update

Author: bowang007

.github/workflows/build-test-linux-x86_64.yml

@@ -426,7 +426,7 @@ jobs:
         set -euo pipefail
         pushd .
         cd tests/py/dynamo
-        python -m pytest -ra --junitxml=${RUNNER_TEST_RESULTS_DIR}/l2_dynamo_plugins_tests_results.xml automatic_plugin/
+        python -m pytest -ra -n 0 --junitxml=${RUNNER_TEST_RESULTS_DIR}/l2_dynamo_plugins_tests_results.xml automatic_plugin/
         popd
 
   L2-torchscript-tests:

tests/py/dynamo/automatic_plugin/test_automatic_plugin.py

@@ -1,4 +1,3 @@
-import os
 import unittest
 from typing import Tuple
 
@@ -13,9 +12,6 @@
 
 from ..conversion.harness import DispatchTestCase
 
-os.environ["CUDA_LAUNCH_BLOCKING"] = "1"
-
-
 @triton.jit
 def elementwise_mul_kernel(X, Y, Z, BLOCK_SIZE: tl.constexpr):
     # Program ID determines the block of data each thread will process

tests/py/dynamo/automatic_plugin/test_automatic_plugin_with_attrs.py

@@ -1,4 +1,3 @@
-import os
 import unittest
 from typing import Tuple
 
@@ -13,9 +12,6 @@
 
 from ..conversion.harness import DispatchTestCase
 
-os.environ["CUDA_LAUNCH_BLOCKING"] = "1"
-
-
 @triton.jit
 def elementwise_scale_mul_kernel(X, Y, Z, a, b, BLOCK_SIZE: tl.constexpr):
     pid = tl.program_id(0)

tests/py/dynamo/automatic_plugin/test_flashinfer_rmsnorm.py

@@ -34,7 +34,7 @@ def _(input: torch.Tensor, weight: torch.Tensor, b: float = 1e-6) -> torch.Tenso
     )
 
 
-# @unittest.skip("Not Available")
+@unittest.skip("Not Available")
 @unittest.skipIf(
     not importlib.util.find_spec("flashinfer")
     or torch_tensorrt.ENABLED_FEATURES.tensorrt_rtx,