[MISC] rename CudaMemoryProfiler to DeviceMemoryProfiler (#8703)

vllm/utils.py

@@ -757,7 +757,7 @@ def is_pin_memory_available() -> bool:
     return True
 
 
-class CudaMemoryProfiler:
+class DeviceMemoryProfiler:
 
     def __init__(self, device: Optional[torch.types.Device] = None):
         self.device = device

vllm/worker/model_runner.py

@@ -45,7 +45,7 @@ from vllm.prompt_adapter.worker_manager import (
     LRUCacheWorkerPromptAdapterManager)
 from vllm.sampling_params import SamplingParams
 from vllm.sequence import IntermediateTensors, SequenceGroupMetadata
-from vllm.utils import (CudaMemoryProfiler, PyObjectCache, async_tensor_h2d,
+from vllm.utils import (DeviceMemoryProfiler, PyObjectCache, async_tensor_h2d,
                         flatten_2d_lists, is_hip, is_pin_memory_available,
                         supports_dynamo)
 from vllm.worker.model_runner_base import (
@@ -1012,7 +1012,7 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
 
     def load_model(self) -> None:
         logger.info("Starting to load model %s...", self.model_config.model)
-        with CudaMemoryProfiler() as m:
+        with DeviceMemoryProfiler() as m:
             self.model = get_model(model_config=self.model_config,
                                    device_config=self.device_config,
                                    load_config=self.load_config,

vllm/worker/xpu_model_runner.py

@@ -21,7 +21,7 @@ from vllm.multimodal import (MULTIMODAL_REGISTRY, BatchedTensorInputs,
                              MultiModalInputs, MultiModalRegistry)
 from vllm.sampling_params import SamplingParams
 from vllm.sequence import IntermediateTensors, SequenceGroupMetadata
-from vllm.utils import CudaMemoryProfiler, make_tensor_with_pad
+from vllm.utils import DeviceMemoryProfiler, make_tensor_with_pad
 from vllm.worker.model_runner import AttentionMetadata, SamplingMetadata
 from vllm.worker.model_runner_base import (
     ModelRunnerBase, ModelRunnerInputBase, ModelRunnerInputBuilderBase,
@@ -391,7 +391,7 @@ class XPUModelRunner(ModelRunnerBase[ModelInputForXPUWithSamplingMetadata]):
         self.model: nn.Module  # Set after init_Model
 
     def load_model(self) -> None:
-        with CudaMemoryProfiler() as m:
+        with DeviceMemoryProfiler() as m:
             self.model = get_model(
                 model_config=self.model_config,
                 device_config=self.device_config,