[core][distributed] exact ray placement control (vllm-project#12732)

Signed-off-by: youkaichao <[email protected]>
Signed-off-by: Felix Marty <[email protected]>

.buildkite/test-pipeline.yaml

@@ -128,6 +128,7 @@ steps:
   - tests/spec_decode/e2e/test_integration_dist_tp4
   - tests/compile
   - examples/offline_inference/rlhf.py
+  - examples/offline_inference/ray_placement.py
   commands:
   - pytest -v -s distributed/test_utils.py
   - pytest -v -s compile/test_basic_correctness.py
@@ -136,6 +137,7 @@ steps:
   # TODO: create a dedicated test section for multi-GPU example tests
   # when we have multiple distributed example tests
   - python3 ../examples/offline_inference/rlhf.py
+  - RAY_DEDUP_LOGS=0 python3 ../examples/offline_inference/ray_placement.py
 
 - label: Metrics, Tracing Test # 10min
   num_gpus: 2 

examples/offline_inference/ray_placement.py

@@ -0,0 +1,121 @@
+# SPDX-License-Identifier: Apache-2.0
+"""
+a simple demonstration to show how to control
+the placement of the vLLM workers with Ray.
+The key is to set VLLM_RAY_PER_WORKER_GPUS and
+VLLM_RAY_BUNDLE_INDICES properly.
+"""
+import os
+
+import ray
+from ray.util.placement_group import placement_group
+from ray.util.scheduling_strategies import PlacementGroupSchedulingStrategy
+
+from vllm import LLM
+from vllm.worker.worker import Worker
+
+
+class MyWorker(Worker):
+
+    def report_device_id(self) -> str:
+        from vllm.platforms import current_platform
+        return current_platform.get_device_uuid(self.device.index)
+
+
+class MyLLM(LLM):
+
+    def __init__(self, *args, bundle_indices: list, **kwargs):
+        # a hack to make the script work.
+        # stop ray from manipulating CUDA_VISIBLE_DEVICES
+        # at the top-level
+        del os.environ["CUDA_VISIBLE_DEVICES"]
+        # every worker will use 0.4 GPU, so that we can schedule
+        # 2 instances on the same GPUs.
+        os.environ["VLLM_RAY_PER_WORKER_GPUS"] = "0.4"
+        os.environ["VLLM_RAY_BUNDLE_INDICES"] = ",".join(
+            map(str, bundle_indices))
+        print(f"creating LLM with bundle_indices={bundle_indices}")
+        super().__init__(*args, **kwargs)
+
+
+class RayTrainingActor:
+
+    def report_device_id(self) -> str:
+        # the argument for get_device_uuid is the index
+        # of the GPU in the visible devices.
+        # ray will set CUDA_VISIBLE_DEVICES to the assigned GPUs
+        from vllm.platforms import current_platform
+        return current_platform.get_device_uuid(0)
+
+
+# ray manages 4 GPUs
+os.environ["CUDA_VISIBLE_DEVICES"] = "0,1,2,3"
+ray.init()
+
+# we want to co-locate vLLM instance and the training actor
+# on the same set of GPUs.
+# the placement plan is as follows:
+# GPU 0 and 1: training actor 0, 1, and vLLM instance 0 (with TP=2)
+# GPU 2 and 3: training actor 2, 3, and vLLM instance 1 (with TP=2)
+
+pg = placement_group([{"GPU": 1, "CPU": 0}] * 4)
+ray.get(pg.ready())
+print(f"placement group has bundles {pg.bundle_specs=}")
+
+training_actors = []
+training_actor_device_ids = []
+inference_engines = []
+inference_engine_device_ids = []
+
+for bundle_index in [0, 1, 2, 3]:
+    training_actor = ray.remote(
+        num_cpus=0,
+        num_gpus=0.4,
+        scheduling_strategy=PlacementGroupSchedulingStrategy(
+            placement_group=pg,
+            placement_group_capture_child_tasks=True,
+            placement_group_bundle_index=bundle_index,
+        ),
+    )(RayTrainingActor).remote()
+    training_actors.append(training_actor)
+    device_id = ray.get(training_actor.report_device_id.remote())
+    print(f"training actor {bundle_index} is on {device_id}")
+    training_actor_device_ids.append(device_id)
+
+for (i, bundle_indices) in enumerate([[0, 1], [2, 3]]):
+    # IMPORTANT: when creating vLLM instances, we need to
+    # make sure there are no GPU activities on the target GPUs,
+    # otherwise, they will interfere with the vLLM memory profiling,
+    # and cause unexpected behaviors.
+    llm = ray.remote(
+        num_cpus=0,
+        num_gpus=0,
+        scheduling_strategy=PlacementGroupSchedulingStrategy(
+            placement_group=pg,
+            placement_group_capture_child_tasks=True,
+        ),
+    )(MyLLM).remote(
+        model="facebook/opt-125m",
+        enforce_eager=True,
+        worker_cls=MyWorker,
+        tensor_parallel_size=2,
+        distributed_executor_backend="ray",
+        gpu_memory_utilization=0.4,
+        bundle_indices=bundle_indices,
+    )
+    inference_engines.append(llm)
+    # don't call any method on the inference engine here,
+    # otherwise it will block until the vLLM instance is created.
+
+for i, llm in enumerate(inference_engines):
+    inference_engine_device_ids.append(
+        ray.get(llm.collective_rpc.remote("report_device_id", args=tuple())))
+    print(f"inference engine {i} is on {inference_engine_device_ids[-1]}")
+
+# check the placement
+# the first two training actors should be
+# on the same GPUs as the first inference engine
+assert training_actor_device_ids[:2] == inference_engine_device_ids[0]
+# the last two training actors should be
+# on the same GPUs as the second inference engine
+assert training_actor_device_ids[2:] == inference_engine_device_ids[1]

vllm/envs.py

@@ -85,6 +85,8 @@
     VLLM_MLA_DISABLE_REQUANTIZATION: bool = False
     VLLM_MLA_CUDA_MEM_ALIGN_KV_CACHE: bool = True
     VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON: bool = False
+    VLLM_RAY_PER_WORKER_GPUS: float = 1.0
+    VLLM_RAY_BUNDLE_INDICES: str = ""
 
 
 def get_default_cache_root():
@@ -550,6 +552,18 @@ def maybe_convert_int(value: Optional[str]) -> Optional[int]:
     lambda: bool(int(os.getenv("VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON", "0"))
                  ),
 
+    # Number of GPUs per worker in Ray, if it is set to be a fraction,
+    # it allows ray to schedule multiple actors on a single GPU,
+    # so that users can colocate other actors on the same GPUs as vLLM.
+    "VLLM_RAY_PER_WORKER_GPUS":
+    lambda: float(os.getenv("VLLM_RAY_PER_WORKER_GPUS", "1.0")),
+
+    # Bundle indices for Ray, if it is set, it can control precisely
+    # which indices are used for the Ray bundle, for every worker.
+    # Format: comma-separated list of integers, e.g. "0,1,2,3"
+    "VLLM_RAY_BUNDLE_INDICES":
+    lambda: os.getenv("VLLM_RAY_BUNDLE_INDICES", ""),
+
     # When on a Nvidia GPU aligns single entries (within a page) so they are 256
     # byte aligned for better performance, this increases the memory usage of
     # the cache. Currently this only affects MLA that results in non-256

vllm/executor/ray_distributed_executor.py

@@ -129,13 +129,7 @@ def _get_env_vars_to_be_updated(self):
 
     def _init_workers_ray(self, placement_group: "PlacementGroup",
                           **ray_remote_kwargs):
-        if (self.parallel_config.tensor_parallel_size == 1
-                and self.parallel_config.pipeline_parallel_size == 1):
-            # For single GPU case, we use a ray worker with constrained memory.
-            num_gpus = self.cache_config.gpu_memory_utilization
-        else:
-            # Otherwise, the ray workers are allocated with a full GPU.
-            num_gpus = 1
+        num_gpus = envs.VLLM_RAY_PER_WORKER_GPUS
 
         # The driver dummy worker does not actually use any resources.
         # It holds the resource for the driver worker.
@@ -155,12 +149,29 @@ def _init_workers_ray(self, placement_group: "PlacementGroup",
         logger.info("use_ray_spmd_worker: %s", self.use_ray_spmd_worker)
 
         # Create the workers.
-        driver_ip = get_ip()
-        rank = 0
+        bundle_indices: List[int]
+        if envs.VLLM_RAY_BUNDLE_INDICES:
+            # Use the bundle indices specified by the user.
+            bundle_indices = list(
+                map(int, envs.VLLM_RAY_BUNDLE_INDICES.split(",")))
+            assert len(bundle_indices) == self.parallel_config.world_size, \
+            ("VLLM_RAY_BUNDLE_INDICES must have the same size"
+            f" as the world size, but got {bundle_indices=} "
+            f"and {self.parallel_config.world_size=}")
+            assert len(set(bundle_indices)) == len(bundle_indices), \
+            ("VLLM_RAY_BUNDLE_INDICES cannot have duplicate values,"
+            f" but got {bundle_indices=}")
+        else:
+            # use the first N bundles that have GPU resources.
+            bundle_indices = []
+            for bundle_id, bundle in enumerate(placement_group.bundle_specs):
+                if bundle.get(current_platform.ray_device_key, 0):
+                    bundle_indices.append(bundle_id)
+            bundle_indices = bundle_indices[:self.parallel_config.world_size]
+
         worker_metadata: List[RayWorkerMetaData] = []
-        for bundle_id, bundle in enumerate(placement_group.bundle_specs):
-            if not bundle.get(current_platform.ray_device_key, 0):
-                continue
+        driver_ip = get_ip()
+        for rank, bundle_id in enumerate(bundle_indices):
             scheduling_strategy = PlacementGroupSchedulingStrategy(
                 placement_group=placement_group,
                 placement_group_capture_child_tasks=True,
@@ -187,7 +198,6 @@ def _init_workers_ray(self, placement_group: "PlacementGroup",
                                            rpc_rank=rank)
             worker_metadata.append(
                 RayWorkerMetaData(worker=worker, created_rank=rank))
-            rank += 1
 
         worker_ips = ray.get([
             each.worker.get_node_ip.remote()  # type: ignore[attr-defined]

vllm/platforms/cuda.py

@@ -275,6 +275,14 @@ def get_device_name(cls, device_id: int = 0) -> str:
         physical_device_id = device_id_to_physical_device_id(device_id)
         return cls._get_physical_device_name(physical_device_id)
 
+    @classmethod
+    @lru_cache(maxsize=8)
+    @with_nvml_context
+    def get_device_uuid(cls, device_id: int = 0) -> str:
+        physical_device_id = device_id_to_physical_device_id(device_id)
+        handle = pynvml.nvmlDeviceGetHandleByIndex(physical_device_id)
+        return pynvml.nvmlDeviceGetUUID(handle)
+
     @classmethod
     @lru_cache(maxsize=8)
     @with_nvml_context

vllm/platforms/interface.py

@@ -183,6 +183,11 @@ def get_device_name(cls, device_id: int = 0) -> str:
         """Get the name of a device."""
         raise NotImplementedError
 
+    @classmethod
+    def get_device_uuid(cls, device_id: int = 0) -> str:
+        """Get the uuid of a device, e.g. the PCI bus ID."""
+        raise NotImplementedError
+
     @classmethod
     def get_device_total_memory(cls, device_id: int = 0) -> int:
         """Get the total memory of a device in bytes."""