[FIX]Fix Machete compile via ENABLE_MACHETE (#3727)

* add ENABLE_MACHETE

* fix

* revert

* update

* pre_commit

* fix

* fix

---------

Co-authored-by: Ayakouji <yuhongh@qq.com>
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>
Co-authored-by: aquagull <hongyuh@qq.com>

fastdeploy/model_executor/layers/quantization/ops/machete_mm.py

@@ -26,8 +26,14 @@ def get_sm_version():
     return cc
 
 
+_ENABLE_MACHETE = False
 if current_platform.is_cuda() and get_sm_version() == 90:
-    from fastdeploy.model_executor.ops.gpu import machete_mm, machete_prepack_B
+    try:
+        from fastdeploy.model_executor.ops.gpu import machete_mm, machete_prepack_B
+
+        _ENABLE_MACHETE = True
+    except Exception:
+        pass
 
 
 def get_pack_factor(num_bits):

fastdeploy/model_executor/layers/quantization/weight_only.py

@@ -34,12 +34,6 @@ from ..utils import get_tensor
 from .quant_base import QuantConfigBase, QuantMethodBase
 
 
-def get_sm_version():
-    prop = paddle.device.cuda.get_device_properties()
-    cc = prop.major * 10 + prop.minor
-    return cc
-
-
 class WeightOnlyConfig(QuantConfigBase):
     """
     Quantization config for weight only
@@ -139,10 +133,14 @@ class WeightOnlyConfig(QuantConfigBase):
                 else:
                     raise ValueError(f"Unsupported MOE backend {layer.use_method}")
             else:
+                from fastdeploy.model_executor.layers.quantization.ops.machete_mm import (
+                    _ENABLE_MACHETE,
+                )
+
                 if (
                     self.name() == "wint4"
+                    and _ENABLE_MACHETE
                     and envs.FD_USE_MACHETE == "1"
-                    and get_sm_version() == 90
                     and layer.weight_shape[1]
                     and layer.weight_shape[1] % 128 == 0
                 ):