[ROCm] [Perf] Add AITER Flash Attention (#941)

Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
This commit is contained in:
TJian authored and GitHub committed 2026-01-27 20:14:49 +08:00
1 parent cadd77262d
commit 46e96e18e3
3 files changed
+43 -2

No files matched your search

@@ -2,7 +2,6 @@
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import torch
from fa3_fwd_interface import flash_attn_func, flash_attn_varlen_func
from vllm.logger import init_logger
from vllm_omni.diffusion.attention.backends.abstract import (
@@ -10,7 +9,13 @@ from vllm_omni.diffusion.attention.backends.abstract import (
AttentionImpl,
AttentionMetadata,
)
from vllm_omni.diffusion.attention.backends.utils.fa import _pad_input, _unpad_input, _upad_input
from vllm_omni.diffusion.attention.backends.utils.fa import (
_pad_input,
_unpad_input,
_upad_input,
flash_attn_func,
flash_attn_varlen_func,
)
logger = init_logger(__name__)