Skip to content

vllm.v1.worker.xpu_model_runner

logger module-attribute 

logger = init_logger(__name__)

XPUModelRunner

Bases: GPUModelRunner

A model runner for XPU devices.

Source code in vllm/v1/worker/xpu_model_runner.py 
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
class XPUModelRunner(GPUModelRunner):
    """A model runner for XPU devices."""

    def __init__(
        self,
        vllm_config: VllmConfig,
        device: torch.device,
    ):
        super().__init__(vllm_config, device)
        # FIXME: To be verified.
        self.cascade_attn_enabled = False

    def _init_device_properties(self) -> None:
        pass

    def _sync_device(self) -> None:
        torch.xpu.synchronize()

cascade_attn_enabled instance-attribute 

cascade_attn_enabled = False

__init__ 

__init__(vllm_config: VllmConfig, device: device)
Source code in vllm/v1/worker/xpu_model_runner.py 
20
21
22
23
24
25
26
27
def __init__(
    self,
    vllm_config: VllmConfig,
    device: torch.device,
):
    super().__init__(vllm_config, device)
    # FIXME: To be verified.
    self.cascade_attn_enabled = False

_init_device_properties 

_init_device_properties() -> None
Source code in vllm/v1/worker/xpu_model_runner.py 
29
30
def _init_device_properties(self) -> None:
    pass

_sync_device 

_sync_device() -> None
Source code in vllm/v1/worker/xpu_model_runner.py 
32
33
def _sync_device(self) -> None:
    torch.xpu.synchronize()