diff --git a/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py b/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
index 3976dec4be04b6929736b054f6f87c6308b50d68..a2f1d77460819a933e8ab536d0ebc7dd32bedf3f 100644
--- a/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
+++ b/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
@@ -175,9 +175,6 @@ class TestMNIST(TestParallelExecutorBase):
             for use_fast_executor in (False, True):
                 self.check_batchnorm_fc_convergence(use_cuda, use_fast_executor)
 
-    # FIXME(wuyi): should checkout why this fails when merging
-    # https://github.com/PaddlePaddle/Paddle/pull/16545
-    @unittest.skip("should fix this later")
     def test_batchnorm_fc_with_new_strategy(self):
         # NOTE: the computation result of nccl_reduce is non-deterministic,
         # related issue: https://github.com/NVIDIA/nccl/issues/157