Signed-off-by: yihong0618 <zouzou0208@gmail.com>
This commit is contained in:
@@ -1143,8 +1143,8 @@ class GPUModelRunnerBase(ModelRunnerBase[TModelInputForGPU]):
|
||||
time_after_load = time.perf_counter()
|
||||
|
||||
self.model_memory_usage = m.consumed_memory
|
||||
logger.info("Model loading took %.4f GB and %.6f seconds",
|
||||
self.model_memory_usage / float(2**30),
|
||||
logger.info("Model loading took %.4f GiB and %.6f seconds",
|
||||
self.model_memory_usage / GiB_bytes,
|
||||
time_after_load - time_before_load)
|
||||
if self.prompt_adapter_config:
|
||||
self.prompt_adapter_manager = LRUCacheWorkerPromptAdapterManager(
|
||||
|
||||
@@ -25,7 +25,7 @@ from vllm.multimodal import (MULTIMODAL_REGISTRY, BatchedTensorInputs,
|
||||
MultiModalRegistry)
|
||||
from vllm.sampling_params import SamplingParams
|
||||
from vllm.sequence import IntermediateTensors, SequenceGroupMetadata
|
||||
from vllm.utils import DeviceMemoryProfiler, make_tensor_with_pad
|
||||
from vllm.utils import DeviceMemoryProfiler, GiB_bytes, make_tensor_with_pad
|
||||
from vllm.worker.model_runner import AttentionMetadata, SamplingMetadata
|
||||
from vllm.worker.model_runner_base import (
|
||||
ModelRunnerBase, ModelRunnerInputBase, ModelRunnerInputBuilderBase,
|
||||
@@ -422,8 +422,8 @@ class XPUModelRunner(ModelRunnerBase[ModelInputForXPUWithSamplingMetadata]):
|
||||
self.model = get_model(vllm_config=self.vllm_config)
|
||||
|
||||
self.model_memory_usage = m.consumed_memory
|
||||
logger.info("Loading model weights took %.4f GB",
|
||||
self.model_memory_usage / float(2**30))
|
||||
logger.info("Loading model weights took %.4f GiB",
|
||||
self.model_memory_usage / GiB_bytes)
|
||||
|
||||
def get_model(self) -> nn.Module:
|
||||
return self.model
|
||||
|
||||
Reference in New Issue
Block a user