[bugfix] Fix mooncake kvpool accuracy issue (#4976)

### What this PR does / why we need it?

The current KVPool has a accuracy issue
https://github.com/vllm-project/vllm-ascend/issues/4412. This PR aims to
fix the precision problem without impacting prefill performance.

Note：Due to a bug in ADXL, calling `current_event.synchronize()` may
occasionally hang. This issue will be fixed in Cann version 8.5.rc1. You
can manually build the master branch of the project at
https://gitcode.com/cann/hixl to resolve this issue before the 8.5.RC1
release.


- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: LCAIZJ <leichao139636@163.com>

vllm_ascend/distributed/kvpool/config_data.py

@@ -1,6 +1,7 @@
 from dataclasses import dataclass
 from typing import Iterable, List, Optional, Tuple, Union
 
+import torch
 from vllm.distributed.kv_transfer.kv_connector.v1.base import \
     KVConnectorMetadata
 from vllm.logger import logger
@@ -284,6 +285,8 @@ class ReqMeta:
 
     is_last_chunk: Optional[bool] = None
 
+    current_event: Optional[torch.npu.Event] = None
+
     @staticmethod
     def from_request_tracker(
         tracker: RequestTracker,
@@ -375,3 +378,4 @@ class LasyerMultiBlockReqMeta:
     block_ids: list[int]
     layer_id: int
     is_last_chunk: Optional[bool] = True
+    current_event: Optional[torch.npu.Event] = None

vllm_ascend/distributed/kvpool/kv_transfer.py

@@ -114,6 +114,7 @@ class KVCacheStoreSendingThread(KVTransferThread):
         block_ids = req_meta.block_ids
         req_id = req_meta.req_id
         is_last_chunk = req_meta.is_last_chunk
+        current_event = req_meta.current_event
         starts = []
         ends = []
         keys = []
@@ -161,6 +162,14 @@ class KVCacheStoreSendingThread(KVTransferThread):
             addrs.append(addr)
             sizes.append(size)
         if keys:
+            """
+            Note: Due to a bug in ADXL, calling current_event.synchronize() may occasionally hang.
+            This issue will be fixed in CANN version 8.5.rc1.
+            You can manually build the master branch of the project at https://gitcode.com/cann/hixl
+            to resolve this issue before the 8.5.RC1 release.
+            """
+            if current_event is not None:
+                current_event.synchronize()
             self.m_store.put(keys, addrs, sizes)
 
         if is_last_chunk:
@@ -235,6 +244,7 @@ class KVCacheStoreLayerSendingThread(KVTransferThread):
         ends = req_meta.ends
         keys = req_meta.keys
         layer_id = req_meta.layer_id
+        current_event = req_meta.current_event
         total_block = len(keys)
         is_last_chunk = req_meta.is_last_chunk
         if not self.dcp_size > 1:
@@ -270,6 +280,8 @@ class KVCacheStoreLayerSendingThread(KVTransferThread):
             addr_list.append(addr)
             size_list.append(size)
 
+        if current_event is not None:
+            current_event.synchronize()
         self.m_store.put(key_list, addr_list, size_list)
 
         if layer_id == self.final_layer_id and is_last_chunk:

vllm_ascend/distributed/kvpool/pool_worker.py

@@ -251,12 +251,20 @@ class KVPoolWorker:
                       connector_metadata: AscendConnectorMetadata) -> None:
         if self.current_layer == 0:
             self.layerwise_storers = []
+            current_event = None
+            for request in connector_metadata.requests:
+                can_save = request.can_save
+                if can_save is None or not can_save:
+                    continue
+                current_event = torch.npu.Event()
+                current_event.record()
+                break
             for request in connector_metadata.requests:
                 can_save = request.can_save
                 if can_save is None or not can_save:
                     continue
 
-                layerwise_storer = self.store_layer(request)
+                layerwise_storer = self.store_layer(request, current_event)
                 self.layerwise_storers.append(layerwise_storer)
         for layerwise_storer in self.layerwise_storers:
             try:
@@ -266,11 +274,21 @@ class KVPoolWorker:
         self.current_layer = self.current_layer + 1
 
     def wait_for_save(self, connector_metadata: AscendConnectorMetadata):
+        current_event = None
+        for request in connector_metadata.requests:
+            can_save = request.can_save
+            if can_save is None or not can_save:
+                continue
+            current_event = torch.npu.Event()
+            current_event.record()
+            break
+
         for request in connector_metadata.requests:
             can_save = request.can_save
             if can_save is None or not can_save:
                 continue
 
+            request.current_event = current_event
             self.kv_send_thread.add_request(  # type: ignore[union-attr]
                 request, )
 
@@ -347,6 +365,7 @@ class KVPoolWorker:
     def store_layer(
         self,
         request: ReqMeta,
+        current_event: Optional[torch.npu.Event],
     ) -> Generator[None, None, None]:
         """
         Store the KV cache in a layerwise manner.
@@ -385,7 +404,8 @@ class KVPoolWorker:
                                                    keys_multi_chunk, starts,
                                                    ends, request.block_ids,
                                                    layer_id,
-                                                   request.is_last_chunk)
+                                                   request.is_last_chunk,
+                                                   current_event)
                 self.kv_send_thread.add_request(  # type: ignore[union-attr, call-arg]
                     req_meta)  # type: ignore[union-attr, call-arg, arg-type]
                 yield