[Chore] Separate out optional dependency checks from vllm.utils (#27207)

Signed-off-by: dongbo910220 <1275604947@qq.com>
Signed-off-by: dongbo910220 <32610838+dongbo910220@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
This commit is contained in:
dongbo910220
2025-10-22 10:44:21 -04:00
committed by GitHub
co-authored by Wentao Ye Cyrus Leung
parent 49c00fe304
commit 3ae082c373
22 changed files with 60 additions and 64 deletions
@@ -14,8 +14,9 @@ from vllm.model_executor.layers.quantization.utils.ocp_mx_utils import (
OCP_MX_Scheme,
)
from vllm.model_executor.layers.quantization.utils.quant_utils import GroupShape
from vllm.utils import cdiv, has_triton_kernels
from vllm.utils import cdiv
from vllm.utils.flashinfer import has_flashinfer_cutlass_fused_moe
from vllm.utils.import_utils import has_triton_kernels
logger = init_logger(__name__)
@@ -26,12 +26,12 @@ from vllm.model_executor.layers.fused_moe.utils import _resize_cache
from vllm.model_executor.layers.quantization.utils.fp8_utils import (
per_token_group_quant_fp8,
)
from vllm.utils import has_deep_gemm
from vllm.utils.deep_gemm import (
get_mk_alignment_for_contiguous_layout,
m_grouped_fp8_gemm_nt_contiguous,
)
from vllm.utils.func_utils import run_once
from vllm.utils.import_utils import has_deep_gemm
logger = init_logger(__name__)
@@ -13,7 +13,7 @@ from vllm.model_executor.layers.fused_moe.topk_weight_and_reduce import (
TopKWeightAndReduceNoOP,
)
from vllm.triton_utils import tl, triton
from vllm.utils import has_triton_kernels
from vllm.utils.import_utils import has_triton_kernels
logger = init_logger(__name__)
@@ -55,8 +55,9 @@ from vllm.model_executor.layers.quantization.utils.flashinfer_utils import (
from vllm.model_executor.utils import set_weight_attrs
from vllm.platforms import current_platform
from vllm.platforms.interface import CpuArchEnum
from vllm.utils import cdiv, has_deep_ep, has_pplx, round_up
from vllm.utils import cdiv, round_up
from vllm.utils.flashinfer import has_flashinfer_cutlass_fused_moe
from vllm.utils.import_utils import has_deep_ep, has_pplx
from vllm.utils.torch_utils import current_stream, direct_register_custom_op
from vllm.v1.worker.ubatching import dbo_current_ubatch_id
@@ -93,7 +93,6 @@ from vllm.model_executor.parameter import (
from vllm.model_executor.utils import set_weight_attrs
from vllm.platforms import current_platform
from vllm.scalar_type import scalar_types
from vllm.utils import has_deep_gemm
from vllm.utils.deep_gemm import (
fp8_gemm_nt,
get_col_major_tma_aligned_tensor,
@@ -102,6 +101,7 @@ from vllm.utils.deep_gemm import (
should_use_deepgemm_for_fp8_linear,
)
from vllm.utils.flashinfer import has_flashinfer_moe
from vllm.utils.import_utils import has_deep_gemm
if TYPE_CHECKING:
from vllm.model_executor.models.utils import WeightsMapper
@@ -48,11 +48,9 @@ from vllm.model_executor.layers.quantization.utils.quant_utils import is_layer_s
from vllm.model_executor.utils import set_weight_attrs
from vllm.platforms import current_platform
from vllm.scalar_type import scalar_types
from vllm.utils import (
has_triton_kernels,
round_up,
)
from vllm.utils import round_up
from vllm.utils.flashinfer import has_flashinfer
from vllm.utils.import_utils import has_triton_kernels
from vllm.utils.torch_utils import is_torch_equal_or_newer
logger = init_logger(__name__)