Revert "[Disagg][Perf] Use NPU event sync instead of blocking tolist (#3194)

…to avoid unintentional copy ops blocking across different NPU streams,
improving disagg TTIT/TTFT (#2788)"



### What this PR does / why we need it?
This reverts commit 6995a7bc5b. We'll add
it back once the issue is fixed.

related issue: https://github.com/vllm-project/vllm-ascend/issues/3195

### How was this patch tested?

- vLLM version: v0.10.2
- vLLM main:
52d0cb8458

vllm_ascend/worker/model_runner_v1.py

@@ -248,7 +248,6 @@ class NPUModelRunner(LoRAModelRunnerMixin):
         self.block_size = vllm_config.cache_config.block_size
         self.max_num_blocks_per_req = cdiv(self.model_config.max_model_len,
                                            self.block_size)
-        self.max_model_len = self.model_config.max_model_len
         self.max_num_tokens = self.scheduler_config.max_num_batched_tokens
         decode_max_num_seqs = getattr(self.scheduler_config,
                                       'decode_max_num_seqs', 0)
@@ -435,12 +434,6 @@ class NPUModelRunner(LoRAModelRunnerMixin):
         # Cached outputs.
         self._draft_token_ids: Optional[Union[list[list[int]],
                                               torch.Tensor]] = None
-        self.transfer_event = torch_npu.npu.Event()
-        self.sampled_token_ids_pinned_cpu = torch.empty(
-            (self.max_model_len, 1),
-            dtype=torch.int64,
-            device="cpu",
-            pin_memory=True)
 
         # NOTE: we need to use `in_profile_run` to determine whether `enable_force_load_balance` is True
         self.in_profile_run = False
@@ -2095,7 +2088,7 @@ class NPUModelRunner(LoRAModelRunnerMixin):
                 max_gen_len = sampled_token_ids.shape[-1]
                 if max_gen_len == 1:
                     # No spec decode tokens.
-                    valid_sampled_token_ids = self._to_list(sampled_token_ids)
+                    valid_sampled_token_ids = sampled_token_ids.tolist()
                 else:
                     # Includes spec decode tokens.
                     valid_sampled_token_ids = self.rejection_sampler.parse_output(
@@ -3547,18 +3540,3 @@ class NPUModelRunner(LoRAModelRunnerMixin):
 
     def _build_drafter_prepare_inputs_torchair_param(self):
         return False
-
-    def _to_list(self, sampled_token_ids: torch.Tensor) -> list[list[int]]:
-        # This is a short term mitigation for issue mentioned in
-        # https://github.com/vllm-project/vllm/issues/22754.
-        # `tolist` would trigger a npu wise stream sync, which
-        # would block other copy ops from other npu streams.
-        # A npu event sync would avoid such a situation. Since
-        # this is in the critical path of every single model
-        # forward loop, this has caused perf issue for a disagg
-        # setup.
-        pinned = self.sampled_token_ids_pinned_cpu[:sampled_token_ids.shape[0]]
-        pinned.copy_(sampled_token_ids, non_blocking=True)
-        self.transfer_event.record()
-        self.transfer_event.synchronize()
-        return pinned.tolist()