diff --git a/vllm_ascend/attention/sfa_v1.py b/vllm_ascend/attention/sfa_v1.py
index 393ddb25eda..24ffcdd4b69 100644
--- a/vllm_ascend/attention/sfa_v1.py
+++ b/vllm_ascend/attention/sfa_v1.py
@@ -152,6 +152,8 @@ class AscendSFAMetadata:
     dsa_cp_context: DSACPContext | None = None
     reshape_cache_event: torch.npu.Event = None
     sfa_cp_metadata: AscendPCPMetadata | None = None
+    # Shared top-k indices reused by IndexCache layers in the same forward pass.
+    shared_topk_indices: torch.Tensor | None = None
     num_decodes: int = 0
     num_decode_tokens: int = 0
     num_prefills: int = 0
@@ -447,6 +449,7 @@ def __init__(
         if self.vllm_config.model_config.hf_config.model_type in ["glm_moe_dsa"]:
             self.is_rope_neox_style = False
             self.use_torch_npu_lightning_indexer = True
+        self.skip_topk = kwargs.get("skip_topk", False)
 
         # dsa c8
         self.use_sparse_c8_indexer = ascend_config.is_sparse_c8_layer(self.layer_name)
@@ -1229,16 +1232,20 @@ def forward(
             if self.is_kv_producer:
                 attn_metadata.reshape_cache_event.record()
 
-        topk_indices = self.indexer_select_post_process(
-            x=hidden_states,
-            q_c=q_c,
-            kv_cache=kv_cache,
-            attn_metadata=attn_metadata,
-            cos=cos,
-            sin=sin,
-            actual_seq_lengths_query=actual_seq_lengths_query,
-            actual_seq_lengths_key=actual_seq_lengths_key,
-        )
+        if self.skip_topk and attn_metadata.shared_topk_indices is not None:
+            topk_indices = attn_metadata.shared_topk_indices
+        else:
+            topk_indices = self.indexer_select_post_process(
+                x=hidden_states,
+                q_c=q_c,
+                kv_cache=kv_cache,
+                attn_metadata=attn_metadata,
+                cos=cos,
+                sin=sin,
+                actual_seq_lengths_query=actual_seq_lengths_query,
+                actual_seq_lengths_key=actual_seq_lengths_key,
+            )
+            attn_metadata.shared_topk_indices = topk_indices
 
         attn_output = self._execute_sparse_flash_attention_process(
             ql_nope, q_pe, kv_cache, topk_indices, attn_metadata, actual_seq_lengths_query, actual_seq_lengths_key
diff --git a/vllm_ascend/ops/mla.py b/vllm_ascend/ops/mla.py
index 689ed0cd672..2b9672d6a12 100644
--- a/vllm_ascend/ops/mla.py
+++ b/vllm_ascend/ops/mla.py
@@ -80,6 +80,7 @@ def __init__(
         cache_config: CacheConfig | None = None,
         quant_config: QuantizationConfig | None = None,
         prefix: str = "",
+        skip_topk: bool = False,
     ) -> None:
         nn.Module.__init__(self)
         self.hidden_size = hidden_size
@@ -122,6 +123,7 @@ def __init__(
             kv_a_layernorm=mla_modules.kv_a_layernorm,
             o_proj=mla_modules.o_proj,
             layer_name=f"{prefix}.attn",
+            skip_topk=skip_topk,
         )
 
         original_process_weights = self.mla_attn.process_weights_after_loading