len,
cu_seqlens_k=attn_metadata.seq_start_loc, cu_seqlens_k=attn_metadata.seq_start_loc,
...@@ -220,8 +222,6 @@ def unified_v1_flash_attention( ...@@ -220,8 +222,6 @@ def unified_v1_flash_attention(
block_table=attn_metadata.block_table, block_table=attn_metadata.block_table,
softcap=logits_soft_cap, softcap=logits_soft_cap,
) )
# TODO(woosuk): Remove this unnecessary copy.
output[:num_actual_tokens].copy_(attn_output)
def unified_v1_flash_attention_fake( def unified_v1_flash_attention_fake(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment