[MM][OOT] Support CPU `seq_lens` for OOT MMEncoderAttention kernels (#36605)

Signed-off-by: shen-shanshan <467638484@qq.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>

tests/kernels/attention/test_mha_attn.py

@@ -297,11 +297,10 @@ def test_mha_attn_varlen_forward_flashinfer(
         hidden_size = num_heads * head_size
         tp_size = 1
 
-        sequence_lengths_np = MMEncoderAttention.maybe_compute_sequence_lengths(
-            AttentionBackendEnum.FLASHINFER, cu_seqlens_np
-        )
-        sequence_lengths = torch.from_numpy(sequence_lengths_np).to(
-            device, dtype=torch.int32, non_blocking=True
+        sequence_lengths = MMEncoderAttention.maybe_compute_seq_lens(
+            AttentionBackendEnum.FLASHINFER,
+            cu_seqlens_np,
+            device,
         )
 
         max_seqlen_val = MMEncoderAttention.compute_max_seqlen(
@@ -309,14 +308,12 @@ def test_mha_attn_varlen_forward_flashinfer(
         )
         max_seqlen = torch.tensor(max_seqlen_val, device=device, dtype=torch.int32)
 
-        cu_seqlens_np = MMEncoderAttention.maybe_recompute_cu_seqlens(
+        cu_seqlens = MMEncoderAttention.maybe_recompute_cu_seqlens(
             AttentionBackendEnum.FLASHINFER,
             cu_seqlens_np,
             hidden_size,
             tp_size,
-        )
-        cu_seqlens = torch.from_numpy(cu_seqlens_np).to(
-            device, dtype=torch.int32, non_blocking=True
+            device,
         )
 
         scale = 1.0 / head_size**0.5

vllm/model_executor/custom_op.py

@@ -22,6 +22,12 @@ op_registry: dict[str, type["CustomOp"] | type["PluggableLayer"]] = {}
 op_registry_oot: dict[str, type["CustomOp"] | type["PluggableLayer"]] = {}
 
 
+def get_oot_class_by_name(class_name: str) -> type | None:
+    if class_name in op_registry_oot:
+        return op_registry_oot[class_name]
+    return None
+
+
 class PluggableLayer(nn.Module):
     """
     Base class for pluggable layers.

vllm/model_executor/layers/attention/mm_encoder_attention.py

@@ -6,7 +6,7 @@ import numpy as np
 import torch
 
 from vllm.logger import init_logger
-from vllm.model_executor.custom_op import CustomOp
+from vllm.model_executor.custom_op import CustomOp, get_oot_class_by_name
 from vllm.model_executor.models.vision import get_vit_attn_backend
 from vllm.utils.math_utils import round_up
 from vllm.v1.attention.backends.fa_utils import get_flash_attn_version
@@ -119,17 +119,25 @@ class MMEncoderAttention(CustomOp):
         return max_seqlen
 
     @classmethod
-    def maybe_compute_sequence_lengths(
+    def maybe_compute_seq_lens(
         cls,
         attn_backend: AttentionBackendEnum,
         cu_seqlens: np.ndarray,
-    ) -> np.ndarray | None:
+        device: torch.device,
+    ) -> torch.Tensor | None:
+        if (oot_class := get_oot_class_by_name(cls.__name__)) is not None:
+            return oot_class.maybe_compute_seq_lens(attn_backend, cu_seqlens, device)  # type: ignore[attr-defined]
+
         if attn_backend != AttentionBackendEnum.FLASHINFER:
             return None
+
         sequence_lengths = cu_seqlens[1:] - cu_seqlens[:-1]
         sequence_lengths = add_padding_to_seqlens(
             sequence_lengths, len(sequence_lengths), 0
         )
+        sequence_lengths = torch.from_numpy(sequence_lengths).to(
+            device, non_blocking=True
+        )
         return sequence_lengths
 
     @classmethod
@@ -139,10 +147,14 @@ class MMEncoderAttention(CustomOp):
         cu_seqlens: np.ndarray,
         hidden_size: int,
         tp_size: int,
-    ) -> np.ndarray:
-        if attn_backend != AttentionBackendEnum.FLASHINFER:
-            return cu_seqlens
+        device: torch.device,
+    ) -> torch.Tensor:
+        if (oot_class := get_oot_class_by_name(cls.__name__)) is not None:
+            return oot_class.maybe_recompute_cu_seqlens(  # type: ignore[attr-defined]
+                attn_backend, cu_seqlens, hidden_size, tp_size, device
+            )
 
+        if attn_backend == AttentionBackendEnum.FLASHINFER:
             batch_size = len(cu_seqlens) - 1
             scale = hidden_size // tp_size
             cu_seqlens = cu_seqlens * scale
@@ -156,7 +168,10 @@ class MMEncoderAttention(CustomOp):
             cu_seqlens_v = add_padding_to_seqlens(
                 cu_seqlens_v, batch_size, cu_seqlens_v[-1]
             )
-        return np.concatenate([cu_seqlens_qko, cu_seqlens_v])
+            cu_seqlens = np.concatenate([cu_seqlens_qko, cu_seqlens_v])
+
+        cu_seqlens = torch.from_numpy(cu_seqlens).to(device, non_blocking=True)
+        return cu_seqlens
 
     def __init__(
         self,

vllm/model_executor/models/qwen3_omni_moe_thinker.py

@@ -983,12 +983,10 @@ class Qwen3Omni_VisionTransformer(nn.Module):
             grid_thw_np[:, 1] * grid_thw_np[:, 2], grid_thw_np[:, 0]
         ).cumsum(axis=0, dtype=np.int32)
         cu_seqlens_np = np.concatenate([np.zeros(1, dtype=np.int32), cu_seqlens_np])
-        sequence_lengths = MMEncoderAttention.maybe_compute_sequence_lengths(
-            self.attn_backend, cu_seqlens_np
-        )
-        if sequence_lengths is not None:
-            sequence_lengths = torch.from_numpy(sequence_lengths).to(
-                self.device, non_blocking=True
+        sequence_lengths = MMEncoderAttention.maybe_compute_seq_lens(
+            self.attn_backend,
+            cu_seqlens_np,
+            self.device,
         )
 
         hidden_states_list = []

vllm/model_executor/models/qwen3_vl.py

@@ -550,12 +550,8 @@ class Qwen3_VisionTransformer(nn.Module):
             axis=0, dtype=np.int32
         )
         cu_seqlens = np.concatenate([np.zeros(1, dtype=np.int32), cu_seqlens])
-        sequence_lengths = MMEncoderAttention.maybe_compute_sequence_lengths(
-            self.attn_backend, cu_seqlens
-        )
-        if sequence_lengths is not None:
-            sequence_lengths = torch.from_numpy(sequence_lengths).to(
-                self.device, non_blocking=True
+        sequence_lengths = MMEncoderAttention.maybe_compute_seq_lens(
+            self.attn_backend, cu_seqlens, self.device
         )
         max_seqlen = torch.tensor(
             MMEncoderAttention.compute_max_seqlen(self.attn_backend, cu_seqlens),
@@ -567,8 +563,8 @@ class Qwen3_VisionTransformer(nn.Module):
             cu_seqlens,
             self.hidden_size,
             self.tp_size,
+            self.device,
         )
-        cu_seqlens = torch.from_numpy(cu_seqlens).to(self.device, non_blocking=True)
         hidden_states = hidden_states.unsqueeze(1)
 
         deepstack_feature_lists = []