[Hardware][openvino] is_openvino --> current_platform.is_openvino (#9716)

vllm/platforms/__init__.py

@@ -65,6 +65,13 @@ try:
 except ImportError:
     pass
 
+is_openvino = False
+try:
+    from importlib.metadata import version
+    is_openvino = "openvino" in version("vllm")
+except Exception:
+    pass
+
 if is_tpu:
     # people might install pytorch built with cuda but run on tpu
     # so we need to check tpu first
@@ -85,6 +92,9 @@ elif is_cpu:
 elif is_neuron:
     from .neuron import NeuronPlatform
     current_platform = NeuronPlatform()
+elif is_openvino:
+    from .openvino import OpenVinoPlatform
+    current_platform = OpenVinoPlatform()
 else:
     current_platform = UnspecifiedPlatform()
 

vllm/platforms/interface.py

@@ -11,6 +11,7 @@ class PlatformEnum(enum.Enum):
     XPU = enum.auto()
     CPU = enum.auto()
     NEURON = enum.auto()
+    OPENVINO = enum.auto()
     UNSPECIFIED = enum.auto()
 
 
@@ -52,6 +53,9 @@ class Platform:
     def is_neuron(self) -> bool:
         return self._enum == PlatformEnum.NEURON
 
+    def is_openvino(self) -> bool:
+        return self._enum == PlatformEnum.OPENVINO
+
     def is_cuda_alike(self) -> bool:
         """Stateless version of :func:`torch.cuda.is_available`."""
         return self._enum in (PlatformEnum.CUDA, PlatformEnum.ROCM)

vllm/platforms/openvino.py

@@ -0,0 +1,31 @@
+import torch
+
+import vllm.envs as envs
+from vllm.utils import print_warning_once
+
+from .interface import Platform, PlatformEnum
+
+
+class OpenVinoPlatform(Platform):
+    _enum = PlatformEnum.OPENVINO
+
+    @classmethod
+    def get_device_name(self, device_id: int = 0) -> str:
+        return "openvino"
+
+    @classmethod
+    def inference_mode(self):
+        return torch.inference_mode(mode=True)
+
+    @classmethod
+    def is_openvino_cpu(self) -> bool:
+        return "CPU" in envs.VLLM_OPENVINO_DEVICE
+
+    @classmethod
+    def is_openvino_gpu(self) -> bool:
+        return "GPU" in envs.VLLM_OPENVINO_DEVICE
+
+    @classmethod
+    def is_pin_memory_available(self) -> bool:
+        print_warning_once("Pin memory is not supported on OpenViNO.")
+        return False