diff --git a/vllm/v1/core/sched/scheduler.py b/vllm/v1/core/sched/scheduler.py index f6e96c6774..40b5899f04 100644 --- a/vllm/v1/core/sched/scheduler.py +++ b/vllm/v1/core/sched/scheduler.py @@ -1961,7 +1961,6 @@ class Scheduler(SchedulerInterface): num_waiting_reqs=len(self.waiting), num_skipped_waiting_reqs=len(self.skipped_waiting), kv_cache_usage=self.kv_cache_manager.usage, - encoder_cache_usage=self._get_encoder_cache_usage(), prefix_cache_stats=prefix_cache_stats, connector_prefix_cache_stats=connector_prefix_cache_stats, kv_cache_eviction_events=eviction_events, @@ -1971,14 +1970,6 @@ class Scheduler(SchedulerInterface): perf_stats=perf_stats, ) - def _get_encoder_cache_usage(self) -> float: - """Get encoder cache usage as a fraction (0.0 to 1.0).""" - ecm = self.encoder_cache_manager - if ecm.cache_size == 0: - return 0.0 - used_slots = ecm.cache_size - ecm.num_free_slots - return used_slots / ecm.cache_size - def make_spec_decoding_stats( self, spec_decoding_stats: SpecDecodingStats | None, diff --git a/vllm/v1/metrics/stats.py b/vllm/v1/metrics/stats.py index df50b78cc0..78faa73a88 100644 --- a/vllm/v1/metrics/stats.py +++ b/vllm/v1/metrics/stats.py @@ -181,7 +181,6 @@ class SchedulerStats: current_wave: int = 0 kv_cache_usage: float = 0.0 - encoder_cache_usage: float = 0.0 prefix_cache_stats: PrefixCacheStats = field(default_factory=PrefixCacheStats) connector_prefix_cache_stats: PrefixCacheStats | None = None