[MISC][Bugfix] Use less CPU when message queue has been empty for some time (#16226)

Signed-off-by: Povilas Kanapickas <povilas@radix.lt>

tests/basic_correctness/test_basic_correctness.py

@@ -128,15 +128,21 @@ def test_models(
 @multi_gpu_test(num_gpus=2)
 @pytest.mark.parametrize(
     "model, distributed_executor_backend, attention_backend, "
-    "test_suite", [
+    "test_suite, extra_env", [
-        ("distilbert/distilgpt2", "ray", "", "L4"),
+        ("distilbert/distilgpt2", "ray", "", "L4", {}),
-        ("distilbert/distilgpt2", "mp", "", "L4"),
+        ("distilbert/distilgpt2", "mp", "", "L4", {}),
-        ("meta-llama/Llama-3.2-1B-Instruct", "ray", "", "L4"),
+        ("distilbert/distilgpt2", "ray", "", "L4", {
-        ("meta-llama/Llama-3.2-1B-Instruct", "mp", "", "L4"),
+            "VLLM_SLEEP_WHEN_IDLE": "1"
-        ("distilbert/distilgpt2", "ray", "", "A100"),
+        }),
-        ("distilbert/distilgpt2", "mp", "", "A100"),
+        ("distilbert/distilgpt2", "mp", "", "L4", {
-        ("distilbert/distilgpt2", "mp", "FLASHINFER", "A100"),
+            "VLLM_SLEEP_WHEN_IDLE": "1"
-        ("meta-llama/Meta-Llama-3-8B", "ray", "FLASHINFER", "A100"),
+        }),
+        ("meta-llama/Llama-3.2-1B-Instruct", "ray", "", "L4", {}),
+        ("meta-llama/Llama-3.2-1B-Instruct", "mp", "", "L4", {}),
+        ("distilbert/distilgpt2", "ray", "", "A100", {}),
+        ("distilbert/distilgpt2", "mp", "", "A100", {}),
+        ("distilbert/distilgpt2", "mp", "FLASHINFER", "A100", {}),
+        ("meta-llama/Meta-Llama-3-8B", "ray", "FLASHINFER", "A100", {}),
     ])
 @pytest.mark.parametrize("enable_prompt_embeds", [True, False])
 def test_models_distributed(
@@ -148,6 +154,7 @@ def test_models_distributed(
     distributed_executor_backend: str,
     attention_backend: str,
     test_suite: str,
+    extra_env: dict[str, str],
     enable_prompt_embeds: bool,
 ) -> None:
 
@@ -173,6 +180,9 @@ def test_models_distributed(
                 attention_backend,
             )
 
+        for k, v in extra_env.items():
+            monkeypatch_context.setenv(k, v)
+
         dtype = "half"
         max_tokens = 5
 

vllm/distributed/device_communicators/shm_broadcast.py

@@ -28,6 +28,43 @@ VLLM_RINGBUFFER_WARNING_INTERVAL = envs.VLLM_RINGBUFFER_WARNING_INTERVAL
 logger = init_logger(__name__)
 
 
+class SpinTimer:
+
+    def record_activity(self):
+        pass
+
+    def spin(self):
+        sched_yield()
+
+
+class SpinSleepTimer(SpinTimer):
+    """
+    In setups which have long inactivity periods it is desirable to reduce
+    system power consumption when vllm does nothing. This would lead to more
+    CPU thermal headroom when a request eventually comes, especially when
+    multiple GPUs are connected as each GPU would otherwise pin one thread at
+    100% CPU usage.
+
+    The simplest solution is to reduce polling frequency when there is no
+    activity for a certain period of time.
+    """
+
+    def __init__(self, busy_loop_s: float = 3.0, wait_sleep_s: float = 0.1):
+        self.last_activity = time.monotonic()
+        self.busy_loop_s = busy_loop_s
+        self.wait_sleep_s = wait_sleep_s
+
+    def record_activity(self):
+        self.last_activity = time.monotonic()
+
+    def spin(self):
+        curr_time = time.monotonic()
+        if curr_time >= self.last_activity + self.busy_loop_s:
+            time.sleep(self.wait_sleep_s)
+        else:
+            sched_yield()
+
+
 class ShmRingBuffer:
 
     def __init__(self,
@@ -238,6 +275,7 @@ class MessageQueue:
         self.local_reader_rank = -1
         # rank does not matter for remote readers
         self._is_remote_reader = False
+        self._read_spin_timer = SpinTimer()
 
         self.handle = Handle(
             local_reader_ranks=local_reader_ranks,
@@ -276,6 +314,9 @@ class MessageQueue:
             self.local_socket.connect(socket_addr)
 
             self.remote_socket = None
+
+            self._read_spin_timer = SpinSleepTimer(
+            ) if envs.VLLM_SLEEP_WHEN_IDLE else SpinTimer()
         else:
             self.buffer = None  # type: ignore
             self.current_idx = -1
@@ -407,7 +448,7 @@ class MessageQueue:
                     # we need to wait until it is written
 
                     # Release the processor to other threads
-                    sched_yield()
+                    self._read_spin_timer.spin()
 
                     # if we wait for a long time, log a message
                     if (time.monotonic() - start_time
@@ -438,6 +479,8 @@ class MessageQueue:
                 metadata_buffer[self.local_reader_rank + 1] = 1
                 self.current_idx = (self.current_idx +
                                     1) % self.buffer.max_chunks
+
+                self._read_spin_timer.record_activity()
                 break
 
     def enqueue(self, obj, timeout: Optional[float] = None):

vllm/envs.py

@@ -122,6 +122,7 @@ if TYPE_CHECKING:
     VLLM_ALL2ALL_BACKEND: str = "naive"
     VLLM_MAX_TOKENS_PER_EXPERT_FP4_MOE: int = 163840
     VLLM_TOOL_PARSE_REGEX_TIMEOUT_SECONDS: int = 1
+    VLLM_SLEEP_WHEN_IDLE: bool = False
 
 
 def get_default_cache_root():
@@ -841,6 +842,11 @@ environment_variables: dict[str, Callable[[], Any]] = {
     # Regex timeout for use by the vLLM tool parsing plugins.
     "VLLM_TOOL_PARSE_REGEX_TIMEOUT_SECONDS":
     lambda: int(os.getenv("VLLM_TOOL_PARSE_REGEX_TIMEOUT_SECONDS", "1")),
+
+    # Reduce CPU usage when vLLM is idle. Enabling this will incur small
+    # latency penalty when a request eventually comes.
+    "VLLM_SLEEP_WHEN_IDLE":
+    lambda: bool(int(os.getenv("VLLM_SLEEP_WHEN_IDLE", "0"))),
 }
 
 # --8<-- [end:env-vars-definition]