xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-10 05:00:07 +08:00
Code Issues Packages Projects Releases Wiki Activity

update

Browse Source 
Signed-off-by: rshaw@neuralmagic.com <robertgshaw2@gmail.com>
This commit is contained in:
rshaw@neuralmagic.com 2025-06-30 01:39:15 +00:00
parent dedb1a5424
commit c481d30c17
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
View File

@ -333,7 +333,8 @@ class NixlConnectorWorker:
self.nixl_wrapper = NixlWrapper(str(uuid.uuid4()), self.nixl_wrapper = NixlWrapper(str(uuid.uuid4()),
None, None,
num_workers=None, num_workers=None,
num_shared_workers=16) # num_shared_workers=16) # setting this > 0 causes the notifs to be recved
num_shared_workers=None)
# Map of engine_id -> {rank0: agent_name0, rank1: agent_name1..}. # Map of engine_id -> {rank0: agent_name0, rank1: agent_name1..}.
self._remote_agents: dict[str, dict[int, str]] = defaultdict(dict) self._remote_agents: dict[str, dict[int, str]] = defaultdict(dict)
@ -819,6 +820,7 @@ class NixlConnectorWorker:
""" """
notified_req_ids: set[str] = set() notified_req_ids: set[str] = set()
for notifs in self.nixl_wrapper.get_new_notifs().values(): for notifs in self.nixl_wrapper.get_new_notifs().values():
print(f"{notifs=}")
for notif in notifs: for notif in notifs:
req_id, tp_ratio = notif.decode("utf-8").rsplit(":", 1) req_id, tp_ratio = notif.decode("utf-8").rsplit(":", 1)
self.consumer_notification_counts_by_req[req_id] += 1 self.consumer_notification_counts_by_req[req_id] += 1
@ -853,6 +855,7 @@ class NixlConnectorWorker:
xfer_state) xfer_state)
# Done. # Done.
print(f"{len(new_handles)=}")
if len(new_handles) == 0: if len(new_handles) == 0:
start = time.perf_counter() start = time.perf_counter()
self.nixl_wrapper.send_notif(agent_name, notif_id) self.nixl_wrapper.send_notif(agent_name, notif_id)