[P/D] Using the cache load operator to replace the index select operator. (#6295)

### What this PR does / why we need it?
Using the cache load operator to replace the index select operator.

- vLLM version: v0.14.1
- vLLM main:
dc917cceb8

---------

Signed-off-by: liziyu <liziyu16@huawei.com>
This commit is contained in:
liziyu
2026-01-30 14:27:53 +08:00
committed by GitHub
parent 70cc5f7969
commit d252e4f5ec
2 changed files with 117 additions and 72 deletions

View File

@@ -735,7 +735,8 @@ class TestHelperFunctions(unittest.TestCase):
)
def test_ensure_zmq_send_success(self, _):
mock_socket = MagicMock()
ensure_zmq_send(mock_socket, b"hello")
path = "127.0.0.1:12345"
ensure_zmq_send(mock_socket, b"hello", path)
mock_socket.send.assert_called_once_with(b"hello")
@patch(
@@ -743,10 +744,11 @@ class TestHelperFunctions(unittest.TestCase):
)
def test_ensure_zmq_send_retry_and_fail(self, _):
mock_socket = MagicMock()
path = "127.0.0.1:12345"
mock_socket.send.side_effect = zmq.ZMQError( # type: ignore
"send failed")
with self.assertRaises(RuntimeError):
ensure_zmq_send(mock_socket, b"hello", max_retries=2)
ensure_zmq_send(mock_socket, b"hello", path, max_retries=2)
self.assertEqual(mock_socket.send.call_count, 2)
@patch(
@@ -759,7 +761,8 @@ class TestHelperFunctions(unittest.TestCase):
mock_poller.poll.return_value = [
(mock_socket, zmq.POLLIN) # type: ignore
]
data = ensure_zmq_recv(mock_socket, mock_poller)
path = "127.0.0.1:12345"
data = ensure_zmq_recv(mock_socket, mock_poller, path)
self.assertEqual(data, b"response")
@patch(
@@ -769,9 +772,11 @@ class TestHelperFunctions(unittest.TestCase):
mock_socket = MagicMock()
mock_poller = MagicMock()
mock_poller.poll.return_value = []
path = "127.0.0.1:12345"
with self.assertRaises(RuntimeError):
ensure_zmq_recv(mock_socket,
mock_poller,
path,
timeout=0.01,
max_retries=2)