[hardware][misc] introduce platform abstraction (#6080)

This commit is contained in:
youkaichao
2024-07-02 20:12:22 -07:00
committed by GitHub
parent 9d6a8daa87
commit 482045ee77
16 changed files with 113 additions and 29 deletions
+3 -2
View File
@@ -35,7 +35,8 @@ from vllm.model_executor.model_loader.weight_utils import (
from vllm.model_executor.models.interfaces import (supports_lora,
supports_vision)
from vllm.model_executor.utils import set_weight_attrs
from vllm.utils import get_device_capability_stateless, is_tpu
from vllm.platforms import current_platform
from vllm.utils import is_tpu
logger = init_logger(__name__)
@@ -46,7 +47,7 @@ def _get_quantization_config(
"""Get the quantization config."""
if model_config.quantization is not None:
quant_config = get_quant_config(model_config, load_config)
capability = get_device_capability_stateless()
capability = current_platform.get_device_capability()
capability = capability[0] * 10 + capability[1]
if capability < quant_config.get_min_capability():
raise ValueError(