chunked prefill, access splitfuse op

vllm_ascend/attention/attention_mask.py

@@ -39,11 +39,13 @@ def __init__(
         self,
         max_seq_len: int,
         dtype: torch.dtype,
+        device: torch.device,
     ):
         attn_mask = _generate_attn_mask(max_seq_len, dtype)
 
         self._seq_len_cached = attn_mask.shape[0]
         self.attn_mask_cache = attn_mask
+        self.chunked_prefill_attn_mask = torch.triu(torch.ones(2048, 2048), diagonal=1).to(torch.int8).to(device)
 
     @staticmethod
     def get_mask_scale_factor(dtype: torch.dtype = torch.float16):
@@ -66,24 +68,8 @@ def get_attn_mask(self, max_seq_len: int, dtype: torch.dtype,
 
     def get_splitfuse_attn_mask(
         self,
-        seq_lens: torch.Tensor,
-        position: torch.Tensor,
-        dtype: torch.dtype,
-        device: torch.device,
     ) -> torch.Tensor:
-        if dtype not in [torch.float16, torch.bfloat16]:
-            raise ValueError(
-                "splitfuse_attn_mask now only supports bf16 and fp16")
-        max_seq_len = max(seq_lens, default=0)
-        self._update_attn_cache(max_seq_len, dtype)
-        # FIXME: Currently the mask value of chunked-prefill situation and Prefill-Only situation
-        # is not the same. Fix this in the future when kernel is ready.
-        mask_scale_factor = AttentionMaskBuilder.get_mask_scale_factor(dtype)
-        attn_mask = torch.index_select(self.attn_mask_cache,
-                                       dim=0,
-                                       index=position)[:, :max_seq_len]
-        attn_mask *= mask_scale_factor
-        return attn_mask.contiguous().to(device, non_blocking=True)
+        return self.chunked_prefill_attn_mask
 
     def _update_attn_cache(self, seqlen: int, dtype: torch.dtype):
         if seqlen > self._seq_len_cached:

vllm_ascend/attention/attention_v1.py

@@ -455,18 +455,25 @@
             attn_metadata.seq_lens = \
                 attn_metadata.seq_lens.to(device=query.device)
 
-        torch_npu._npu_paged_attention_splitfuse(
+        num_block, block_size, head_num, head_dim = self.key_cache.shape
+        key = self.key_cache.view(num_block, block_size, -1)
+        value = self.value_cache.view(num_block, block_size, -1)
+
+        output, _ = torch_npu.npu_fused_infer_attention_score(
             query=query,
-            key_cache=self.key_cache,
-            value_cache=self.value_cache,
-            mask=attn_metadata.attn_mask,
+            key=key,
+            value=value,
+            atten_mask=attn_metadata.attn_mask,
             block_table=attn_metadata.block_tables,
-            seq_len=attn_metadata.query_lens,
-            context_lens=attn_metadata.seq_lens,
-            num_kv_heads=self.num_kv_heads,
+            input_layout="TND",
+            block_size=block_size,
+            actual_seq_lengths=attn_metadata.query_start_loc[1:],
+            actual_seq_lengths_kv=attn_metadata.seq_lens,
+            num_key_value_heads=self.num_kv_heads,
             num_heads=self.num_heads,
-            scale_value=self.scale,
-            out=output)
+            scale=self.scale,
+            sparse_mode=3,
+            )
         return output
 
     def forward(

vllm_ascend/worker/model_runner_v1.py

@@ -293,7 +293,7 @@ def __init__(self, vllm_config: VllmConfig, device: torch.device):
         )
 
         self.attn_mask_builder = AttentionMaskBuilder(
-            self.model_config.max_model_len, self.dtype)
+            self.model_config.max_model_len, self.dtype, self.device)
 
         # Set up speculative decoding.
         self.spec_attn_mask = None
@@ -797,8 +797,7 @@ def _make_attention_mask(self, seq_lens, position,
                              attn_state) -> torch.Tensor:
         # Chunk Prefill situation.
         if attn_state == AscendAttentionState.ChunkedPrefill and not self.vllm_config.model_config.use_mla:
-            return self.attn_mask_builder.get_splitfuse_attn_mask(
-                seq_lens, position, self.dtype, self.device)
+            return self.attn_mask_builder.get_splitfuse_attn_mask()
         # Prefill without cache situation.
         elif attn_state == AscendAttentionState.PrefillNoCache:
             max_seq_len = max(seq_lens, default=0)