[HiCache] Merge HiCache event checks to reduce decode overhead (#30511)

Co-authored-by: alphabetc1 <47200617+alphabetc1@users.noreply.github.com>
This commit is contained in:
huangtingwei
2026-07-29 23:48:48 +08:00
committed by GitHub
co-authored by alphabetc1
parent bfc450248e
commit 50029f05a3
8 changed files with 219 additions and 104 deletions
@@ -66,16 +66,15 @@ class TestLoadBackDurationMetric(CustomTestCase):
num_tokens=1024,
timing_enabled=True,
)
stub = SimpleNamespace(
cache_controller=SimpleNamespace(ack_load_queue=[ack]),
ongoing_load_back={1: object(), 2: object()},
dec_lock_ref=MagicMock(),
metrics_collector=MagicMock(),
pp_rank=0,
_all_reduce=MagicMock(),
)
stub = object.__new__(HiRadixCache)
stub.cache_controller = SimpleNamespace(ack_load_queue=[ack])
stub.ongoing_load_back = {1: object(), 2: object()}
stub.dec_lock_ref = MagicMock()
stub.metrics_collector = MagicMock()
stub.pp_rank = 0
stub._all_reduce = MagicMock()
HiRadixCache.loading_check(stub)
stub.loading_check()
stub.metrics_collector.increment_load_back_num_tokens.assert_called_once_with(
1024
@@ -102,16 +101,15 @@ class TestLoadBackDurationMetric(CustomTestCase):
num_tokens=512,
timing_enabled=False,
)
stub = SimpleNamespace(
cache_controller=SimpleNamespace(ack_load_queue=[ack]),
ongoing_load_back={7: object()},
dec_lock_ref=MagicMock(),
metrics_collector=MagicMock(),
pp_rank=0,
_all_reduce=MagicMock(),
)
stub = object.__new__(HiRadixCache)
stub.cache_controller = SimpleNamespace(ack_load_queue=[ack])
stub.ongoing_load_back = {7: object()}
stub.dec_lock_ref = MagicMock()
stub.metrics_collector = MagicMock()
stub.pp_rank = 0
stub._all_reduce = MagicMock()
HiRadixCache.loading_check(stub)
stub.loading_check()
stub.metrics_collector.increment_load_back_num_tokens.assert_called_once_with(
512