diff --git a/qwen3_6_scripts/paged_attn.py b/qwen3_6_scripts/paged_attn.py
index 23118933..00525269 100644
--- a/qwen3_6_scripts/paged_attn.py
+++ b/qwen3_6_scripts/paged_attn.py
@@ -2290,10 +2290,8 @@ class PagedAttention:
         kv_caches: List[torch.Tensor],
         src_to_dists: torch.Tensor,
     ) -> None:
-        key_caches = [kv_cache[0] for kv_cache in kv_caches]
-        value_caches = [kv_cache[1] for kv_cache in kv_caches]
-        # BI100 CoreX 3.2.3: ixformer exposes vllm_copy_blocks, not
-        # copy_blocks.  Call the vendor symbol directly instead of going
-        # through ops.copy_blocks (which hits the missing name).
-        import ixformer.functions as _ixf
-        _ixf.vllm_copy_blocks(key_caches, value_caches, src_to_dists)
\ No newline at end of file
+        for kv_cache in kv_caches:
+            src = src_to_dists[:, 0].long()
+            dst = src_to_dists[:, 1].long()
+            kv_cache[0][dst] = kv_cache[0][src]
+            kv_cache[1][dst] = kv_cache[1][src]
\ No newline at end of file