From ac3d821bc0b2186b598db2b73434250a9e4d0124 Mon Sep 17 00:00:00 2001
From: Leo Chen <chenqiuliang@baidu.com>
Date: Thu, 4 Mar 2021 15:01:43 +0800
Subject: [PATCH] [NPU] add npu kernel for equal op (#31393)

* add npu kernel for equal op

* refine code

* add more ut

* update year
---
 .../operators/controlflow/compare_op_npu.cc   | 53 +++++++++++
 .../unittests/npu/test_compare_op_npu.py      | 90 +++++++++++++++++++
 2 files changed, 143 insertions(+)
 create mode 100644 paddle/fluid/operators/controlflow/compare_op_npu.cc
 create mode 100644 python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py

diff --git a/paddle/fluid/operators/controlflow/compare_op_npu.cc b/paddle/fluid/operators/controlflow/compare_op_npu.cc
new file mode 100644
index 0000000000..58401302bc
--- /dev/null
+++ b/paddle/fluid/operators/controlflow/compare_op_npu.cc
@@ -0,0 +1,53 @@
+/* Copyright (c) 2021 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 <algorithm>
+#include <string>
+#include <vector>
+
+#include "paddle/fluid/framework/op_registry.h"
+#include "paddle/fluid/framework/op_version_registry.h"
+#include "paddle/fluid/operators/controlflow/compare_op.h"
+#include "paddle/fluid/operators/elementwise/elementwise_op_function.h"
+#include "paddle/fluid/operators/npu_op_runner.h"
+
+namespace paddle {
+namespace operators {
+
+template <typename T>
+class EqualNPUKernel : public framework::OpKernel<T> {
+ public:
+  void Compute(const framework::ExecutionContext& ctx) const override {
+    auto* x = ctx.Input<framework::LoDTensor>("X");
+    auto* y = ctx.Input<framework::LoDTensor>("Y");
+    auto* out = ctx.Output<framework::LoDTensor>("Out");
+    out->mutable_data<bool>(ctx.GetPlace());
+
+    auto runner = NpuOpRunner("Equal", {*x, *y}, {*out}, {});
+    auto stream =
+        ctx.template device_context<paddle::platform::NPUDeviceContext>()
+            .stream();
+    runner.Run(stream);
+  }
+};
+
+}  // namespace operators
+}  // namespace paddle
+
+namespace ops = paddle::operators;
+namespace plat = paddle::platform;
+
+REGISTER_OP_NPU_KERNEL(equal, ops::EqualNPUKernel<float>,
+                       ops::EqualNPUKernel<plat::float16>,
+                       ops::EqualNPUKernel<int>);
diff --git a/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py
new file mode 100644
index 0000000000..c82897d54b
--- /dev/null
+++ b/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py
@@ -0,0 +1,90 @@
+#  Copyright (c) 2021 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 numpy as np
+import unittest
+import sys
+sys.path.append("..")
+from op_test import OpTest
+import paddle
+import paddle.fluid as fluid
+
+paddle.enable_static()
+SEED = 2021
+
+
+@unittest.skipIf(not paddle.is_compiled_with_npu(),
+                 "core is not compiled with NPU")
+class TestEqual(OpTest):
+    def setUp(self):
+        self.set_npu()
+        self.op_type = "equal"
+        self.place = paddle.NPUPlace(0)
+
+        self.init_dtype()
+        np.random.seed(SEED)
+        x = np.random.uniform(1, 2, [11, 17]).astype(self.dtype)
+        y = np.random.uniform(1, 2, [11, 17]).astype(self.dtype)
+        out = x == y  # all elements are not equal
+
+        self.inputs = {
+            'X': OpTest.np_dtype_to_fluid_dtype(x),
+            'Y': OpTest.np_dtype_to_fluid_dtype(y)
+        }
+        self.outputs = {'Out': out}
+
+    def set_npu(self):
+        self.__class__.use_npu = True
+
+    def init_dtype(self):
+        self.dtype = np.float32
+
+    def test_check_output(self):
+        self.check_output_with_place(self.place, check_dygraph=False)
+
+
+class TestEqual2(TestEqual):
+    def setUp(self):
+        self.set_npu()
+        self.op_type = "equal"
+        self.place = paddle.NPUPlace(0)
+
+        self.init_dtype()
+        np.random.seed(SEED)
+        x = np.random.uniform(1, 2, [11, 17]).astype(self.dtype)
+        y = x.copy()
+        y[0][1] = 1
+        out = x == y  # all elements are equal, except position [0][1]
+
+        self.inputs = {
+            'X': OpTest.np_dtype_to_fluid_dtype(x),
+            'Y': OpTest.np_dtype_to_fluid_dtype(y)
+        }
+        self.outputs = {'Out': out}
+
+
+class TestEqual2FP16(TestEqual2):
+    def init_dtype(self):
+        self.dtype = np.float16
+
+
+class TestEqual2Int(TestEqual2):
+    def init_dtype(self):
+        self.dtype = np.int32
+
+
+if __name__ == '__main__':
+    unittest.main()
-- 
GitLab