[Optimization] Use a separate driver when using Triton with Paddle (#6897)

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

fastdeploy/model_executor/ops/triton_ops/triton_utils.py

@@ -18,29 +18,42 @@ import inspect
 import os
 import re
 import sys
-from importlib.metadata import PackageNotFoundError, distribution
 
 import paddle
 import triton
 from paddle.base.framework import OpProtoHolder
 
 from fastdeploy import envs
+from fastdeploy.utils import _is_package_installed
 
 compile_file = triton.__path__[0] + "/tools/compile.py"
 link_file = triton.__path__[0] + "/tools/link.py"
 python_path = sys.executable
 
+if _is_package_installed("torch"):
+    with paddle.use_compat_guard(enable=True, silent=True):
+        from triton.runtime.driver import _create_driver
 
-def _is_package_installed(dist_name: str) -> bool:
-    try:
-        distribution(dist_name)
-        return True
-    except PackageNotFoundError:
-        return False
+        paddle_driver = _create_driver()
+
+
+def swap_driver_guard(fn):
+    from triton.runtime.driver import driver
+
+    # A lightweight wrapper to enable compatibility for triton kernel
+    def wrapped_fn(*args, **kwargs):
+        driver.set_active(paddle_driver)
+        try:
+            return fn(*args, **kwargs)
+        finally:
+            driver.reset_active()
+
+    return wrapped_fn
 
 
 def enable_compat_on_triton_kernel(triton_kernel):
     # When torch is not installed, this decorator does not do anything, just return the original triton kernel.
+    # Because the `paddle.enable_compat(scope={"triton"})` already enabled in `__init__.py`, it will take zero runtime overhead.
     if not _is_package_installed("torch"):
         return triton_kernel
 
@@ -49,7 +62,7 @@ def enable_compat_on_triton_kernel(triton_kernel):
             self.kernel = kernel
 
         def __getitem__(self, index):
-            return paddle.use_compat_guard(enable=True, silent=True)(self.kernel[index])
+            return swap_driver_guard(self.kernel[index])
 
     return WrappedTritonKernel(triton_kernel)