Skip to content

[ROCm] Make Whisper causal attention backend-agnostic #34631

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
laudney wants to merge 1 commit into
base: main
Choose a base branch
from
+8 −45
Open

[ROCm] Make Whisper causal attention backend-agnostic #34631

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 8 additions & 45 deletions vllm/model_executor/models/whisper_causal.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,14 +30,6 @@
CommonAttentionMetadata,
subclass_attention_backend_with_overrides,
)
from vllm.v1.attention.backends.flash_attn import FlashAttentionBackend

try:
from vllm.v1.attention.backends.rocm_aiter_fa import AiterFlashAttentionBackend
except ImportError:
AiterFlashAttentionBackend = None
from vllm.v1.attention.backends.rocm_attn import RocmAttentionBackend
from vllm.v1.attention.backends.triton_attn import TritonAttentionBackend
from vllm.v1.attention.selector import get_attn_backend
from vllm.v1.kv_cache_interface import AttentionSpec

Expand Down Expand Up @@ -211,36 +203,6 @@ def forward(
output_block_scale,
)

_SUPPORTED_BACKENDS = tuple(
b
for b in (
AiterFlashAttentionBackend,
FlashAttentionBackend,
RocmAttentionBackend,
TritonAttentionBackend,
)
if b is not None
)

if not issubclass(underlying_attn_backend, _SUPPORTED_BACKENDS):
raise NotImplementedError(
f"{underlying_attn_backend} is not yet supported."
"Contributions to support more backends are much "
"appreciated."
)

if not issubclass(underlying_attn_backend, FlashAttentionBackend):
logger.info(
"Using %s for Whisper causal attention with block pooling. "
"This backend was recently enabled for this model. "
"If you encounter any accuracy or performance issues, "
"please open an issue at "
"https://github.com/vllm-project/vllm/issues "
"with the [ROCm] tag so it can be triaged by the "
"appropriate team.",
underlying_attn_backend.get_name(),
)

attn_backend = subclass_attention_backend_with_overrides(
name_prefix=prefix,
attention_backend_cls=underlying_attn_backend,
Expand All @@ -251,13 +213,14 @@ def forward(
block_size,
num_kv_heads,
head_size,
cache_dtype_str: underlying_attn_backend.get_kv_cache_shape(
num_blocks,
# we stretch each block by `block_pool_size`
block_size * block_pool_size,
num_kv_heads // block_pool_size,
head_size,
cache_dtype_str,
cache_dtype_str: (
underlying_attn_backend.get_kv_cache_shape(
num_blocks,
block_size * block_pool_size,
num_kv_heads // block_pool_size,
head_size,
cache_dtype_str,
)
),
"forward_includes_kv_cache_update": True,
},
Expand Down
Loading