diff --git a/paddle/fluid/operators/truncated_gaussian_random_op_mlu.cc b/paddle/fluid/operators/truncated_gaussian_random_op_mlu.cc
new file mode 100644
index 0000000000000000000000000000000000000000..6d5d9f8a3b1d86ae1f952bc1c7dd0944af786382
--- /dev/null
+++ b/paddle/fluid/operators/truncated_gaussian_random_op_mlu.cc
@@ -0,0 +1,62 @@
+/* Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License. */
+
+#include <limits>
+#include <random>
+
+#include "paddle/fluid/framework/generator.h"
+#include "paddle/fluid/framework/op_registry.h"
+#include "paddle/fluid/operators/truncated_gaussian_random_op.h"
+
+namespace paddle {
+namespace operators {
+
+template <typename T>
+class TruncatedGaussianRandomMLUKernel : public framework::OpKernel<T> {
+ public:
+  void Compute(const framework::ExecutionContext& context) const override {
+    float mean = context.Attr<float>("mean");
+    float std = context.Attr<float>("std");
+    auto* tensor = context.Output<framework::Tensor>("Out");
+    tensor->mutable_data<T>(context.GetPlace());
+
+    framework::Tensor cpu_tensor(tensor->dtype());
+    cpu_tensor.Resize(tensor->dims());
+    T* data_cpu = cpu_tensor.mutable_data<T>(platform::CPUPlace());
+
+    std::uniform_real_distribution<T> dist(std::numeric_limits<float>::min(),
+                                           1.0);
+    TruncatedNormal<T> truncated_normal(mean, std);
+    int64_t size = tensor->numel();
+
+    unsigned int seed = static_cast<unsigned int>(context.Attr<int>("seed"));
+    auto engine = framework::GetCPURandomEngine(seed);
+
+    for (int64_t i = 0; i < size; ++i) {
+      data_cpu[i] = truncated_normal(dist(*engine));
+    }
+
+    auto& dev_ctx =
+        context.template device_context<platform::MLUDeviceContext>();
+    framework::TensorCopy(cpu_tensor, context.GetPlace(), dev_ctx, tensor);
+    dev_ctx.Wait();
+  }
+};
+
+}  // namespace operators
+}  // namespace paddle
+
+namespace ops = paddle::operators;
+REGISTER_OP_MLU_KERNEL(truncated_gaussian_random,
+                       ops::TruncatedGaussianRandomMLUKernel<float>);
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_truncated_gaussian_random_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_truncated_gaussian_random_op_mlu.py
new file mode 100644
index 0000000000000000000000000000000000000000..e1752029ef97c4eff75aa6813da75971f53f00be
--- /dev/null
+++ b/python/paddle/fluid/tests/unittests/mlu/test_truncated_gaussian_random_op_mlu.py
@@ -0,0 +1,91 @@
+#   Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import print_function
+
+import unittest
+import numpy
+
+import paddle
+import paddle.fluid as fluid
+import paddle.fluid.core as core
+import sys
+
+sys.path.append("..")
+from op_test import OpTest
+from paddle.fluid.op import Operator
+from paddle.fluid.executor import Executor
+from paddle.fluid.framework import _test_eager_guard
+
+paddle.enable_static()
+
+
+class TestTrunctedGaussianRandomOp(unittest.TestCase):
+
+    def setUp(self):
+        self.op_type = "truncated_gaussian_random"
+        self.inputs = {}
+        self.attrs = {
+            "shape": [10000],
+            "mean": .0,
+            "std": 1.,
+            "seed": 10,
+        }
+        self.outputs = ["Out"]
+
+    def test_cpu(self):
+        self.gaussian_random_test(place=fluid.CPUPlace())
+        self.gaussian_random_test_eager(place=fluid.CPUPlace())
+
+    def test_mlu(self):
+        if core.is_compiled_with_mlu():
+            self.gaussian_random_test(place=fluid.MLUPlace(0))
+            # self.gaussian_random_test_eager(place=fluid.MLUPlace(0))
+
+    def gaussian_random_test(self, place):
+
+        program = fluid.Program()
+        block = program.global_block()
+        vout = block.create_var(name="Out")
+        op = block.append_op(type=self.op_type,
+                             outputs={"Out": vout},
+                             attrs=self.attrs)
+
+        op.desc.infer_var_type(block.desc)
+        op.desc.infer_shape(block.desc)
+
+        fetch_list = []
+        for var_name in self.outputs:
+            fetch_list.append(block.var(var_name))
+
+        exe = Executor(place)
+        outs = exe.run(program, fetch_list=fetch_list)
+        tensor = outs[0]
+        self.assertAlmostEqual(numpy.mean(tensor), .0, delta=0.1)
+        self.assertAlmostEqual(numpy.var(tensor), 0.773, delta=0.1)
+
+    # TruncatedNormal.__call__ has no return value, so here call _C_ops api
+    # directly
+    def gaussian_random_test_eager(self, place):
+        with fluid.dygraph.guard(place):
+            with _test_eager_guard():
+                out = paddle._C_ops.final_state_truncated_gaussian_random(
+                    self.attrs["shape"], self.attrs["mean"], self.attrs["std"],
+                    self.attrs["seed"], core.VarDesc.VarType.FP32, place)
+                self.assertAlmostEqual(numpy.mean(out.numpy()), .0, delta=0.1)
+                self.assertAlmostEqual(numpy.var(out.numpy()), 0.773, delta=0.1)
+
+
+if __name__ == "__main__":
+    unittest.main()