avoid performing index selection of sin/cos cache every layer

Signed-off-by: whx-sjtu <2952154980@qq.com>

Author: whx-sjtu

vllm_ascend/attention/mla_v1.py

@@ -9,6 +9,7 @@
                                               MLAAttentionImpl)
 from vllm.attention.backends.utils import PAD_SLOT_ID
 from vllm.config import get_current_vllm_config
+from vllm.forward_context import get_forward_context
 from vllm.model_executor.layers.linear import (LinearBase,
                                                UnquantizedLinearMethod)
 from vllm.utils import cdiv, round_down
@@ -93,6 +94,8 @@ class ChunkedContextMetadata:
     max_query_len: int
     max_seq_lens: int
     chunked_context: Optional[ChunkedContextMetadata] = None
+    sin: torch.Tensor = None
+    cos: torch.Tensor = None
 
 
 @dataclass
@@ -106,6 +109,8 @@ class AscendMLADecodeMetadata:
     seq_lens_list: list[int]
     actual_seq_q_lens: Optional[list[int]] = None
     attn_mask: Optional[torch.Tensor] = None
+    sin: torch.Tensor = None
+    cos: torch.Tensor = None
 
 
 @dataclass
@@ -217,6 +222,9 @@ def __init__(self,
             )
         ascend_config = get_ascend_config()
         self.torchair_graph_enabled = ascend_config.torchair_graph_config.enabled
+        self.rope_dim = self.runner.model_config.hf_text_config.qk_rope_head_dim
+        self.cos_cache = None
+        self.sin_cache = None
 
     def reorder_batch(self, input_batch: "InputBatch",
                       scheduler_output: "SchedulerOutput") -> bool:
@@ -348,6 +356,18 @@ def build_torchair_graph_dummy(
         else:
             attn_state = AscendAttentionState.DecodeOnly
             num_decode_tokens = 1
+        sin = torch.ones(num_reqs,
+                         1,
+                         1,
+                         self.rope_dim,
+                         dtype=self.runner.dtype,
+                         device=device)
+        cos = torch.ones(num_reqs,
+                         1,
+                         1,
+                         self.rope_dim,
+                         dtype=self.runner.dtype,
+                         device=device)
         decode_metadata = AscendMLADecodeMetadata(
             input_positions=input_positions,
             block_table=block_table,
@@ -356,7 +376,8 @@ def build_torchair_graph_dummy(
             max_seq_lens=1,
             attn_mask=self.runner.spec_attn_mask,
             actual_seq_q_lens=self.runner.actual_seq_q_lens[:num_reqs],
-        )
+            sin=sin,
+            cos=cos)
         return self.metadata_cls(  # type: ignore
             num_input_tokens=num_actual_tokens,
             num_actual_tokens=num_actual_tokens,
@@ -408,6 +429,14 @@ def build(
         max_query_len = query_lens.max().item()
         max_seq_lens = seq_lens.max().item()
         query_start_loc = common_attn_metadata.query_start_loc
+        if self.cos_cache is None:
+            self.cos_cache = self.runner.get_model(
+            ).model.layers[0].self_attn.rotary_emb.cos_cached
+            self.sin_cache = self.runner.get_model(
+            ).model.layers[0].self_attn.rotary_emb.sin_cached
+        if self.cos_cache.dtype != self.runner.dtype:
+            self.cos_cache = self.cos_cache.to(self.runner.dtype)
+            self.sin_cache = self.sin_cache.to(self.runner.dtype)
 
         prefill_metadata = None
         chunked_context_metadata = None
@@ -454,18 +483,24 @@ def build(
                     chunk_seq_lens=chunk_seq_lens,
                     workspace=self.chunked_prefill_workspace,
                 )
-
+            prefill_input_positions = input_positions[tokens_start:]
+            cos = self.cos_cache[prefill_input_positions].unsqueeze(
+                1).unsqueeze(2)
+            sin = self.sin_cache[prefill_input_positions].unsqueeze(
+                1).unsqueeze(2)
             prefill_metadata = AscendMLAPrefillMetadata(
                 attn_mask=self.runner.attn_mask,
                 query_lens=query_lens[tokens_start:],
                 seq_lens=seq_lens,
                 context_lens=seq_lens[tokens_start:],
-                input_positions=input_positions[tokens_start:],
+                input_positions=prefill_input_positions,
                 block_table=block_table[reqs_start:, ...],
                 max_query_len=max_query_len,
                 max_seq_lens=max_seq_lens,
                 query_start_loc=prefill_query_start_loc,
                 chunked_context=chunked_context_metadata,
+                sin=sin,
+                cos=cos,
             )
 
         decode_metadata = None
@@ -510,8 +545,11 @@ def build(
                 actual_seq_q_lens = query_start_loc[1:].tolist(
                 ) + self.runner.actual_seq_q_lens[num_reqs:num_reqs +
                                                   num_reqs_pad_size]
+                cos = self.cos_cache[input_positions].unsqueeze(1).unsqueeze(2)
+                sin = self.sin_cache[input_positions].unsqueeze(1).unsqueeze(2)
             else:
                 seq_lens_list = seq_lens.tolist()
+                cos, sin = None, None
 
             decode_metadata = AscendMLADecodeMetadata(
                 input_positions=input_positions,
@@ -521,7 +559,8 @@ def build(
                 max_seq_lens=max_seq_lens,
                 attn_mask=self.runner.spec_attn_mask,
                 actual_seq_q_lens=actual_seq_q_lens,
-            )
+                sin=sin,
+                cos=cos)
 
         return self.metadata_cls(  # type: ignore
             num_actual_tokens=num_actual_tokens,
@@ -1113,15 +1152,8 @@ def forward(
             decode_k_nope = None
             assert attn_metadata.decode is not None
             if self.running_in_graph:
-                seq_len = self.rotary_emb.max_position_embeddings * self.rotary_emb.scaling_factor
-                cos = self.rotary_emb.cos_cached[:seq_len].to(
-                    dtype=decode_hs_or_q_c.dtype)
-                sin = self.rotary_emb.sin_cached[:seq_len].to(
-                    dtype=decode_hs_or_q_c.dtype)
-                cos = cos[attn_metadata.decode.input_positions]
-                sin = sin[attn_metadata.decode.input_positions]
-                cos = cos[:, None, None, :]
-                sin = sin[:, None, None, :]
+                cos = attn_metadata.decode.cos
+                sin = attn_metadata.decode.sin
                 # Without explicitly controlling the order, IndexByTensor operations
                 # would be placed after `matmul W_KV_T` hindering the overlapping of
                 # KvRmsNormRopeCache and SingleRope.
@@ -1156,15 +1188,8 @@ def forward(
             prefill_q_nope = prefill_q[..., :self.qk_nope_head_dim]
             if self.torchair_graph_enabled:
                 num_tokens = prefill_hs_or_q_c.shape[0]
-                seq_len = self.rotary_emb.max_position_embeddings * self.rotary_emb.scaling_factor
-                cos = self.rotary_emb.cos_cached[:seq_len].to(
-                    dtype=prefill_q_pe.dtype)
-                sin = self.rotary_emb.sin_cached[:seq_len].to(
-                    dtype=prefill_q_pe.dtype)
-                cos = cos[attn_metadata.prefill.input_positions]
-                sin = sin[attn_metadata.prefill.input_positions]
-                cos = cos[:, None, None, :]
-                sin = sin[:, None, None, :]
+                cos = attn_metadata.prefill.cos
+                sin = attn_metadata.prefill.sin
 
                 prefill_q_pe = self.rope_single(prefill_q_pe, cos, sin)
                 prefill_k_pe, prefill_k_nope = self.exec_kv_prefill(

vllm_ascend/worker/model_runner_v1.py

@@ -1647,6 +1647,8 @@ def _dummy_run(
                             attn_metadata.decode.block_table)
                         torch._dynamo.mark_static(
                             attn_metadata.decode.input_positions)
+                        torch._dynamo.mark_static(attn_metadata.decode.sin)
+                        torch._dynamo.mark_static(attn_metadata.decode.cos)
                         torch._dynamo.mark_static(attn_metadata.slot_mapping)
                         for kv in self.kv_caches:
                             assert isinstance(