refactor hard coded device string in test files under tests/v1 and tests/lora (#37566)
Signed-off-by: Liao, Wei <wei.liao@intel.com>
This commit is contained in:
@@ -9,10 +9,13 @@ import vllm.lora.ops.torch_ops as torch_ops
|
||||
import vllm.lora.ops.triton_ops as triton_ops
|
||||
from vllm.lora.ops.triton_ops import LoRAKernelMeta
|
||||
from vllm.lora.ops.triton_ops.utils import _LORA_A_PTR_DICT, _LORA_B_PTR_DICT
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils.torch_utils import set_random_seed
|
||||
|
||||
from .utils import PunicaTensors, assert_close, generate_data_for_nslices
|
||||
|
||||
DEVICE_TYPE = current_platform.device_type
|
||||
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def reset_device(reset_default_device):
|
||||
@@ -146,7 +149,9 @@ def check_lora_shrink_kernel(
|
||||
|
||||
# Setup metadata information for the LoRA kernel.
|
||||
lora_meta = LoRAKernelMeta.make(
|
||||
max_loras=num_loras, max_num_tokens=token_nums, device="cuda"
|
||||
max_loras=num_loras,
|
||||
max_num_tokens=token_nums,
|
||||
device=DEVICE_TYPE,
|
||||
)
|
||||
lora_meta.prepare_tensors(data.token_lora_mapping)
|
||||
|
||||
@@ -219,7 +224,9 @@ def check_lora_expand_kernel(
|
||||
|
||||
# Setup metadata information for the LoRA kernel.
|
||||
lora_meta = LoRAKernelMeta.make(
|
||||
max_loras=num_loras, max_num_tokens=token_nums, device="cuda"
|
||||
max_loras=num_loras,
|
||||
max_num_tokens=token_nums,
|
||||
device=DEVICE_TYPE,
|
||||
)
|
||||
lora_meta.prepare_tensors(data.token_lora_mapping)
|
||||
|
||||
@@ -367,7 +374,7 @@ test_params = {
|
||||
}
|
||||
|
||||
DTYPES = [torch.float16, torch.bfloat16]
|
||||
DEVICES = [f"cuda:{0}"]
|
||||
DEVICES = [f"{DEVICE_TYPE}:{0}"]
|
||||
SEED = [0]
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user