xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-11 13:42:20 +08:00
Code Issues Packages Projects Releases Wiki Activity

updated

Browse Source 
Signed-off-by: rshaw@neuralmagic.com <robertgshaw2@gmail.com>
This commit is contained in:
rshaw@neuralmagic.com 2025-06-30 01:45:03 +00:00
parent c481d30c17
commit 8283d7b85c
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
View File

@ -333,8 +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) # setting this > 0 causes the notifs to be recved num_shared_workers=16) # setting this > 0 causes the notifs to be recved
num_shared_workers=None) # 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)
@ -864,7 +864,7 @@ class NixlConnectorWorker:
end = time.perf_counter() end = time.perf_counter()
print(f"========= SEND NOTIF TIME: {end - start} =========") print(f"========= SEND NOTIF TIME: {end - start} =========")
else: else:
transfers[req_id] = (new_handles, notif_id, agent_name) transfers[req_id] = (new_handles, agent_name, notif_id)
return done_req_ids return done_req_ids
@ -977,7 +977,7 @@ class NixlConnectorWorker:
assert len(local_block_descs_ids) == len(remote_block_descs_ids) assert len(local_block_descs_ids) == len(remote_block_descs_ids)
# Prepare transfer with Nixl. # Prepare transfer with Nixl.
CHUNK_SIZE = 1000 CHUNK_SIZE = 100
handles = [] handles = []
for i in range(0, len(local_block_descs_ids), CHUNK_SIZE): for i in range(0, len(local_block_descs_ids), CHUNK_SIZE):
handles.append( handles.append(