Remove mindie_turbo (#4896)

mindie_turbo is out of data for long time. This PR remove the related register method.

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

tests/ut/test_utils.py

@@ -112,27 +112,6 @@ class TestUtils(TestBase):
         output_tensor = utils.aligned_16(input_tensor)
         self.assertEqual(output_tensor.shape[0], 32)
 
-    @mock.patch('importlib.util.find_spec')
-    @mock.patch('importlib.import_module')
-    def test_try_register_lib(self, mock_import_module, mock_find_spec):
-        # import OK
-        mock_find_spec.return_value = mock.MagicMock()
-        mock_import_module.return_value = mock.MagicMock()
-        lib_name = "existing_lib"
-        lib_info = "Library found and imported successfully"
-        utils.try_register_lib(lib_name, lib_info)
-
-        # Can't find lib
-        mock_find_spec.return_value = None
-        lib_name = "non_existing_lib"
-        utils.try_register_lib(lib_name)
-
-        # import error
-        mock_find_spec.return_value = mock.MagicMock()
-        mock_import_module.side_effect = ImportError("import error")
-        lib_name = "error_lib"
-        utils.try_register_lib(lib_name)
-
     def test_enable_custom_op(self):
         result = utils.enable_custom_op()
         self.assertTrue(result)

tests/ut/worker/test_worker_v1.py

@@ -52,14 +52,12 @@ class TestNPUWorker(TestBase):
     @patch("vllm_ascend.worker.worker_v1.get_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.init_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.check_ascend_device_type")
-    @patch("vllm_ascend.worker.worker_v1.try_register_lib")
     @patch(init_cached_hf_modules_path)
     @patch("vllm_ascend.worker.worker_v1.NPUWorker._init_profiler")
     def test_init_npu_worker_normal_case(
         self,
         mock_init_profiler,
         mock_init_cached_hf_modules,
-        mock_try_register_lib,
         mock_check_ascend_device_type,
         mock_init_ascend_config,
         mock_get_ascend_config,
@@ -94,12 +92,6 @@ class TestNPUWorker(TestBase):
         mock_init_ascend_config.assert_called_once_with(self.vllm_config_mock)
         mock_check_ascend_device_type.assert_called_once()
 
-        # Verify try_register_lib call
-        mock_try_register_lib.assert_called_once_with(
-            "mindie_turbo",
-            "MindIE Turbo is installed. vLLM inference will be accelerated with MindIE Turbo.",
-        )
-
         # Verify cache_dtype setting
         self.assertEqual(worker.cache_dtype, torch.float16)
         mock_init_profiler.assert_called_once()
@@ -114,14 +106,12 @@ class TestNPUWorker(TestBase):
     @patch("vllm_ascend.worker.worker_v1.get_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.init_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.check_ascend_device_type")
-    @patch("vllm_ascend.worker.worker_v1.try_register_lib")
     @patch(init_cached_hf_modules_path)
     @patch("vllm_ascend.worker.worker_v1.NPUWorker._init_profiler")
     def test_init_npu_worker_with_trust_remote_code(
         self,
         mock_init_profiler,
         mock_init_cached_hf_modules,
-        mock_try_register_lib,
         mock_check_ascend_device_type,
         mock_init_ascend_config,
         mock_get_ascend_config,
@@ -159,14 +149,12 @@ class TestNPUWorker(TestBase):
     @patch("vllm_ascend.worker.worker_v1.get_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.init_ascend_config")
     @patch("vllm_ascend.worker.worker_v1.check_ascend_device_type")
-    @patch("vllm_ascend.worker.worker_v1.try_register_lib")
     @patch(init_cached_hf_modules_path)
     @patch("vllm_ascend.worker.worker_v1.NPUWorker._init_profiler")
     def test_init_npu_worker_with_custom_cache_dtype(
         self,
         mock_init_profiler,
         mock_init_cached_hf_modules,
-        mock_try_register_lib,
         mock_check_ascend_device_type,
         mock_init_ascend_config,
         mock_get_ascend_config,

vllm_ascend/utils.py

@@ -219,19 +219,6 @@ def aligned_16(tensor: torch.Tensor):
     return new_tensor
 
 
-def try_register_lib(lib_name: str, lib_info: str = ""):
-    import importlib
-    import importlib.util
-    try:
-        module_spec = importlib.util.find_spec(lib_name)
-        if module_spec is not None:
-            importlib.import_module(lib_name)
-            if lib_info:
-                logger.info(lib_info)
-    except Exception:
-        pass
-
-
 def enable_custom_op():
     """
     Enable lazy init for vllm_ascend_C to avoid early initialization of CANN's RTS component.

vllm_ascend/worker/worker_v1.py

@@ -53,8 +53,7 @@ from vllm_ascend.distributed.parallel_state import init_ascend_model_parallel
 from vllm_ascend.ops.triton.triton_utils import init_device_properties_triton
 from vllm_ascend.platform import NPUPlatform
 from vllm_ascend.utils import (check_ascend_device_type, enable_sp,
-                               is_enable_nz, register_ascend_customop,
+                               is_enable_nz, register_ascend_customop)
-                               try_register_lib)
 from vllm_ascend.worker.model_runner_v1 import NPUModelRunner
 
 torch._dynamo.trace_rules.clear_lru_cache()  # noqa: E402
@@ -111,11 +110,6 @@ class NPUWorker(WorkerBase):
             except Exception:
                 logger.info("Skip binding cpu.")
 
-        # Try to import mindie_turbo to accelerate vLLM inference.
-        try_register_lib(
-            "mindie_turbo",
-            "MindIE Turbo is installed. vLLM inference will be accelerated with MindIE Turbo."
-        )
         if self.cache_config.cache_dtype == "auto":
             self.cache_dtype = self.model_config.dtype
         else: